**README**
**Task:** Evaluating the impacts of SYG laws across the US, 1999-2017
**Data:** CDC’S Vital Statistics on Multiple Causes of Death (MCOD), 1999-2017
**Author:** Michelle Degli Esposti
**Date:** 02 Feb 2021
**Data Steps:**
**STEP 1.** “01.dataprep&descriptives” includes cleaning, visualising, and describing MCOD data, 1999-2017 for: homicide, firearm homicide, suicide and firearm suicide.
Subfolder 1.5 includes imputed race data for Texas Sept 2007-Feb 2009 due to suspected errors in data coding for homicide and firearm homicide. Subfolder 1.6 includes exploratory modelling of non-linear trends at the national and state-level, identifying non-linear trends that are best fit using cubic polynomials in GLMMs. The main manuscript is based on imputed values. Note that if values are re-imputed then this will slightly impact new estimates for SYG law effects by race
**STEP 2.** “02.mainmodels” includes fitting the main models of the impacts of SYG laws on states across the US, 1999-2017 for: homicide, firearm homicide, suicide and firearm suicide. Main models are: GAMs (constrained and differential), GLMM-splines (linear, quadratic, cubic), GLMM polynomials (linear, quadratic, cubic), ITS plus meta-analysis (linear and cubic and a restricted 3 year linear analysis). These models all specify fixed effects of SYG laws. The analyses are based on 23 intervention SYG states.
NB: GLMM-polynomial with cubic terms for firearm homicide did not converge (quadratic term instead used)
*Figures from manuscript and supplementary material: Fig 1, Fig S6, Fig S10, Fig S11*
**STEP 3.** “03.equitymodels” includes fitting the equity models for examining the impacts of SYG across population subgroups (race, sex, age) for: homicide and firearm homicide including not imputed and imputed data to account for outlier Texas race data. Estimates are primarily based on stratified models while GAMs with interaction terms (difference-in-difference models) assess whether the impacts of SYG laws are conditional on race, age group, or sex. *Figures from manuscript: Fig 2*
