Changes in version 1.0.4 (2024-03-31) - Test suite changes for M1 mac stay current on CRAN. Changes in version 1.0.2 (2024-01-17) - Minor documentation changes to stay current on CRAN. Changes in version 1.0.0 (2022-07-04) - Version bump to coincide with DeclareDesign package version 1.0.0 - Tests edited Changes in version 0.30.6 (2022-01-31) - Fix tests to address CRAN failures Changes in version 0.30.4 (2021-11-07) - Bug fix of tidy handling of conf.level - Bug fix of lh_robust tidy Changes in version 0.30.2 (2021-01-17) - Remove lfe from tests Changes in version 0.30.0 (2021-01-07) - Test suite changes (skip if not installed for checking against other packages) Changes in version 0.28.0 (2020-11-19) - Test suite changes Changes in version 0.26.0 (2020-09-07) - Test suite changes Changes in version 0.24.0 (2020-09-02) - tidy: rename nobs, nclusters, nblocks - tidy: new arguments conf.int, conf.level - Added update.iv_robust() - Bug fix regarding fixed effects with large numbers Changes in version 0.22.0 (2020-03-19) - Bug fixes Changes in version 0.20.0 (2019-09-09) - Added support for emmeans (thanks @rvlenth)! - Fixed bug when estimating diagnostics in iv_robust() without explicitly specifying se_type (issue #310) - Support for rlang 0.4.0 Changes in version 0.18.0 (2019-05-26) - Fixed bug where collinear covariates caused fixed effects estimator to crash (issue #294) - Added glance.lh_robust() and fixed some issues with printing and summarizing lh_robust() objects (issues #295 and #296) - Fixes CRAN errors in testing with new clubSandwich package Changes in version 0.16.0 - Add diagnostics to iv_robust() - Add glance() methods for all estimators - Add lh_robust() for easy interface to car::linearHypothesis() - Fixed minor bug with a formula such as is.na(var) in the covariates formula in lm_lin() (issue #283) Changes in version 0.14.0 - Removes broom hack for tidy method and instead relies on importing generics Changes in version 0.12.0 - Fixed ambiguity about how interacted covariates were centered in lm_lin - A series of fixes for bugs that occurred with multiple outcomes (multivariate regression): - Fixed bug pointed out by James Pustejovsky via the sandwich version 2.5-0 and off-diagonal blocks of multivariate regression vcov matrices - Fixed bugs in lm_lin preventing multivariate regression - Fixed bug that truncated degrees of freedom with "CR2" standard errors - Fixed bug that returned incorrect R-squared for the second or later outcomes - Fixed bug preventing integration with latest version of margins - Fixed bug with difference_in_means when using condition1 and condition2 to subset a treatment vector with more than two treatment conditions. Previous estimates and standard errors were incorrect. Changes in version 0.10.0 (2018-07-12) - Changed names of confidence interval columns in tidied data from ci.lower and ci.upper to conf.low and conf.high to be in line with other tidy methods - Added support for fixed_effects that are just one block - Added support for specifing condition_prs in horvitz_thompson() as a single number - Added t- and z-statistics to output - Limit unnecessary messaging in horvitz_thompson() Changes in version 0.8.0 (2018-06-02) - Added support for absorbing fixed effects in lm_robust and iv_robust - Added commarobust and starprep for stargazer integration - Added texreg support for 2SLS IV models - Fixed bugs for incorrect F-statistics with robust standard errors - Refactor of main fitting engine for linear models Changes in version 0.6.0 (2018-03-28) - Added support for multivariate linear models - Added support for instrumental variables regression - Major change to name of object output elements to mostly match with broom::tidy - old -> new - coefficient_names -> term - se -> std.error - p -> p.values - ci_lower -> ci.lower - ci_upper -> ci.upper - All of the above changes are also made to the column names on the output of tidy; furthermore for tidy objects one further name change from coefficients -> estimate has been made - Fixed bug that caused variances, standard errors, and p-values to be wrong for weighted "CR2" variance estimation - Fixed incorrect estimates when both weights and blocks were passed to difference_in_means - Rewrite NSE handling to be done by rlang - Rewrite na.omit handler in R - Major refactor of C++ underlying regression estimators Changes in version 0.4.0 (2018-02-15) - Changed suffix added to centered variables in lm_lin() from _bar to _c - Added all vignettes to .Rbuildignore, only available on website now - Fixed lm_robust_helper.cpp algorithm to not catch own exception and to deal with valgrind memory errors - Bugfix where passing a formula as an object within a function would fail - Simplified some tests for various CRAN test platforms Changes in version 0.2.0 (2018-01-29) - First CRAN upload