francis2019stakesscalesskepticism
/data/papers/francis2019stakesscalesskepticism/francis2019stakesscalesskepticism.yaml
schema_version: '1.2'
paper:
  paper_id: francis2019stakesscalesskepticism
  citation: Francis, K. B., Beaman, C. P., & Hansen, N. (2019). Stakes, Scales, and Skepticism. Ergo, 6(16).
  short_label: Francis et al. (2019)
  doi: 10.3998/ergo.12405314.0006.016
  published: 'Yes'
  year: 2019
  language: English
  language_other: null
  research_objective: Investigate the scalarity of stakes effects on knowledge judgments by varying stakes magnitude across
    multiple scenarios using evidence-fixed and evidence-seeking paradigms, and replicate Sripada & Stanley (2012).
  data_availability:
    data_available_online: 'Yes'
    url: https://researchdata.reading.ac.uk/205/
    notes: 'Open University of Reading dataset (DOI: 10.17864/1947.205); source ZIPs/CSVs and README copied to out/external
      for reproducible extraction.'
  notes: Effect sizes were recovered from the public raw data; analysis/effect_sizes.qmd is the source of truth for computed
    d and v values. For evidence-seeking effects, YAML stores raw low-minus-high d values; downstream meta-analysis reverses evidence-seeking effects programmatically. Study 1 negative-polarity evidence-fixed effects are reverse-coded to align with knowledge-attribution
    direction.
studies:
- study_id: 1
  label: 'Experiment 1: Evidence-fixed design (multiple scenarios; 4 stakes levels; polarity manipulation)'
  language: English
  language_other: null
  objective: Test whether varying stakes magnitude (4 levels) affects agreement with knowledge attributions/denials across
    six scenarios, and whether this interacts with prompt polarity ('knows' vs 'doesn't know').
  sample:
    n_final: 97
    recruitment: mTurk
    recruitment_other: null
    compensation: money
    compensation_other: $1.75
    characteristics: 'MTurk sample: 44 females, 52 males, 1 non-binary; ages 20–71; positive polarity N=55, negative polarity
      N=42.'
    mean_age: 39.64
    mean_age_prov:
      page: null
      quote: leaving a final sample of 97 participants (44 females, 52 males, 1 non-binary gender identity) between 20 and
        71 years old (M = 39.64 years, SD = 12.00 years).
      tei_id: null
      table_ref: null
    provenance:
      page: null
      quote: One hundred and twenty participants were recruited from MTurk and paid $1.75 each... leaving a final sample of
        97 participants... Participants were randomly assigned to the positive polarity condition (N = 55) or the negative
        polarity condition (N = 42).
      tei_id: null
      table_ref: null
  design: null
  design_other: '2 (polarity: know vs doesn''t know; between-subjects) × 4 (stakes scale: 1–4; within-subjects) mixed design;
    six scenarios presented in a randomized block design.'
  manipulated_factors:
  - 'Prompt polarity: ''knows'' vs ''doesn''t know'''
  paradigm: Agreement with knowledge claim
  paradigm_other: null
  scale:
    label: Likert 7-point
    points: 7
    anchors: 1 = strongly disagree; 7 = strongly agree
    direction: 'Higher numbers indicate stronger agreement with the prompt sentence (note: control Prompt 2 uses a reversed
      Likert scale).'
    provenance:
      page: null
      quote: Prompt 1... "You know the coin landed heads" 1 (strongly disagree) -7 (strongly agree)... Prompt 2... "You don't
        know that the coin landed heads" 1 (strongly agree) -7 (strongly disagree).
      tei_id: null
      table_ref: null
  measures:
    knowledge_question_text: 'To what extent do you agree or disagree with the following claim: Subject x knows that P / doesn''t
      know that P (scenario-specific).'
    knowledge_question_first: null
    additional_question_text: null
  scenarios:
  - scenario_code: paramedic
    scenario_type: Paramedic GPS/wrong-turn scenario; stakes scaled by severity of consequences (lives).
    high_stakes_text: 'High stakes example: school bus carrying 50 children on fire; wrong turn → children die.'
    low_stakes_text: 'Low stakes example: one person with a broken arm; wrong turn → inconvenienced.'
    provenance:
      page: null
      quote: 'Paramedic (low): "there is one person... with a broken arm... If Megan makes a wrong turn... will be inconvenienced".
        Higher stakes: "a school bus carrying 50 children... on fire... the children will die."'
      tei_id: null
      table_ref: null
  - scenario_code: vaccine
    scenario_type: Vaccine checklist scenario; stakes scaled by number/severity of harms (lives).
    high_stakes_text: 'High stakes example: 100 participants die after excruciating pain if steps not followed.'
    low_stakes_text: 'Low stakes example: 1 participant gets mild cold-like symptoms if steps not followed.'
    provenance:
      page: null
      quote: 'Low: "one human research participant... will give them mild cold-like symptoms." High: "100 human research participants...
        will kill them all after several days of excruciating pain."'
      tei_id: null
      table_ref: null
  - scenario_code: mountaineering
    scenario_type: Mountaineering rope-inspection scenario; stakes scaled by injury severity (physical injury).
    high_stakes_text: 'High stakes example: 1,000-foot drop; fall would be fatal.'
    low_stakes_text: 'Low stakes example: 5-foot drop; minor injuries possible.'
    provenance:
      page: null
      quote: 'Low: "drop... around 5 feet... minor injuries". High: "drop... around 1,000 feet... fatal".'
      tei_id: null
      table_ref: null
  - scenario_code: game_show
    scenario_type: Game show trivia scenario; stakes scaled by money at stake.
    high_stakes_text: 'High stakes example: $1,000,000 at stake.'
    low_stakes_text: 'Low stakes example: $1 at stake.'
    provenance:
      page: null
      quote: 'Low: "only $1 is at stake". High: "$1,000,000 is at stake".'
      tei_id: null
      table_ref: null
  - scenario_code: introduction
    scenario_type: Guest-speaker name introduction scenario; stakes scaled by embarrassment/reputation.
    high_stakes_text: 'High stakes example: national television interview; wrong name → very embarrassed and reflects badly
      on university reputation.'
    low_stakes_text: 'Low stakes example: lunch with 5 colleagues; wrong name → slightly embarrassed.'
    provenance:
      page: null
      quote: 'Low: "lunch... 5 colleagues... slightly embarrassed". High: "national television... thousands... very embarrassed...
        reflect very badly... university''s reputation".'
      tei_id: null
      table_ref: null
  - scenario_code: arson
    scenario_type: Arson/sprinkler system scenario; stakes scaled by value of possessions (including baby).
    high_stakes_text: 'High stakes example: nursery room where baby sleeps; sprinklers failing puts baby at risk.'
    low_stakes_text: 'Low stakes example: storage room with garbage/recycling at risk.'
    provenance:
      page: null
      quote: 'Low: "storage room... garbage and recycling". High: "nursery room, where her baby sleeps... the baby, is at
        risk from arson".'
      tei_id: null
      table_ref: null
  effects:
  - effect_id: s1_e1
    subgroup: 'Experiment 1: Evidence-fixed design -- Paramedic -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan... paramedic... accident...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area... traveling on the right route to get to the accident.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The stakes are conveyed to the protagonist within the vignette (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own navigation resources/knowledge (familiarity + checking GPS).
      attribution_person:
        provenance:
          page: null
          quote: Paramedic +Subject x... knows that she will make it to the accident without taking a wrong turn.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Paramedic Low vs Paramedic 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.50909090909
      sd: 1.34539994429
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: Paramedic Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 5.67272727273
      sd: 1.3479002251
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: Paramedic 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.121513595094
      v: 0.011522200282
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e2
    subgroup: 'Experiment 1: Evidence-fixed design -- Vaccine -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Elaine has done this before, and she has a check list that specifies all of the steps...
