This codebook documents the complete R analysis pipeline used in a Master of Public Administration capstone study examining predictors of preventable custodial death across four Texas county jails Bexar, Dallas, Harris, and Travis from 2015 to 2025 (N = 390). All data were sourced from the Texas Office of Attorney General custodial death reporting records. The codebook is organized into nine sections covering the full analytical workflow: library setup, data loading and variable preparation, summary dataset construction, descriptive statistics, bivariate testing, correlation analysis, logistic regression modeling, data visualization, and a key results reference. All code is fully annotated and reproducible. Four logistic regression models are documented, including a demographics-only baseline model, a model adding mental health and custody time variables, a primary binary model coding Bexar County as an indicator variable, and a supplemental all-county comparison model with Bexar as the reference group. Significant predictors of preventable death identified in the primary model include Bexar County incarceration (OR = 3.303, p = .001), suicidal ideation (OR = 5.474, p = .006), single-cell housing (OR = 2.580, p = .004), and younger age (OR = 0.928 per year, p < .001). Six figures are included: preventable death rates by county, odds ratio forest plots for both primary and all-county models, and supporting graphs examining suicidal ideation, age distribution, and housing type by preventable death status. This analysis was conducted as part of a Master of Public Administration capstone project at the University of Texas at San Antonio. The full paper examines institutional drivers of preventable custodial death and develops policy recommendations for statewide reform of Texas county jail death prevention standards

This analysis examines institutional and individual-level predictors of preventable custodial death across four Texas county jail systems Bexar, Dallas, Harris, and Travis using data from 390 deaths recorded between 2015 and 2025. Data were sourced from the Texas Office of Attorney General custodial death reporting records and analyzed using binary logistic regression in R. Raw preventable death rates reveal striking county-level disparities: Travis County (27.9%) and Bexar County (27.7%) each classify more than one in four custodial deaths as preventable, while Dallas County (7.8%) and Harris County (8.6%) maintain substantially lower rates. Together, Bexar and Travis counties account for 70% of all preventable deaths in the sample despite representing 42% of total deaths. Two logistic regression models were estimated. The primary model coded Bexar County as a binary indicator variable and found that incarceration in Bexar County was associated with 3.3 times greater odds of preventable death (OR = 3.303, 95% CI [1.669, 6.676], p = .001) after controlling for age, sex, race, mental health status, suicidal ideation, and housing type. A supplemental all-county comparison model, with Bexar as the reference group, found that Dallas County inmates had 79% lower odds of preventable death (OR = 0.208, p = .005) and Harris County inmates had 70% lower odds (OR = 0.304, p = .004) compared to Bexar. Travis County did not differ significantly from Bexar (OR = 0.445, p = .119), confirming both counties share comparably elevated preventable death risk. At the individual level, suicidal ideation was the strongest predictor of preventable death (OR = 5.474, p = .006), followed by single-cell housing (OR = 2.580, p = .004) and younger age (OR = 0.928 per year, p < .001). Mental health status alone was not a significant predictor in the full model, suggesting that general documentation of mental health problems is insufficient without targeted institutional response protocols for acute crisis indicators. These findings provide empirical support for the argument that institutional practices not individual characteristics are the primary drivers of variation in preventable custodial death rates across Texas county jails. The existence of counties achieving preventable death rates below 9% demonstrates that current conditions in higher-rate counties represent institutional failures, not inevitable outcomes. This analysis was conducted as part of a Master of Public Administration capstone project at the University of Texas at San Antonio. The full paper examines policy implications and recommendations for statewide reform of Texas county jail death prevention standards.

Analysis of 390 custodial deaths across Texas's four largest county jails (2015-2025) reveals a 3.6 fold variation in preventable death rates, from 7.8% in Dallas County to 27.9% in Travis County. Suggesting that institutional factors, not individual characteristics, determine outcomes. This research provides evidence-based recommendations for statewide jail reform to prevent unnecessary deaths in custody.