Alternatively, consider taking logs.
It may be that you just have a skewed
variable better examined on a log scale,
on which mean and median could well be closer,
relatively speaking, so that the apparent problem fades away.
Nick
[email protected]
Roger Newson replied to Carmine Ornaghi
> > I have to compare the mean and the median of a
> >variable for two groups of firms (acquired and
> >acquirer) but this variable is not normally
> >distributed (so that I cannot use ttest).
> >I know the command ksmirnov and median to compare the
> >distribution and the median but they just tell me that
> >they are not equal. When I look at the main statistics
> >(using "summarize, detail" for the two groups), I have
> >that the mean is grater in one group but the median is
> >lower. Any suggestion?
> You might use the program -cendif-, part of the -somersd-
> package, to
> calculate confidence intervals for median differences or
> ratios. The
> package is distributed with .pdf manuals. In Stata, type
> -ssc desc somersd-
> to find out more.
*
* 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/