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: Evidence is based on the agent’s own checking/procedure-following (consulting a checklist).
      attribution_person:
        provenance:
          page: null
          quote: Elaine knows that she is making the vaccine correctly
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Elaine knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vaccine Low vs Vaccine 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.98181818182
      sd: 1.14650691864
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: Vaccine Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 6.07272727273
      sd: 1.11976428496
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: Vaccine 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.0802223158773
      v: 0.0153833192653
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e3
    subgroup: 'Experiment 1: Evidence-fixed design -- Mountaineering -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... mountain climbing expedition... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing... the drop... If not tied together securely... injuries...
          tei_id: null
          table_ref: null
        reason: The stakes are described within the scenario (agent is in the situation and consequences are described).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: Mountaineering +Subject x... knows that the rope is tied securely.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mountaineering Low vs Mountaineering 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.83636363636
      sd: 1.21355975243
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: Mountaineering Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 5.96363636364
      sd: 1.03572548135
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: Mountaineering 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.112815214964
      v: 0.0146515699543
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e4
    subgroup: 'Experiment 1: Evidence-fixed design -- Game show -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?" ... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Emma has recently read a list of the most obscure world capitals and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described as part of the protagonist’s situation (game show winnings/losses).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory (first-person evidence).
      attribution_person:
        provenance:
          page: null
          quote: Game show +Subject x... knows that the capital of Tanzania is Dodoma.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: GameShow low vs GameShow 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.27272727273
      sd: 1.45874813726
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: GameShow low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 5.21818181818
      sd: 1.37019745929
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: GameShow 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0385435142968
      v: 0.00648211042106
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e5
    subgroup: 'Experiment 1: Evidence-fixed design -- Introduction -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... introduce a guest speaker... "Dr. Woodbridge"'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... it will reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario as consequences of misnaming the speaker.
      evidence:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook with the name).
      attribution_person:
        provenance:
          page: null
          quote: Introduction +Subject x... knows that the guest speakers name is "Dr. Woodbridge".
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.98181818182
      sd: 1.22460740634
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 6.12727272727
      sd: 1.01934157134
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.129102557916
      v: 0.0126670930616
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e6
    subgroup: 'Experiment 1: Evidence-fixed design -- Possessions/Arson -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Only a functioning sprinkler system can stop a fire set by an arsonist.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own prior checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: Arson +Subject x... knows that the sprinklers are working in the x room.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Personal Val Low vs Personal Val 3.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 55
      mean: 5.90909090909
      sd: 1.00503781526
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; low
          column: Personal Val Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 55
      mean: 5.89090909091
      sd: 1.19679707232
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Pos_data_post_removal.csv; high
          column: Personal Val 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0164528734539
      v: 0.0159562267938
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
    quality_flags:
    - raw_data_recovered
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data.
  - effect_id: s1_e7
    subgroup: 'Experiment 1: Evidence-fixed design -- Paramedic -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan... paramedic... accident...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area... traveling on the right route to get to the accident.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The stakes are conveyed to the protagonist within the vignette (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own navigation resources/knowledge (familiarity + checking GPS).
      attribution_person:
        provenance:
          page: null
          quote: Paramedic +Subject x... knows that she will make it to the accident without taking a wrong turn.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Paramedic Low vs Paramedic 3. For this negative-polarity evidence-fixed effect, group means are reverse-coded
        knowledge-attribution scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.42857142857
      sd: 1.54829342941
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: Paramedic Low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.64285714286
      sd: 1.30330590108
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: Paramedic 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.149740006799
      v: 0.0210717079245
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  - effect_id: s1_e8
    subgroup: 'Experiment 1: Evidence-fixed design -- Vaccine -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Elaine has done this before, and she has a check list that specifies all of the steps...
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: Evidence is based on the agent’s own checking/procedure-following (consulting a checklist).
      attribution_person:
        provenance:
          page: null
          quote: Elaine knows that she is making the vaccine correctly
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Elaine knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vaccine Low vs Vaccine 3. For this negative-polarity evidence-fixed effect, group means are reverse-coded
        knowledge-attribution scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.61904761905
      sd: 1.43054451431
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: Vaccine Low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.54761904762
      sd: 1.51741726905
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: Vaccine 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0484386042375
      v: 0.0103158241577
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  - effect_id: s1_e9
    subgroup: 'Experiment 1: Evidence-fixed design -- Mountaineering -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... mountain climbing expedition... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing... the drop... If not tied together securely... injuries...
          tei_id: null
          table_ref: null
        reason: The stakes are described within the scenario (agent is in the situation and consequences are described).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: Mountaineering +Subject x... knows that the rope is tied securely.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mountaineering Low vs Mountaineering 3. For this negative-polarity evidence-fixed effect, group means are
        reverse-coded knowledge-attribution scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.40476190476
      sd: 1.79510885341
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: Mountaineering Low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.33333333333
      sd: 1.72027602247
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: Mountaineering 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0406284919837
      v: 0.0125311746321
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  - effect_id: s1_e10
    subgroup: 'Experiment 1: Evidence-fixed design -- Game show -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?" ... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Emma has recently read a list of the most obscure world capitals and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described as part of the protagonist’s situation (game show winnings/losses).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory (first-person evidence).
      attribution_person:
        provenance:
          page: null
          quote: Game show +Subject x... knows that the capital of Tanzania is Dodoma.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: GameShow low vs GameShow 3. For this negative-polarity evidence-fixed effect, group means are reverse-coded
        knowledge-attribution scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.33333333333
      sd: 1.60284300262
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: GameShow low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.11904761905
      sd: 1.65577159012
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: GameShow 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.131502174867
      v: 0.0104410020413
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  - effect_id: s1_e11
    subgroup: 'Experiment 1: Evidence-fixed design -- Introduction -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... introduce a guest speaker... "Dr. Woodbridge"'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... it will reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario as consequences of misnaming the speaker.
      evidence:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook with the name).
      attribution_person:
        provenance:
          page: null
          quote: Introduction +Subject x... knows that the guest speakers name is "Dr. Woodbridge".
