Renzo Comolli
>
> I would like to test that the log of income is normal.
> I am familiar with -sfrancia- and -swilk-
> But my variable (income) comes in categories 0-5000
> 5000-10,000 ... More than 25,000
>
> I read
> Patrick Royston "A Toolkit for Testing for Non-Normality in
> Complete and
> Censored Samples" Statistician, Vol. 42, No. 1. (1993), pp. 37-43
> Which provides the solution from the theoretical point of
> view (although
> you have to combine the case for groups with the case for
> the open ended
> "more than").
>
> Do you know whether anybody has implemented Royston's solution in an
> .ado file?
> I used -findit- but it does not look like it
A check with Patrick Royston (who
is not a member of Statalist) indicates
that he has no such program up his sleeve.
You could play with a chi-square using frequencies
observed and expected if lognormal.
That would need assumptions about mean and SD of
log income, so far as I can see, and if you
guessed those parameter values would have
severe limitations as a significance test.
But it might signal something of interest.
Nick
[email protected]
*
* 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/
