diff: Simplifying the causal inference analysis with difference-in-differences
Most of the microeconometrics studies are being based on the causal inference analysis. diff provides to the researcher an easy-to-use tool to perform the difference-in-differences estimation from a two-period panel dataset designed for an impact evaluation. It combines the conditional independence of the outcome and the treatment given unobservable characteristics that do not vary over time with the use of observable covariates. diff is endowed with four estimands: the single diff-in-diff; diff-in-diff accounting for covariates; the kernel propensity score diff-in-diff, which allows the selection of the bandwidth, the use of probit or logit, the provision of the propensity score, and the estimation on the common support; and the quantile diff-in-diff at the specified quantile, which is also available for continuous outcomes and it is combinable with the kernel option. Finally, option test performs the balancing test of covariates between treatment and comparison groups in the base-line period, generating a simple table based on Stata’s ttest.