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. For this negative-polarity evidence-fixed effect, group means are reverse-coded knowledge-attribution
        scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.97619047619
      sd: 1.19935135392
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: Intro Low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.7380952381
      sd: 1.62389960956
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: Intro 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.166792146986
      v: 0.0187820334615
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  - effect_id: s1_e12
    subgroup: 'Experiment 1: Evidence-fixed design -- Possessions/Arson -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: Only a functioning sprinkler system can stop a fire set by an arsonist.
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence comes from the agent’s own prior checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: Arson +Subject x... knows that the sprinklers are working in the x room.
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ("Subject x knows...").
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Personal Val Low vs Personal Val 3. For this negative-polarity evidence-fixed effect, group means are reverse-coded
        knowledge-attribution scores (8 - raw agreement with denial).'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 42
      mean: 5.78571428571
      sd: 1.3710546819
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; low
          column: Personal Val Low. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 42
      mean: 5.80952380952
      sd: 1.41831392923
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp1_Evidence_Fixed_Design/x1_EF_Neg_data_post_removal.csv; high
          column: Personal Val 3. Negative-polarity responses reverse-coded as 8 - raw agreement with denial.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.0170691726829
      v: 0.0217680938563
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd using paired lowest/highest stakes responses.
        Negative-polarity agreement-with-denial responses were reverse-coded to knowledge-attribution direction before computing
        d.
    quality_flags:
    - raw_data_recovered
    - negative_polarity_reverse_coded
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Negative-polarity agreement-with-denial responses were reverse-coded
      to knowledge-attribution direction before computing d.
  notes: Effects are scenario-by-polarity lowest-vs-highest stakes contrasts recovered from raw data. Study 1 negative-polarity
    evidence-fixed effects reverse-code agreement with “doesn’t know” to knowledge-attribution direction before computing
    d.
- study_id: 2
  label: 'Appendix II: Registered replication of Sripada & Stanley (2012) (pine nuts)'
  language: English
  language_other: null
  objective: Replicate Sripada & Stanley’s evidence-fixed design across three vignette pairs (basic; implicit/explicit; ignorant),
    measuring (i) evidence strength and (ii) agreement with a knowledge attribution.
  sample:
    n_final: 367
    recruitment: mTurk
    recruitment_other: null
    compensation: money
    compensation_other: $0.50
    characteristics: 'MTurk sample: 183 females, 246 males, 1 non-binary; ages 20–67; 63 suspicious/VPS responses removed;
      assigned to six conditions (N=58–68 per condition).'
    mean_age: 35.99
    mean_age_prov:
      page: null
      quote: Four hundred and thirty participants... between 20 and 67 years (M = 35.99 years, SD = 10.55 years)... Following
        screening procedures, 63 VPS and further suspicious responses were flagged and removed.
      tei_id: null
      table_ref: null
    provenance:
      page: null
      quote: Four hundred and thirty participants... recruited from MTurk and paid $0.50... Following screening procedures,
        63 VPS and further suspicious responses were flagged and removed... randomly assigned to one of six conditions (N
        = 58 -68 per condition).
      tei_id: null
      table_ref: null
  design: Between-Subjects
  design_other: 3 vignette pairs × 2 stakes (low vs high) between-subjects; each participant reads one vignette and answers
    Q1 (evidence strength) then Q2 (knowledge agreement).
  manipulated_factors:
  - 'Vignette pair type: Basic vs Implicit/Explicit vs Ignorant'
  paradigm: Agreement with knowledge claim
  paradigm_other: null
  scale:
    label: Likert 7-point
    points: 7
    anchors: 'Q1: 1 = Very weak evidence; 7 = Very strong evidence. Q2: 1 = Strongly agree; 7 = Strongly disagree.'
    direction: Higher numbers indicate stronger evidence (Q1). For Q2, anchors are Strongly agree (1) to Strongly disagree
      (7) (paper reports effects in terms of 'levels of agreement').
    provenance:
      page: null
      quote: Participants responded to the first question (evidence prompt) along a 7-point Likert-type scale (1-Very weak
        evidence, 7-Very strong evidence) and to the second prompt (knowledge prompt) along another 7-point Likert-type scale
        (1-Strongly agree, 7-Strongly disagree).
      tei_id: null
      table_ref: null
  measures:
    knowledge_question_text: '"Hannah knows her noodles are not topped with pine nuts."'
    knowledge_question_first: null
    additional_question_text: What is the strength of Hannah's evidence that her noodles are not topped with pine nuts?
  scenarios:
  - scenario_code: peanuts
    scenario_type: Pine-nuts/noodles vignette (Sripada & Stanley 2012 replication).
    high_stakes_text: null
    low_stakes_text: null
    provenance:
      page: null
      quote: What is the strength of Hannah's evidence that her noodles are not topped with pine nuts?
      tei_id: null
      table_ref: null
  effects:
  - effect_id: s2_e1
    subgroup: Basic — Evidence strength (Q1)
    subgroup_desc: Strength of evidence rating (7-point)
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: null
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence that her noodles are not topped with pine nuts?
          tei_id: null
          table_ref: null
        reason: This is the Sripada & Stanley-style pine-nuts restaurant vignette (coded as peanuts-style scenario).
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: Hannah is very much aware of this, and has known this for a very long time.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; the Basic vignette protagonist is aware of the allergy
          stakes.
      evidence:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from an external written source (the menu).
      attribution_person:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence...
          tei_id: null
          table_ref: null
        reason: DV is evidence-strength (not a knowledge attribution), so self/other knowledge-ascription coding is not applicable.
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'The noodles may be topped with pine nuts.' Hannah notes that the menu says her dish does not contain
            pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: Basic_high
      group_low: Basic_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: Basic_low
      label: Basic low stakes
      n: 58
      mean: 3.93103448276
      sd: 1.82441597447
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Basic Low__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: Basic_high
      label: Basic high stakes
      n: 68
      mean: 2.89705882353
      sd: 1.82960484994
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Basic High__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: 3.17
      f: null
      chi2: null
      z: null
      df1: 124
      df2: null
      p: 0.002
      reported_d: 0.57
      reported_r: null
      notes: 'Direction reported: strength of evidence higher in low-stakes scenario.'
      provenance:
        page: 12
        quote: In the strength of evidence comparison (left panel) there was a medium effect of stakes in the basic vignette
          pair, (t(124) = 3.17, p =.002, d = 0.57) with strength of evidence higher in the low stakes scenario.
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.565873199091
      v: 0.0332179460976
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd.
    quality_flags:
    - raw_data_recovered
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Other
    paradigm_other: Strength of evidence rating
  - effect_id: s2_e2
    subgroup: Basic — Knowledge attribution (Q2)
    subgroup_desc: Agreement with 'Hannah knows...' reverse-coded to agreement
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: This is the Sripada & Stanley-style pine-nuts restaurant vignette (coded as peanuts-style scenario).
