Data and R code for: Does Greenwashing Undermine Green Purchase Intention? A Meta-Analysis of Green Trust, Skepticism, and Consumer Response

Description

This dataset contains the complete coded database and analysis code underlying the meta-analysis “Does Greenwashing Undermine Green Purchase Intention? A Meta-Analysis of Green Trust, Skepticism, and Consumer Response.” The synthesis covers 63 peer-reviewed studies reporting 70 independent consumer samples (total N = 27,567; 27 countries; 2013–2026) and 110 effect sizes linking consumer-perceived greenwashing to green trust, green skepticism, and green purchase intention. Records were identified through a Scopus search executed on August 17, 2026, and screened against pre-specified eligibility criteria described in the article. File 1 — Greenwashing_Meta_Coded_Database.xlsx (9 sheets). README: sheet guide. Samples: one row per independent sample (70 rows) with country, economy type (IMF World Economic Outlook classification), product category, sample type, study design, publication year, sample size, and the reliability of the greenwashing measure. Effects: all 110 coded effect sizes, each with the exact table or page location of the source value, the statistic as reported, the conversion method applied, and the final Pearson r and Fisher z. Pooled, Bias, Moderators, and StructuralModel: the numeric results behind Tables 2–5 of the article. Exclusions: all 138 full-text exclusions with reasons. Audit: adjudication decisions on signs, conversions, and the duplicate-sample exclusion. File 2 — Analysis_Package_Greenwashing_Meta.zip. Input data (analysis_data.csv; samples.csv), the main analysis script (analysis.R: REML random-effects models with 95% confidence and prediction intervals, heterogeneity statistics, three-level robustness models, Egger’s regression test, trim-and-fill, PET-PEESE, leave-one-out and influence diagnostics, mixed-effects subgroup and meta-regression moderator analyses, and a two-stage structural model on the pooled correlation matrix with Monte Carlo confidence intervals from 20,000 draws), an optional strict two-stage TSSEM replication script (metasem_replication.R), figure-generation scripts (fig_color.R; make_figs.py), the complete console log of the analysis run (R_output_log.txt), the numeric outputs behind Tables 2–5, the 300-dpi article figures, the exclusion log, the adjudication log, bibliographic records of all included studies (appendix_refs.json), and a README manifest. Screening and coding were performed with a large-language-model-assisted pipeline in two independent structured passes, with every included value re-verified against the source text; because each coded value carries its exact source location, all data can be independently re-verified against the published articles. The dataset contains only coded numeric data and bibliographic metadata; no copyrighted full texts are included. The files permit full reproduction of every statistic, table, and figure in the article, as well as re-analysis under alternative inclusion decisions documented in the Audit sheet.

Steps to reproduce

Requirements: R 4.3.3 with metafor 4.4.0 (the optional metasem_replication.R additionally requires the metaSEM package); Python 3 with matplotlib for Figures 1, 2, and 5. Unzip Analysis_Package_Greenwashing_Meta.zip into a single directory and set it as the working directory. Run analysis.R to reproduce all pooled, publication-bias, moderator, and structural-model results (it rewrites out_tab2_pooled.csv, out_tab3_bias.csv, out_tab4_moderators.csv, out_tab5_masem.csv, and R_output_log.txt). Run fig_color.R to redraw the forest and contour-enhanced funnel plots, and make_figs.py to redraw Figures 1, 2, and 5. The Monte Carlo procedure uses a fixed random seed, so results replicate exactly. Cross-check any coded value against its source using the table/page provenance in the Effects sheet of Greenwashing_Meta_Coded_Database.xlsx.

Institutions

• Padjadjaran University

  West Java, Bandung

Categories

Licence