Causal inference using STATA
This work has two main objectives: first to provide a short overview of available analytical methods that estimate Causal Effect measures when “association is not causation” and then to introduce a set of programs which estimate them. The methods used are: Outcome Regression adjustment, Inverse Weig...
Guardado en:
| Autor principal: | |
|---|---|
| Otros Autores: | |
| Formato: | Tesis de maestría acceptedVersion |
| Lenguaje: | Español |
| Publicado: |
Universidad Torcuato Di Tella
2017
|
| Materias: | |
| Acceso en línea: | http://repositorio.utdt.edu/handle/utdt/1334 |
| Aporte de: |
| Sumario: | This work has two main objectives: first to provide a short overview of available analytical methods that estimate Causal Effect measures when “association is not causation” and then to introduce a set of programs which estimate them. The methods used are: Outcome Regression adjustment, Inverse Weighted probability, Double Robust bounded and Stratification by the propensity score. In order to implement such methods we have developed five programs using STATA software for both continuous and binary outcomes. When the outcome variable is binary the programs outputs estimators of the Average Treatment effect (ATE), the Causal Risk ratio (CCR) and the Causal Odd ratio (COR) while if the outcome variable is continuous it only outputs the ATE. In addition we constructed a special program (prop_score.ado) for the evaluation of the propensity score fit in order to use it in the propensity score stratification method. These programs are: t_out_reg.ado, t_ipw.ado, t_prop_stat.ado, the dr_bounded.ado and the t_prop_score.ado. |
|---|