This code restricts BRFSS 2011 to Texas respondents with a valid diagnosed-diabetes outcome and constructs predictor-specific analytic samples with appropriate recoding and missing-data handling. It then fits separate bivariate survey-weighted logistic regressions of diabetes on each predictor and reports sample sizes and tidy coefficient estimates.

This script loads Texas BRFSS 2011–2023, harmonizes age into consistent five-year groups, constructs a binary diagnosed-diabetes outcome, and runs survey-weighted logistic regressions of diabetes on age group for each year plus pooled periods. It then extracts the most statistically significant age-group coefficient per year (and per pooled period) and plots those log-odds estimates with 95% error bars, marking 2014 (ACA) and 2020 (COVID).

This code constructs a Texas-only BRFSS 2011 analytic sample, rigorously cleans and recodes health, behavioral, socioeconomic, and access-to-care variables, applies multiple imputation for missing covariates, and defines survey-weighted designs consistent with the BRFSS sampling structure. It then estimates a sequence of survey-weighted logistic regression models of diagnosed diabetes, progressively adding regional fixed effects, interaction terms, and nonlinear specifications to assess robustness and heterogeneity.

Statistical analysis (logistic regression, decision tree) on publicly accessible kaggle dataset regarding diabetes factors