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: Hannah is very much aware of this, and has known this for a very long time.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; the Basic vignette protagonist is aware of the allergy
          stakes.
      evidence:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from an external written source (the menu).
      attribution_person:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ('Hannah knows...').
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'The noodles may be topped with pine nuts.' Hannah notes that the menu says her dish does not contain
            pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: Basic_high
      group_low: Basic_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: Basic_low
      label: Basic low stakes
      n: 58
      mean: 3.72413793103
      sd: 1.84289154423
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Basic Low__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: Basic_high
      label: Basic high stakes
      n: 68
      mean: 2.91176470588
      sd: 1.70806073258
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Basic High__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: -2.57
      f: null
      chi2: null
      z: null
      df1: 124
      df2: null
      p: 0.011
      reported_d: 0.46
      reported_r: null
      notes: 'Direction reported: levels of agreement higher in low-stakes scenario.'
      provenance:
        page: 12
        quote: For the levels of agreement comparison (right panel) there was a smaller effect of stakes in the basic vignette
          pair, (t(124) = -.2.57, p =.011, d = 0.46) with levels of agreement are higher in the low stakes scenario.
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.458627340663
      v: 0.0327819403839
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; Q2 was reverse-coded to agreement before computing
        low minus high.
    quality_flags:
    - raw_data_recovered
    - q2_reverse_coded_to_agreement
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Agreement with knowledge claim
    paradigm_other: null
  - effect_id: s2_e3
    subgroup: Implicit/Explicit — Evidence strength (Q1)
    subgroup_desc: Strength of evidence rating (7-point)
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: null
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence that her noodles are not topped with pine nuts?
          tei_id: null
          table_ref: null
        reason: Same pine-nuts/noodles vignette family; coded as peanuts-style scenario.
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: 'Implicit Low Stakes: "Hannah likes the taste of most foods and is not a very picky eater."'
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no ignorance manipulation is present.
      evidence:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from an external written source (the menu).
      attribution_person:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence...
          tei_id: null
          table_ref: null
        reason: DV is evidence-strength (not a knowledge attribution), so self/other knowledge-ascription coding is not applicable.
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'The noodles may be topped with pine nuts.' Hannah notes that the menu says her dish does not contain
            pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: ImplicitExplicit_high
      group_low: ImplicitExplicit_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: ImplicitExplicit_low
      label: Implicit/Explicit low stakes
      n: 58
      mean: 3.62068965517
      sd: 1.94510258789
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Implicit Low__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: ImplicitExplicit_high
      label: Implicit/Explicit high stakes
      n: 61
      mean: 3.4262295082
      sd: 1.96179353648
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Explicit High__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: 0.54
      f: null
      chi2: null
      z: null
      df1: 117
      df2: null
      p: 0.588
      reported_d: null
      reported_r: null
      notes: Effect not significant per Figure 2 caption.
      provenance:
        page: 12
        quote: 'The effect of stakes was not significant in the other vignettes [Implicit/Explicit: t(117) = 0.54, p = .588;
          ...].'
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.0995353183474
      v: 0.0336764491585
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd.
    quality_flags:
    - raw_data_recovered
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Other
    paradigm_other: Strength of evidence rating
  - effect_id: s2_e4
    subgroup: Implicit/Explicit — Knowledge attribution (Q2)
    subgroup_desc: Agreement with 'Hannah knows...' reverse-coded to agreement
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: Same pine-nuts/noodles vignette family; coded as peanuts-style scenario.
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: 'Implicit Low Stakes: "Hannah likes the taste of most foods and is not a very picky eater."'
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no ignorance manipulation is present.
      evidence:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from an external written source (the menu).
      attribution_person:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ('Hannah knows...').
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'The noodles may be topped with pine nuts.' Hannah notes that the menu says her dish does not contain
            pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: ImplicitExplicit_high
      group_low: ImplicitExplicit_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: ImplicitExplicit_low
      label: Implicit/Explicit low stakes
      n: 58
      mean: 3.79310344828
      sd: 1.92635116117
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Implicit Low__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: ImplicitExplicit_high
      label: Implicit/Explicit high stakes
      n: 61
      mean: 3.27868852459
      sd: 1.87199668394
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Explicit High__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: -1.48
      f: null
      chi2: null
      z: null
      df1: 117
      df2: null
      p: 0.142
      reported_d: null
      reported_r: null
      notes: Effect not significant per Figure 2 caption.
      provenance:
        page: 12
        quote: 'Once again there was no significant effect of stakes in the other vignettes [Implicit/Explicit: t(117) = -1.48,
          p = .142; ...].'
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.270934142758
      v: 0.0339432476044
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; Q2 was reverse-coded to agreement before computing
        low minus high.
    quality_flags:
    - raw_data_recovered
    - q2_reverse_coded_to_agreement
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Agreement with knowledge claim
    paradigm_other: null
  - effect_id: s2_e5
    subgroup: Ignorant — Evidence strength (Q1)
    subgroup_desc: Strength of evidence rating (7-point)
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'No'
      evidence: External
      attribution_person: null
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence that her noodles are not topped with pine nuts?
          tei_id: null
          table_ref: null
        reason: Same pine-nuts/noodles vignette family; coded as peanuts-style scenario.
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: the protagonist being ignorant of the stakes involved (ignorant low/ignorant high)
          tei_id: null
          table_ref: null
        reason: Paper explicitly describes the Ignorant manipulation as the protagonist being unaware of the stakes.
      evidence:
        provenance:
          page: null
          quote: Sarah says, 'I heard that Mongolian dishes are often served topped with Mongolian pine nuts.' Hannah notes that
            the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from external sources (testimony + menu).
      attribution_person:
        provenance:
          page: null
          quote: What is the strength of Hannah's evidence...
          tei_id: null
          table_ref: null
        reason: DV is evidence-strength (not a knowledge attribution), so self/other knowledge-ascription coding is not applicable.
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'I heard that Mongolian dishes are often served topped with Mongolian pine nuts.' Hannah notes that
            the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: Ignorant_high
      group_low: Ignorant_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: Ignorant_low
      label: Ignorant low stakes
      n: 62
      mean: 4.25806451613
      sd: 1.88118800858
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Ignorant Low__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: Ignorant_high
      label: Ignorant high stakes
      n: 60
      mean: 4.03333333333
      sd: 1.88631707048
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Ignorant High__Strength_Evidence.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: 0.84
      f: null
      chi2: null
      z: null
      df1: 120
      df2: null
      p: 0.511
      reported_d: null
      reported_r: null
      notes: Effect not significant per Figure 2 caption.
      provenance:
        page: 12
        quote: 'The effect of stakes was not significant... [ ... Ignorant: t(120) = 0.84, p =.511].'
