Hi,
I'm currently running regression models on cross-sectional data for which the outcome variables (I have two) are both clustered at zero. Of the 240 observations, around 100 are zero, while the others span from -60 to +40 for one dependent variable, and -2.5 to +2.5 (ordinal variable, iterations of 0.5) for the other dependent variable. Thus, the normality assumption is violated (as well as the homoskedasticity for one of the models). I will run both parametric tests (multiple regression) and nonparametric tests. To deal with the non-normality and heteroskedasticity, I've see suggestions to use vcs(robust) or vcs(bootstrap), or use rreg instead of reg. I'm not sure which to use, can anyone provide explanations of when to use which?
Thanks in advance,
Rens
Related Posts with Using rreg or vce(robust) for non-normal heteroscedastic standard errors
Summary StatisticsDear People, How do I get from the attached STATA file, to the attached output table? Kind regards…
Assumptions for a mediated model with multiple fixed effects regressionsDeal all, I am trying to study a mediating model with multiple fixed effects regressions and before…
Problem with calculating employment (unemployment) duration until first unemployment (employment)Hello everyone, I have trouble working with duration data. I am working with spell-type data. The …
interpreting categorical odds ratiosI ran a logistic regression to look at the difference of an outcome ( low birth weight infant) among…
question about a r(101) errorDear all, I get an r(101) error while running the following code (the error is "factor variables an…
Subscribe to:
Post Comments (Atom)
0 Response to Using rreg or vce(robust) for non-normal heteroscedastic standard errors
Post a Comment