Full-empirical-analysis-skill-R

What it does

Covers an 8-step empirical analysis pipeline in R, leveraging tidyverse and econometrics packages to import/clean data, construct variables, produce descriptive statistics and Table 1, run baseline models (fixest::feols, ivreg, did, eventstudyr, synthdid, rdrobust, MatchIt, WeightIt, grf, DoubleML, mediation), perform robustness checks, extend analyses (interactions, marginal effects, mediation, gsem, dose-response splines, grf CATE), and generate publication-ready tables/figures. It emphasizes reproducible scripts with renv and Quarto, and outputs multi-table reports (T1–T5) plus figures (F1–F4).

How it works

The skill prescribes an 8-step workflow:

1. Data import & cleaning using read_dta/read_csv and helpers; 2) Variable construction with mutate/across and lag/lead; 3) Descriptive statistics & Table 1 via gtsummary and modelsummary; 4) Classical diagnostic tests (shapiro, jarque.bera.test, bptest, dwtest, bgtest, vif, adf.test, kpss.test, Hausman); 5) Baseline modeling with fixest::feols and related methods (ivreg, did, eventstudyr, sun_ab, did_imputation, synthdid, rdrobust, MatchIt, WeightIt, grf::causal_forest, DoubleML, mediation); 6) Robustness battery (modelsummary stack, clubSandwich CRSE, fwildclusterboot, ri2, robomit Oster, bacondecomp, HonestDiD); 7) Further analysis (interactions, marginaleffects, mediation::mediate, gsem via lavaan, dose-response splines, grf CATE); 8) Publication-ready outputs via modelsummary, kableExtra, gt, stargazer, texreg, flextable and associated plotting via ggplot2, iplot, binsreg.

When to use it

Use when an applied economist wants a complete R-based empirical analysis pipeline yielding publication-ready results and tables, with the option to leverage modes for different domains (economics default; epidemiology and ML causal inference subsystems share the same steps but swap Step-5 estimators and Step-6/7 analyses).

What it can touch

The workflow touches data files read_dta/read_csv, variable construction and several modeling/diagnostic steps, and outputs to tables/figures (tables/table1_balance.xlsx, tables/table2_main.xlsx, figures/fig1_trend.png, etc.). It specifies usage of packages and functions such as feols, modelsummary, gtsummary, tbl_summary, sunab, ggdid, bacon, HonestDiD, rdrobust, MatchIt, WeightIt, cobalt, ebal, grf, DoubleML, mediation, marginaleffects, kableExtra, gt, stargazer, texreg, flextable, ggplot2.

Caveats

The description notes a license of NOASSERTION and relies on several external packages; specifics about licensing or risk are not itemized beyond that, and no outcomes are promised beyond the publication-ready outputs outlined (tables/figures).