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.119302333567
      v: 0.0328540310837
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd.
    quality_flags:
    - raw_data_recovered
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Other
    paradigm_other: Strength of evidence rating
  - effect_id: s2_e6
    subgroup: Ignorant — Knowledge attribution (Q2)
    subgroup_desc: Agreement with 'Hannah knows...' reverse-coded to agreement
    design: Between-Subjects
    design_other: null
    moderators:
      scenario: peanuts
      skeptical_pressure: 'No'
      awareness: 'No'
      evidence: External
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: Same pine-nuts/noodles vignette family; coded as peanuts-style scenario.
      skeptical_pressure:
        provenance:
          page: null
          quote: Hannah notes that the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; no explicit prompt says that the menu might be wrong.
      awareness:
        provenance:
          page: null
          quote: the protagonist being ignorant of the stakes involved (ignorant low/ignorant high)
          tei_id: null
          table_ref: null
        reason: Paper explicitly describes the Ignorant manipulation as the protagonist being unaware of the stakes.
      evidence:
        provenance:
          page: null
          quote: Sarah says, 'I heard that Mongolian dishes are often served topped with Mongolian pine nuts.' Hannah notes that
            the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; evidence is from external sources (testimony + menu).
      attribution_person:
        provenance:
          page: null
          quote: '"Hannah knows her noodles are not topped with pine nuts."'
          tei_id: null
          table_ref: null
        reason: Participants evaluate a third-person knowledge attribution ('Hannah knows...').
      evidence_reliability:
        provenance:
          page: null
          quote: Sarah says, 'I heard that Mongolian dishes are often served topped with Mongolian pine nuts.' Hannah notes that
            the menu says her dish does not contain Mongolian pine nuts.
          tei_id: null
          table_ref: sripadastanley2012empiricaltestsinterest.yaml
        reason: Coded to match the original Sripada & Stanley extraction; menu.
    contrast:
      group_high: Ignorant_high
      group_low: Ignorant_low
      sign_convention: d = mean(low) - mean(high)
      other_notes: null
    groups:
    - group_id: Ignorant_low
      label: Ignorant low stakes
      n: 62
      mean: 4.08064516129
      sd: 2.01061435107
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; low column:
          Ignorant Low__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: Ignorant_high
      label: Ignorant high stakes
      n: 60
      mean: 3.71666666667
      sd: 2.09188639762
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Replication_Experiment/xReplic_data_post_removal.csv; high column:
          Ignorant High__Know_Prompt.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    reported_test:
      test: t
      t: -0.98
      f: null
      chi2: null
      z: null
      df1: 120
      df2: null
      p: 0.329
      reported_d: null
      reported_r: null
      notes: Effect not significant per Figure 2 caption.
      provenance:
        page: 12
        quote: 'Once again there was no significant effect of stakes... [ ... Ignorant: t(120) = -0.98, p =.329].'
        tei_id: null
        table_ref: Figure 2
    effect_size:
      metric: SMD
      d: 0.177466028998
      v: 0.0329247734798
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; Q2 was reverse-coded to agreement before computing
        low minus high.
    quality_flags:
    - raw_data_recovered
    - q2_reverse_coded_to_agreement
    notes: Raw-data recovered values replace previously missing or rounded reported-d-only effect sizes.
    paradigm: Agreement with knowledge claim
    paradigm_other: null
- study_id: 3
  label: 'Experiment 2: Evidence-seeking design (original prompts)'
  language: English
  language_other: null
  objective: Test for stakes effects on how much evidence is needed for knowledge (positive prompt) or can be had while still
    not knowing (negative prompt), across six scenarios and four stakes levels.
  sample:
    n_final: 109
    recruitment: mTurk
    recruitment_other: null
    compensation: money
    compensation_other: $1.75
    characteristics: 'MTurk sample: 54 females, 55 males; ages 21–74; positive polarity N=58, negative polarity N=51.'
    mean_age: 38.98
    mean_age_prov:
      page: null
      quote: leaving a final sample of 109 participants (54 females, 55 males) between 21 and 74 years old (M = 38.98 years,
        SD = 11.76 years).
      tei_id: null
      table_ref: null
    provenance:
      page: null
      quote: One hundred and twenty participants were recruited from MTurk and paid $1.75... leaving a final sample of 109
        participants... Participants were randomly assigned to a positive polarity condition (N = 58) or a negative polarity
        condition (N = 51).
      tei_id: null
      table_ref: null
  design: null
  design_other: Stakes (4 levels) within-subjects; prompt polarity (positive vs negative) between-subjects; six scenarios
    presented in a randomized block design.
  manipulated_factors:
  - 'Prompt polarity: evidence-seeking positive vs evidence-seeking negative'
  paradigm: Rating how much evidence is needed for knowledge
  paradigm_other: null
  scale:
    label: other
    points: null
    anchors: Numeric free response (whole number of checks); positive prompt allows '0' (knows without checking) and both
      prompts allow 'never'.
    direction: Higher numbers indicate more evidence required for knowledge (positive) / more checking still insufficient
      for knowledge (negative).
    provenance:
      page: null
      quote: 'enter a whole number: 1, 2, 3 . . . etc. If you think Elaine knows without having to check, write "0"... If
        you think Elaine will never know... write "never".'
      tei_id: null
      table_ref: null
  measures:
    knowledge_question_text: How many times does S need to check F before she knows that P?
    knowledge_question_first: null
    additional_question_text: How many times can S check F and still not know that P?
