Bayesian Cohort-Period Modeling

About This Project¶

This project develops a hierarchical Bayesian model for analyzing period-cohort effects in General Social Survey (GSS) data. The model uses Gaussian Random Walk of order 2 (RW2) priors to provide principled smoothing of cohort and period effects, addressing limitations of previous approaches that used ad hoc smoothing and binned cohorts.

Contents¶

• Happiness in America - Cohort and period effects in GSS happiness data

• Trust and Well-Being - Cohort and period effects in GSS generalized trust

• No Religious Preference - Cohort and period effects for reporting no religious preference (GSS relig)

• Abortion opinion — deadlock or replacement? - Period and cohort effects on GSS legal-abortion items (abdefect–abany)

• End-of-life opinion — physician aid and suicide rights - Draft: letdie1 and GSS suicide-rights battery (suicide1–suicide4)

• Technical Report - Complete technical documentation of the model, implementation, and results

Repository Structure¶

• notebooks/: Model implementation notebooks

• scripts/: Batch processing scripts

• tables/: Output tables and trajectory data

• logs/: Model execution logs

• figs/: Generated figures

• jb/: JupyterBook source files