← ClaudeAtlas

survey-instrument-verificationlisted

Manuscripts evolve terminology that may not match what was actually administered:
fabioc-aloha/Alex_Skill_Mall · ★ 4 · AI & Automation · score 74
Install: claude install-skill fabioc-aloha/Alex_Skill_Mall
# Survey Instrument Verification ## The Problem Manuscripts evolve terminology that may not match what was actually administered: ``` Manuscript says: "We measured perceived safety using a 5-point scale" Raw data shows: 7-point scale with different anchors Manuscript says: "Participants rated their agreement with 12 items" Raw data shows: 14 items (2 were dropped but not documented) ``` ## The Solution Before editing survey-related content, verify against raw data. ### 1. Check Scale Parameters | Manuscript Claims | Verify In Data | |-------------------|----------------| | Number of points | Actual min/max values | | Anchor labels | Codebook or data dictionary | | Number of items | Count of columns/variables | | Response options | Unique values in data | ### 2. Verify Item Wording ``` Manuscript Appendix: "How safe do you feel walking at night?" Actual survey item: "How safe do you feel walking alone at night?" ^^^^^ Missing word! ``` Compare exact wording: - Survey instrument file (if available) - Qualtrics/SurveyMonkey export - IRB-approved protocol ### 3. Check for "Exactly As Administered" Claims If an appendix claims "items shown exactly as administered": ```bash # Line-by-line comparison diff manuscript-appendix.txt original-survey.txt ``` This claim is a testable hypothesis — test it. ## Verification Checklist - [ ] Scale points match (5-point? 7-point? Likert?) - [ ] Anchor