Re: st: Averaging across variables

[David Kantor <dkantor@jhu.edu>]

I should add some warnings to what Ying wrote in reply to Kelly Johnson's 
inquiry.

The command
 egen avginc = rmean(inc78 - inc80)

depends on having the desired variables ordered such that they are covered 
by the varlist
 "inc78 - inc80".  That is, it depends on the present order of the 
variables. (Stata will *not* try to infer that this includes inc79 based on 
the *form* of the varlist expression.) It is generally not a good idea to 
write code with such a dependence.  It would be better to explicitly spell 
out the variables, or use star notation, such as rmean(inc*) -- if inc* 
covers exactly the variables you want.

I often put the varlist into a macro:
 local incvars "inc78 inc79 inc80"
 egen avginc = rmean(`incvars')

(You can also put a spelled-out version of a star-notation varlist into a 
macro using -unab-, and then manipulate the  resulting macro with macrolist 
operation -- such as deleting specific variables.)

Finally, be sure to use rmean, and not mean.
 -egen ... rmean()-  takes a varlist and computes a row mean.
 -egen ... mean()-  takes an expression and computes a column mean.

Thus,
 egen avginc = mean(inc78 - inc80)
looks almost the same as the rmean version, but is vastly different. It 
will take the difference inc78 - inc80 (the result of subtracting), and 
yield the column sum of that difference.

HTH
-- David

At 09:30 AM 7/14/2005 -0400, you wrote:

> Hi,
>
> I am not sure whether this is what you want. Suppose you have income
> varialbe named inc78, inc79 and inc80. You created a new variable called
> avginc which is the mean of (inc78 - inc80).  You use this command.
>
> egen avginc = rmean(inc78 - inc80)
>
> Do a search on Stata by "egen".
>
> Good luck.
>
> Ying
>
>
> [from Kelly Johnson]> hi folks,
> >
> > suppose i have a data set with say, many variables.
> >
> > how can i create a new variable that is the (arithmetic) mean of all the
> > variables starting with certain letters (i.e. using the '*' operator)?
> >
> > thanks again!
> >
> > kj
> >
> [...]
David Kantor
Institute for Policy Studies
Johns Hopkins University
dkantor@jhu.edu
410-516-5404

*
*   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/