Hi all,
I'm constructing a table that shows differences in baseline characteristics between four study groups, before and after probability weighting. I've used chi-square tests for categorical variables using -svyset- and can get p-values for group differences before and after probability weighting.
However, I am struggling to figure out what to do with the two continuous variables I have remaining. The study I've modelled my design after has two groups and used the Wilcoxon Rank-Sum Test. This test, however, doesn't work with more than two groups and I have four. The Kruskal-Wallis test seems to be the multiple group equivalent to the Wilcoxon, but Stata's implementation doesn't work with probability weighted data (it also doesn't store p-values anywhere preventing the use of any of Stata 17's nifty table building functionality). My supervisor suggested looking at statistical tests that look at differences in means - I thought an ANOVA using -oneway- might be in order, but this, too, doesn't work with probability weights.
My question: what options are available to me in Stata to test for between study group differences with a continuous variable before and after applying probability of treatment weights?
Thank you!
I'm constructing a table that shows differences in baseline characteristics between four study groups, before and after probability weighting. I've used chi-square tests for categorical variables using -svyset- and can get p-values for group differences before and after probability weighting.
However, I am struggling to figure out what to do with the two continuous variables I have remaining. The study I've modelled my design after has two groups and used the Wilcoxon Rank-Sum Test. This test, however, doesn't work with more than two groups and I have four. The Kruskal-Wallis test seems to be the multiple group equivalent to the Wilcoxon, but Stata's implementation doesn't work with probability weighted data (it also doesn't store p-values anywhere preventing the use of any of Stata 17's nifty table building functionality). My supervisor suggested looking at statistical tests that look at differences in means - I thought an ANOVA using -oneway- might be in order, but this, too, doesn't work with probability weights.
My question: what options are available to me in Stata to test for between study group differences with a continuous variable before and after applying probability of treatment weights?
Thank you!
Comment