Stimulus Normativity in the Erasure-with-Confound Meta-Analysis: Results, Code and Output of a Preregistered Secondary Analysis

This record contains the results report, analysis code and output of a preregistered secondary analysis on stimulus normativity in the erasure-with-confound meta-analysis. The analysis tested whether the normativity of liking ratings (normativity index, NI) places the stimulus classes with a robust erasure-with-confound effect (colors, foods) at intermediate values relative to the classes without one (Chinese characters, pictures, tones).

The analysis was preregistered on OSF before any data were analyzed:
- OSF project: https://osf.io/e5dzg/overview
- Preregistration: https://osf.io/uagy8
- Amendment to the preregistration (filed before any analysis): https://osf.io/e5dzg/files/cgds6

Files:
1. Results_Report_NI.pdf – results report: data and procedure as executed, class-level and study-level NI estimates, and evaluation of the registered criterion (not met: Chinese characters and tones lie inside the color–food interval; only pictures lie outside it).
2. NI_load.py – loads and harmonizes the rating-level raw data of the 18 confounded limited-set studies.
3. NI_estimate.py – REML estimation of NI per study with parametric bootstrap confidence intervals and sensitivity estimates.
4. NI_pool.py – logit-transformed random-effects pooling per stimulus class and evaluation of the registered criterion.
5. NI_results_by_study.csv – NI, 95% CI and sensitivity estimates for each study.
6. NI_results_by_class.csv – pooled NI per stimulus class with 95% CI.
7. NI_results_figure.png – figure of the class-level results.