Purpose: This study develops a model linking green advertising and environmental concern to green brand purchase intention through guilt and examines the relationships supported by the available archival data. Method: An exploratory secondary analysis was conducted using 60 respondent records from a recycled-paper advertising study involving university students in Yogyakarta. Advertising evaluation served as a limited proxy for green advertising, while guilt and purchase intention were reconstructed from conceptually relevant questionnaire items. Analyses included descriptive statistics, internal consistency assessment, regression with HC3 robust standard errors, and 10,000 bootstrap resamples. Sensitivity analysis incorporated an indicator distinguishing two respondent-order blocks. Finding: The pooled analysis produced an indirect association of 0.483, with a 95% confidence interval of 0.247–0.683. After accounting for respondent-order blocks, this association decreased to 0.075, with an interval of −0.019–0.183. These findings indicate that the mediation evidence was unstable. Environmental concern lacked explicit indicators, preventing estimation of its proposed relationships. Novelty: The study combines conceptual model development with transparent assessment of construct compatibility and sensitivity in archival data reuse. It demonstrates why strong pooled associations require further scrutiny before being interpreted as psychological mechanisms and identifies the measurements needed for subsequent testing of the complete model.