You are putting the blame in the wrong place,
I think.
-insheet- tries hard to convert your "labels", meaning
not Stata labels, but column headers in your file, into variable
names, but "-" is not legal within variable names
and if it is stripped the problem then will be that
"p-1" and "p1" cannot be distinguished. So it seems
that your variable names will get messed up well before
-reshape- ever sees them. The problem therefore
is the implicit assumption that -insheet- will break Stata's
rules to satisfy your preferences.
One work-around is this:
Your .csv file, it seems, has a first line something like this:
permid,p-2,p-1,p0,p1,p2
Change the "permid" to "id".
Change all the "-" signs to underscores "_". "_" is allowed
within variable names. You should be able to do this easily
in any text editor.
Now -insheet- as before.
. reshape long p, i(id) j(qtime) string
The change of name of -permid- to -id- stops -permid-
getting bundled in with variables whose names start
with -p-. The -string- option flags that the underscores
are OK, but time will be a string variable.
Now the underscores must be mapped back again:
replace qtime = subinstr(qtime,"_","-",1)
destring qtime, replace
and you should be nearer where you want to be.
Nick
[email protected]
M. Douglas Berg
> I have a comma seperated (csv) file with quarterly data on
> variable p.
> The column headings are of the form p0, p1, etc for 1960q1,
> 1960q2. For
> periods prior to 1960 I labelled the columns p-1, p-2, etc
> for 1959q4,
> 1959q3.
> I use the command :
>
> insheet using qp.csv
>
> to input the data into Stata.
>
> Then I do the following:
> reshape long p, i(permno) j(qtime)
> format qtime %tq
>
> Stata then creates variable v1, v2, v3, .... for all of the
> dates prior
> to 1960q1.
>
> How should I specify the dates in the csv file so that Stata
> picks them up correctly?
*
* 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/
