Skip to contents

Changelog

Source: NEWS.md

lwdidR 0.2.0

  • Added the large-N path of Lee & Wooldridge (2026a), activated via method = "ra", "ipw", or "ipwra". It estimates ATT(g,t) by cohort and period (regression adjustment / inverse-probability weighting / doubly-robust IPWRA), aggregates to the event-study WATT(r) path and Pre/Post averages, and reports wild cluster bootstrap standard errors with simultaneous (sup-t) confidence bands. New options: method, pre, never, attgt, ydot, reps, seed, level, nose.
  • Added plot() (event-study graph via ggplot2) and print()/summary() methods for the large-N result (class lwdid_large).
  • Influence-function machinery ported verbatim from Stata lwdid v2.4.2; the large-N estimates and inference are validated against Stata on lw_smoking and lw_walmart (see data-raw/stata-reference/).

lwdidR 0.1.2

  • Corrected citation to the published version: Lee, S. J., & Wooldridge, J. M. (2026), “Simple Transformation Approach to Difference-in-Differences Estimation for Panel Data”, Journal of Business & Economic Statistics, 1–27, doi:10.1080/07350015.2026.2683047. Fixed the first author’s name (previously shown incorrectly as “Lee, Y.”).
  • Added inst/CITATION and acknowledged the original Stata lwdid package (https://github.com/Soo-econ/lwdid) in the README and function documentation.
  • Corrected paper attribution for the Castle Doctrine example and the small-sample inference options (wild bootstrap, permutation): these come from the small-N paper Lee & Wooldridge (2026b), “Simple Approaches to Inference with Difference-in-Differences Estimators with Small Cross-Sectional Sample Sizes” (SSRN 5325686), Section 7.2 — not the JBES transformation paper. The package now cites both papers. Removed a reference to a non-existent “Table 7.2” (the Castle results appear in the text of Section 7.2) and corrected the cohort-effect equation reference to (7.10).

lwdidR 0.1.0

Initial release.

  • Implements Lee & Wooldridge (2026a, 2026b) DiD estimator via unit-specific pre-treatment transformations
  • Supports common-timing and staggered adoption designs
  • Transformation methods: demean, detrend, demeanq, detrendq
  • Standard errors: homoskedastic OLS, HC1, HC3, cluster-robust (all manual sandwich)
  • Staggered design: overall pooled ATT, cohort-specific ATT, and (g,r)-level effects
  • Bundles Castle Doctrine dataset for replication of paper Section 7.2
  • Vignette replicating the Castle Doctrine results in Lee & Wooldridge (2026b) Section 7.2