Robust estimators for ANOVA?

[Emma Davies]

I want to compare questionnaire scores across three groups of participants, however, the questionnaire scores are positively skewed. I've been advised to use a robust estimator instead of transforming the data, but there doesn't seem to be an option for this with the oneway or anova command? Does anybody have advice on how I might do this?

Many Thanks!

[Nick Cox]

I'd consider using qreg with suitable predictors.

[Carlo Lazzaro]

Emma:
as a sidelight to Nick's superb advice: why focusing on -anova- when -regress- has so many more flavours?

[Emma Davies]

because I'm quite new to the world of statistics (and stata)! Forgive me for my lack of knowledge! Someone has advised the glm will allow me to use the robust estimator I want, but the glm and anovas (without the robust estimator) are giving me different p values! I thought they were meant to be the same? Is there something wrong with my code:

glm questionnaire group if baseline

anova questionnaire group if baseline

thanks for all your help!!! I really appreciate it.

[Carlo Lazzaro]

Emma:
as per FAQ, please post what you typed and what stata gave you back. Thanks.
Eventually, I'm still not clear with what you mean by robust (to what?).

[Nick Cox]

Two quite different senses of robust here, (a) robustness to (especially) long tails or wild values in data and (b) robust standard errors. You need to read up the difference and work out what you want. I suspect you want (a) and are being side-tracked by (b).

See e.g. http://home.uchicago.edu/~lhansen/ch...ss_stigler.pdf