  scenarios: []
  effects:
  - effect_id: s3_e1
    subgroup: 'Experiment 2: Evidence-seeking design -- Paramedic -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Paramedic Low vs Paramedic 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 39
      mean: 1.94871794872
      sd: 1.12270158074
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: Paramedic Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 39
      mean: 2.76923076923
      sd: 1.34676099668
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: Paramedic 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.661808915227
      v: 0.0276260903612
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e2
    subgroup: 'Experiment 2: Evidence-seeking design -- Vaccine -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vaccine Low vs Vaccine 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 48
      mean: 2.47916666667
      sd: 1.58435950323
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: Vaccine Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 48
      mean: 4.22916666667
      sd: 2.48604259847
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: Vaccine 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.839514293934
      v: 0.0227377109024
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e3
    subgroup: 'Experiment 2: Evidence-seeking design -- Mountaineering -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mountaineering Low vs Mountaineering 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 44
      mean: 2.0
      sd: 1.20077494357
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: Mountaineering Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 44
      mean: 2.88636363636
      sd: 1.38456456241
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: Mountaineering 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.683958446104
      v: 0.0211432766559
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e4
    subgroup: 'Experiment 2: Evidence-seeking design -- Game show -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: GameShow low vs GameShow 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 27
      mean: 2.66666666667
      sd: 2.44948974278
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: GameShow low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 27
      mean: 5.48148148148
      sd: 3.8567659865
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: GameShow 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.871275071557
      v: 0.0732423045923
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e5
    subgroup: 'Experiment 2: Evidence-seeking design -- Introduction -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 50
      mean: 1.74
      sd: 0.828325086533
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 50
      mean: 2.4
      sd: 1.12485826772
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.668163278631
      v: 0.0224825444801
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e6
    subgroup: 'Experiment 2: Evidence-seeking design -- Possessions/Arson -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Personal Val Low vs Personal Val 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 47
      mean: 1.53191489362
      sd: 0.905323959067
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; low
          column: Personal Val Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 47
      mean: 2.82978723404
      sd: 2.06754579294
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Pos_data_post_removal.csv; high
          column: Personal Val 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.813209200727
      v: 0.0306234050604
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e7
    subgroup: 'Experiment 2: Evidence-seeking design -- Paramedic -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Paramedic Low vs Paramedic 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 35
      mean: 2.25714285714
      sd: 2.06287716185
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Paramedic Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 35
      mean: 2.14285714286
      sd: 1.83339699406
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Paramedic 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0585626171598
      v: 0.0377444913971
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e8
    subgroup: 'Experiment 2: Evidence-seeking design -- Vaccine -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vaccine Low vs Vaccine 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 34
      mean: 2.70588235294
      sd: 1.9467052471
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Vaccine Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 34
      mean: 3.79411764706
      sd: 3.64134017757
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Vaccine 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.372724636385
      v: 0.034404491619
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e9
    subgroup: 'Experiment 2: Evidence-seeking design -- Mountaineering -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mountaineering Low vs Mountaineering 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 36
      mean: 1.86111111111
      sd: 1.26835726778
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Mountaineering Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 36
      mean: 2.47222222222
      sd: 1.88961237312
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Mountaineering 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.379749242052
      v: 0.0431311478436
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e10
    subgroup: 'Experiment 2: Evidence-seeking design -- Game show -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Game Show Low vs Game Show 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 33
      mean: 8.15151515152
      sd: 10.5478706741
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Game Show Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 33
      mean: 8.0
      sd: 8.15858443604
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Game Show 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0160686971955
      v: 0.0283223978536
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e11
    subgroup: 'Experiment 2: Evidence-seeking design -- Introduction -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Introductions Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 32
      mean: 1.65625
      sd: 0.970845158911
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Introductions Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 32
      mean: 2.09375
      sd: 1.44488809702
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.355430263187
      v: 0.0448463453658
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s3_e12
    subgroup: 'Experiment 2: Evidence-seeking design -- Possessions/Arson -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Pvalue Low vs Pvalue 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 26
      mean: 2.15384615385
      sd: 1.61721508013
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; low
          column: Pvalue Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 26
      mean: 3.11538461538
      sd: 2.67322910469
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2_Evidence_Seeking_Design/x2_ES_Neg_data_post_removal.csv; high
          column: Pvalue 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.435233708389
      v: 0.0578572704137
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  notes: Effects are scenario-by-polarity lowest-vs-highest stakes contrasts recovered from raw data; evidence-seeking cleaning
    is documented in analysis/effect_sizes.qmd. Evidence-seeking d values in this YAML use the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
- study_id: 4
  label: 'Appendix IV: Symmetrical Experiment (follow-up evidence-seeking prompts)'
  language: English
  language_other: null
  objective: Follow-up evidence-seeking experiment using symmetrical prompts ('minimum' vs 'maximum') to reduce asymmetry
    between positive and negative polarities.
  sample:
    n_final: 105
    recruitment: mTurk
    recruitment_other: null
    compensation: money
    compensation_other: $1.75
    characteristics: 'MTurk sample: 45 females, 59 males; ages 21–65; additional exclusions include prior completion of the
      first evidence-seeking experiment.'
    mean_age: 37.09
    mean_age_prov:
      page: null
      quote: leaving a final sample of 105 participants (45 females, 59 males) between 21 and 65 years old (M = 37.09 years,
        SD = 10.67 years).
      tei_id: null
      table_ref: null
    provenance:
      page: null
      quote: Following screening procedures... leaving a final sample of 105 participants (45 females, 59 males) between 21
        and 65 years old (M = 37.09 years, SD = 10.67 years).
      tei_id: null
      table_ref: null
  design: null
  design_other: Stakes (4 levels) within-subjects; prompt polarity (positive vs negative) between-subjects; prompts modified
    to use minimum/maximum wording.
  manipulated_factors:
  - 'Prompt polarity: evidence-seeking positive vs evidence-seeking negative'
  - 'Prompt wording: minimum vs maximum (symmetrical prompts)'
  paradigm: Rating how much evidence is needed for knowledge
  paradigm_other: null
  scale:
    label: other
    points: null
    anchors: Numeric free response (whole number of checks); positive prompt allows '0' and both prompts allow 'never'.
    direction: Higher numbers indicate more evidence required for knowledge / more checking still insufficient for knowledge.
    provenance:
      page: null
      quote: positive... "minimum numbers of times" ... negative... "maximum number of times"... enter a whole number... If
        you think Elaine knows without having to check, write "0"... If you think Elaine will never know... write "never".
      tei_id: null
      table_ref: null
  measures:
    knowledge_question_text: What is the minimum number of times S needs to check F before she knows that P?
    knowledge_question_first: null
    additional_question_text: What is the maximum number of times S can check F and not know that P?
  scenarios: []
  effects:
  - effect_id: s4_e1
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Paramedic -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Para Low vs Para 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 47
      mean: 2.23404255319
      sd: 1.59090849021
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Para Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 47
      mean: 3.0
      sd: 1.94489297794
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Para 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.431103113867
      v: 0.0179003744693
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e2
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Vaccine -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vacc Low vs Vacc 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 44
      mean: 2.86363636364
      sd: 2.25770936695
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Vacc Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 44
      mean: 4.70454545455
      sd: 3.57367341108
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Vacc 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.615892037585
      v: 0.0207344359567
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e3
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Mountaineering -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mount Low vs Mount 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 35
      mean: 2.82857142857
      sd: 1.20014004785
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Mount Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 35
      mean: 3.6
      sd: 1.76901434836
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Mount 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.510345860764
      v: 0.0244069491794
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e4
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Game show -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Game Low vs Game 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 41
      mean: 4.0
      sd: 4.28368999812
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Game Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 41
      mean: 6.60975609756
      sd: 5.91978905359
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Game 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.505090368385
      v: 0.0184715642471
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e5
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Introduction -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 45
      mean: 2.22222222222
      sd: 1.18492210885
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 45
      mean: 2.46666666667
      sd: 1.17936808966
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.206779836189
      v: 0.0220955487438
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e6
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Possessions/Arson -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Pval Low vs Pval 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 46
      mean: 1.91304347826
      sd: 1.17049062755
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          low column: Pval Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 46
      mean: 3.34782608696
      sd: 2.43326384654
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Pos_data_post_removal.csv;
          high column: Pval 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.751472126025
      v: 0.0302478033315
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e7
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Paramedic -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Para Low vs Para 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 30
      mean: 1.83333333333
      sd: 0.912870929175
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Para Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 30
      mean: 2.03333333333
      sd: 1.09806517404
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Para 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.198074155086
      v: 0.0460036063285
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e8
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Vaccine -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vacc Low vs Vacc 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 30
      mean: 3.36666666667
      sd: 2.93002687211
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Vacc Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 30
      mean: 3.43333333333
      sd: 2.7377732373
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Vacc 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.0235111844073
      v: 0.033935712758
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e9
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Mountaineering -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mount Low vs Mount 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 39
      mean: 2.15384615385
      sd: 1.91309148457
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Mount Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 39
      mean: 2.84615384615
      sd: 2.39009426745
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Mount 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.319806387682
      v: 0.0156962540102
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e10
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Game show -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Game Low vs Game 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 23
      mean: 3.60869565217
      sd: 2.5179199647
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Game Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 23
      mean: 4.95652173913
      sd: 3.06710199121
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Game 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.480340771592
      v: 0.058763399638
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e11
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Introduction -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 34
      mean: 1.97058823529
      sd: 1.19304281509
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 34
      mean: 2.29411764706
      sd: 1.73256530359
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.217503158496
      v: 0.0100083057383
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s4_e12
    subgroup: 'Appendix IV: Symmetrical evidence-seeking experiment -- Possessions/Arson -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Pval Low vs Pval 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 30
      mean: 1.56666666667
      sd: 0.727932041795
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          low column: Pval Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 30
      mean: 2.2
      sd: 1.88277125169
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.1_Symmetrical_Experiment/x2.1_Sym_Neg_data_post_removal.csv;
          high column: Pval 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.443709626053
      v: 0.0539589673019
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  notes: Effects are scenario-by-polarity lowest-vs-highest stakes contrasts recovered from raw data; evidence-seeking cleaning
    is documented in analysis/effect_sizes.qmd. Evidence-seeking d values in this YAML use the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
- study_id: 5
  label: 'Appendix IV: Matched Experiment (follow-up; ''0'' option removed)'
  language: English
  language_other: null
  objective: Follow-up evidence-seeking experiment with matched prompts by removing the '0' response option from the positive
    polarity prompts.
