Repository navigation
Releases
Releases · dvbn/cdid
cdid 1.3.0.0
cdid 1.3.0.0 expands the treatment designs you can estimate with complete or incomplete panels.
- Treatment can stop and restart. Use
att_gt(dname = ...)to estimate immediate effects at treatment entry and exit under no carryover. - More ways to compare adoption cohorts. Use linear cohort-by-date fixed effects or connect treated and control cohorts observed over different date pairs. The method comparison shows how to match specifications in other R and Stata packages.
- Inspect the models behind a fit. Fit with
control = list(keep_nuisance = TRUE)and usenuisance_models()to see the equations, coefficients and standard errors for the outcome, treatment, observation and imputation models available in your fit. - Keep usable comparisons in rotating panels. When a required reference-date covariate is missing, the fit warns and excludes the unit from the affected cohort's comparisons. The unit can still contribute to other cohorts. This resolves #16.
- Less repeated computation. The package reuses comparison inputs, fitted models and influence-function calculations across comparisons where their inputs agree.
Install from GitHub with remotes::install_github("dvbn/cdid@v1.3.0.0").
cdid 1.2.0.1 (archive)
Archived source for cdid 1.2.0.1, the version published on CRAN before the 1.3.0.0 GitHub release.
Install this version from GitHub with remotes::install_github("dvbn/cdid@v1.2.0.1"). Changelog