Dataset and Code Files for When Evidence Is Not Enough
This repository contains the de-identified data and analysis code required to reproduce the results reported in "When Evidence Is Not Enough: Experimental Effects on Health Misinformation Endorsement" (N=3,200, between-subjects randomized experiment, 4 conditions: Control, Scientific Evidence, Framing Effect, Social Contagion). Contents: replication_data.csv — de-identified dataset (N=3,200) containing the 15 outcome variables (agreement scores, Q1–Q15), experimental condition, and the sociodemographic, ideological, and attitudinal covariates used in all models reported in the manuscript.replication_code.R — complete R analysis script reproducing all figures (Figs. 2–7) and supplementary tables in the manuscript: descriptive statistics, ANOVA, Multi-Arm Causal Forest (ATE/CATE) estimation, heterogeneity tests by sex and by continuous covariates (best linear projections), and main-effect models of baseline endorsement. Software: R (≥4.4), packages tidyverse, grf, broom, patchwork, mgcv. Data license note: fully de-identified; no direct identifiers or high-reidentification-risk variables are included