Purpose-in-Life Predicts Mortality but Does Not Buffer Stress: Analysis archive for HRS 2006-2022

This deposit archives the data manifest, analysis code, and analysis output for a study testing whether purpose-in-life moderates the association between discrete stress exposures and all-cause mortality in the Health and Retirement Study (HRS) panel 2006-2022. The submitted manuscript itself is held in the journal submission system pre-acceptance and is not included in this archive; it will be added in a subsequent deposit version upon journal acceptance. The analysis tests nine primary moderator specifications, eleven exploratory robustness specifications, and two exploratory chronic-state loneliness specifications, applies a within-respondent Mundlak trait-state decomposition to the purpose-mortality association, and decomposes pathway effects across eight stress-to-purpose specifications (six discrete-event specifications plus two chronic-state loneliness specifications). The primary finding is the systematic absence of buffer evidence across all twenty-two moderator specifications, paired with substantial robustness of the purpose-mortality main effect across trait, state, and lagged specifications. The exploratory pathway analyses show that within-person purpose is associated with chronic ambient psychosocial states (loneliness) but not with the discrete event transitions tested here. The pattern supports a chronic-versus-acute typology rather than a general purpose-erosion mechanism. Repository contents:- README.md (project landing)- ANALYSIS_NARRATIVE.md (chronological design documentation distinguishing primary, exploratory, and post-Sutin extensions)- DATA_MANIFEST.md (HRS variable list, sample construction, exclusion criteria)- POST_HOC_DISCLOSURE.md (explicit timing disclosure)- code/ (R and Python analysis scripts)- data/ (frozen R analysis objects)- output/ (figures, figure source data, supplementary results) HRS data are not included in this deposit; they are available through the University of Michigan HRS data portal (https://hrsdata.isr.umich.edu) under standard public-use access. All variables used in this analysis are from the public-use RAND HRS Longitudinal File and the public-use HRS Leave-Behind questionnaire modules. The deposit is made concurrently with manuscript submission for transparency and reproducibility. See POST_HOC_DISCLOSURE.md for explicit timing disclosure relative to analysis completion.