Naturally, I wasn't recommending inappropriate
transformations. If your variable is not proportions,
but percents, then the appropriate function call is
logit(`v' / 100). And so on.
Nick
[email protected]
Kit Baum
> Nick points out
>
> I have nothing to add to discussion of Newey-West, but just
> want to point out that mass transformation of a bunch of
> variables is just a -foreach- loop:
>
> foreach v of var <varlist> {
> gen logit_`v' = logit(`v')
> label var logit_`v' "logit `v'"
> }
>
> subject to small problems if variable names are very long.
> You mainly need to work out the <varlist>.
>
>
> Quite so, but automatically transforming variables by ln(x/(1-x))
> assumes that they are in a proper interval (such as inside (0,1)) in
> the first place. I'm not sure how Clive's approach, using the
> unemployment rate (a value always in excess of 2) works very well
> with a logit transformation. Being able to automatically generate a
> large number of variables with all missing values is probably of
> limited utility.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