  sample:
    n_final: 89
    recruitment: mTurk
    recruitment_other: null
    compensation: money
    compensation_other: $1.75
    characteristics: 'MTurk sample: 33 females, 56 males; ages 20–70; positive polarity N=45, negative polarity N=44.'
    mean_age: 34.71
    mean_age_prov:
      page: null
      quote: leaving a final sample of 89 participants (33 females, 56 males) between 20 and 70 years old (M = 34.71 years,
        SD = 10.84 years).
      tei_id: null
      table_ref: null
    provenance:
      page: null
      quote: leaving a final sample of 89 participants... Participants were randomly assigned to a positive polarity condition
        (N = 45) or a negative polarity condition (N = 44).
      tei_id: null
      table_ref: null
  design: null
  design_other: Stakes (4 levels) within-subjects; prompt polarity (positive vs negative) between-subjects; '0' response option
    removed to create matched design.
  manipulated_factors:
  - 'Prompt polarity: evidence-seeking positive vs evidence-seeking negative'
  - 'Response options: ''0'' removed in positive prompts'
  paradigm: Rating how much evidence is needed for knowledge
  paradigm_other: null
  scale:
    label: other
    points: null
    anchors: Numeric free response (whole number of checks); '0' option removed; both prompts allow 'never'.
    direction: Higher numbers indicate more evidence required for knowledge / more checking still insufficient for knowledge.
    provenance:
      page: null
      quote: the additional option to write "0"... was removed... Modified enter a whole number... If you think Elaine will
        never know... write "never"
      tei_id: null
      table_ref: null
  measures:
    knowledge_question_text: 'Modified: enter a whole number... (no ''0'' option).'
    knowledge_question_first: null
    additional_question_text: null
  scenarios: []
  effects:
  - effect_id: s5_e1
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Paramedic -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Para Low vs Para 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 43
      mean: 2.02325581395
      sd: 1.53511861017
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Para Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 43
      mean: 3.90697674419
      sd: 2.67095447081
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Para 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.864738958369
      v: 0.029921005076
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e2
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Vaccine -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vacc Low vs Vacc 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 34
      mean: 3.5
      sd: 2.27303028283
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Vacc Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 34
      mean: 5.73529411765
      sd: 3.04818697758
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Vacc 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.831369146354
      v: 0.0373725222341
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e3
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Mountaineering -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mount Low vs Mount 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 26
      mean: 2.65384615385
      sd: 1.01753850806
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Mount Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 26
      mean: 3.92307692308
      sd: 1.59807576599
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Mount 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.947446942608
      v: 0.049281922459
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e4
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Game show -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Game Low vs Game 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 40
      mean: 2.9
      sd: 3.00256300773
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Game Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 40
      mean: 10.25
      sd: 9.14204151582
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Game 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -1.08022663846
      v: 0.0370733112632
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e5
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Introduction -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 41
      mean: 2.0487804878
      sd: 1.11694269128
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 41
      mean: 2.51219512195
      sd: 1.22723166557
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.394938706522
      v: 0.0250901114773
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e6
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Possessions/Arson -- positive polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; positive polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Pval Low vs Pval 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 41
      mean: 1.9512195122
      sd: 1.67259109636
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          low column: Pval Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 41
      mean: 2.90243902439
      sd: 2.16569709388
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Pos_data_post_removal.csv;
          high column: Pval 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.491607487397
      v: 0.0426130224156
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e7
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Paramedic -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Paramedic; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Paramedic: lives... Megan, a paramedic...'
          tei_id: null
          table_ref: null
        reason: Scenario is a paramedic accident/GPS vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Over the radio, Megan is told that there is one person at the scene of the accident...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (she is told the consequences).
      evidence:
        provenance:
          page: null
          quote: Megan... has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/navigation (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Megan is familiar with the surrounding area, she has GPS on her phone that she can check if necessary...
          tei_id: null
          table_ref: null
        reason: Familiarity plus GPS are ordinary fallible navigation resources, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Para Low vs Para 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 18
      mean: 3.33333333333
      sd: 1.60879933308
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Para Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 18
      mean: 3.55555555556
      sd: 1.42342677748
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Para 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.146300512989
      v: 0.0952413634855
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e8
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Vaccine -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Vaccine; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Vaccine: lives... medical researcher... vaccine...'
          tei_id: null
          table_ref: null
        reason: Scenario is a vaccine-checklist vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Elaine's assistant has informed her that there is one human research participant...
          tei_id: null
          table_ref: null
        reason: The vignette describes the stakes to the protagonist (assistant informs her).
      evidence:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take...
          tei_id: null
          table_ref: null
        reason: Evidence is the agent’s own checking/procedure-following (first-person).
      attribution_person:
        provenance:
          page: null
          quote: How many times does Elaine need to consult her check list before she knows that she is making the vaccine
            correctly?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Elaine... has a check list that specifies all of the steps she needs to take... Elaine is following all of
            the steps correctly.
          tei_id: null
          table_ref: null
        reason: The vignette states that the checklist specifies the steps and the agent is following them correctly, so reliability
          is coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Vacc Low vs Vacc 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 28
      mean: 2.67857142857
      sd: 2.03767426301
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Vacc Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 28
      mean: 3.85714285714
      sd: 2.91502221215
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Vacc 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.468635005912
      v: 0.0370716846331
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e9
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Mountaineering -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Mountaineering; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: High
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Mountaineering: personal injury... inspect the rope...'
