Changelog
Source:NEWS.md
undidR 3.0.2
- Computation of HCCMEs is now faster by using a crossproduct approach rather than constructing a full omega matrix.
- Changed the default HCCME from HC3 to HC1 (i.e.,
hc = 1). - Fixed an issue where standard errors were returning
NAfor certain edge cases.
undidR 3.0.1
CRAN release: 2026-02-16
- Added the
lwdandlegendparameters to the S3plot()method forUnDiDObjwhich controls line width and legend locations, respectively. Removed the default titles of plots. - Also made other minor aesthetic changes to the default plots, such as adding horizontal bars to the ends of the event study plot confidence bands (like the letter I).
- Removed “none” as an aggregation option for staggered adoption.
undidR 3.0.0
CRAN release: 2026-01-26
-
Added new aggregation methods for staggered adoption:
"sgt","time", and"none".-
"sgt"computes subaggregate ATTs for each treatment-time and post-treatment period pair, further separated by silo. -
"time"computes subaggregate ATTs grouped by time since treatment. -
"none"skips subaggregate ATT computation and directly computes the aggregate ATT from the underlying differences.
-
-
undid_stage_three()now returns an object of classUnDiDObj, providing a unified S3 interface. Added support for event study plots via the
plot()method forUnDiDObjobjects.Added randomization inference p-values for subaggregate ATTs.
Fixed a miscalculation of jackknife standard errors and p-values.
-
Expanded weighting options for ATT computation:
"none","diff","att", and"both".-
"diff"applies weights when computing subaggregate ATTs. -
"att"applies weights when aggregating subaggregate ATTs. -
"both"applies weighting at both stages. -
"none"disables weighting entirely.
-
undidR 2.0.0
CRAN release: 2025-05-31
- start_time and end_time columns are now stored as strings in the format specified by date_format column in the empty_diff_df CSV file.
- Fixed
as.Date()calls inundid_stage_two(),undid_stage_three(), andplot_parllel_trends()to ensure compatability for R versions 4.0.0 and onwards.