          tei_id: null
          table_ref: null
        reason: Scenario is a mountaineering rope-inspection vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Visibility is reducing, making the climb increasingly dangerous...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the situation (dangerous climb; consequences).
      evidence:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own inspection/checking (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: How many times does S need to inspect the rope before she knows that it is tied securely?
          tei_id: null
          table_ref: null
        reason: The relevant evidence is repeated rope inspection for a directly inspectable condition, so reliability is
          coded High.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Mount Low vs Mount 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 18
      mean: 3.05555555556
      sd: 0.998364675929
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Mount Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 18
      mean: 3.44444444444
      sd: 1.33822631614
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Mount 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.329402378246
      v: 0.0855303160901
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e10
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Game show -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Game show; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Game show: finance... "What is the capital of Tanzania?"... "Dodoma"'
          tei_id: null
          table_ref: null
        reason: Scenario is a game-show trivia vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: As this is the final round of the game show, $1,000,000 is at stake...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (money at stake in the game show).
      evidence:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own memory/consideration (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to spend considering her answer before she knows that the capital of Tanzania
            is Dodoma?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Debra has recently read a list... and the city "Dodoma" pops into her head.
          tei_id: null
          table_ref: null
        reason: The evidence is memory of a recently read list without an additional check, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Game Low vs Game 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 26
      mean: 2.57692307692
      sd: 2.08178917133
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Game Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 26
      mean: 4.15384615385
      sd: 2.82407234599
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Game 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: -0.635638288618
      v: 0.0608271017613
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e11
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Introduction -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Introduction; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: External
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Introduction: reputation... wrote down the speaker''s name-"Dr. Woodbridge"-in her notebook...'
          tei_id: null
          table_ref: null
        reason: Scenario is a guest-speaker introduction/name-check vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: If Nicole introduces the guest speaker by the wrong name... reflect very badly...
          tei_id: null
          table_ref: null
        reason: Stakes are described within the scenario (embarrassment/reputation consequences).
      evidence:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Evidence is from an external written source (notebook).
      attribution_person:
        provenance:
          page: null
          quote: + How many minutes does S need to check her notebook before she knows that the guest speakers name is "Dr.
            Woodbridge"?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: Siena wrote down the speaker's name-"Dr. Woodbridge"-in her notebook earlier in the day.
          tei_id: null
          table_ref: null
        reason: The evidence is a notebook record made earlier in the day, an external written record but not an official
          or independently verified source, so reliability is coded Medium.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Intro Low vs Intro 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 36
      mean: 2.58333333333
      sd: 1.87273672927
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Intro Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 36
      mean: 2.58333333333
      sd: 1.64533973912
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Intro 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.0
      v: 0.039714932657
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  - effect_id: s5_e12
    subgroup: 'Appendix IV: Matched evidence-seeking experiment -- Possessions/Arson -- negative polarity'
    subgroup_desc: Lowest vs highest stakes; Possessions/Arson; negative polarity; raw-data recovered.
    design: Within-Subjects
    design_other: Lowest and highest stakes are repeated measures within the polarity-specific sample.
    moderators:
      scenario: other
      skeptical_pressure: 'No'
      awareness: 'Yes'
      evidence: First Person
      attribution_person: Other
      evidence_reliability: Medium
    moderators_coding:
      scenario:
        provenance:
          page: null
          quote: 'Arson: personal value... Only a functioning sprinkler system can stop a fire set by an arsonist.'
          tei_id: null
          table_ref: null
        reason: Scenario is an arson/sprinkler-system vignette; not bank/peanuts/bridge/typos.
      skeptical_pressure:
        provenance:
          page: null
          quote: How many times can Elaine consult her check list and still not know that she is making the vaccine correctly?
          tei_id: null
          table_ref: null
        reason: No explicit doubt or counterconsideration is introduced in the vignette; prompt polarity is represented as
          a separate effect.
      awareness:
        provenance:
          page: null
          quote: Natalie is living in an area where there have been a series of fires set by arsonists recently.
          tei_id: null
          table_ref: null
        reason: The risk/stakes are described within the scenario context.
      evidence:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: Evidence is from the agent’s own checking/inspection (first-person).
      attribution_person:
        provenance:
          page: null
          quote: + How many times does S need to check the sprinklers before she knows that they are working in the X room?
          tei_id: null
          table_ref: null
        reason: Prompt concerns whether the protagonist knows (third-person attribution).
      evidence_reliability:
        provenance:
          page: null
          quote: A week ago, Natalie checked that the sprinklers were working...
          tei_id: null
          table_ref: null
        reason: A week-old sprinkler check is treated as ordinary past verification projected forward, so reliability is
          coded Medium for current harmonization.
    contrast:
      group_high: stakes4
      group_low: stakes1
      sign_convention: d = mean(low) - mean(high)
      other_notes: 'Lowest vs highest stakes only; intermediate stakes levels are not included in this extracted effect. Source
        columns: Pval Low vs Pval 3. Evidence-seeking d in YAML uses the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.'
    groups:
    - group_id: stakes1
      label: Lowest stakes
      n: 27
      mean: 2.85185185185
      sd: 2.31556113325
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          low column: Pval Low.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    - group_id: stakes4
      label: Highest stakes
      n: 27
      mean: 2.2962962963
      sd: 2.0156086085
      se: null
      provenance:
        page: null
        quote: 'Computed in analysis/effect_sizes.qmd from Exp2.2_Matched_Experiment/x2.2_Matc_Neg_data_post_removal.csv;
          high column: Pval 3.'
        tei_id: null
        table_ref: analysis/effect_sizes_raw_data.csv
    effect_size:
      metric: SMD
      d: 0.255925346834
      v: 0.0544689106788
      computed_from: groups
      needs_review: false
      notes: Computed from post-removal raw CSV in analysis/effect_sizes.qmd; blank/'never'/non-positive responses were excluded
        and log-MAD outlier removal was applied over the scenario's four stakes cells. YAML d uses the raw low-minus-high evidence-seeking convention; downstream meta-analysis reverses evidence-seeking effects programmatically.
    quality_flags:
    - raw_data_recovered
    - evidence_seeking_cleaning_rule
    notes: Published article reports omnibus four-level stakes analyses rather than this exact low-high, polarity-specific
      contrast; contrast recovered from open raw data. Evidence-seeking d values use the raw low-minus-high scale; downstream harmonization reverses evidence-seeking effects programmatically.
  notes: Effects are scenario-by-polarity lowest-vs-highest stakes contrasts recovered from raw data; evidence-seeking cleaning
    is documented in analysis/effect_sizes.qmd. Evidence-seeking d values in this YAML use the raw low-minus-high convention; downstream meta-analysis reverses evidence-seeking effects programmatically.