Quite so.
-egen- itself is not indifferent to variable type, and I didn't say it
was. It will try to produce the default variable type -- which itself is
-float- by default -- unless you override that default.
So, if you feed -rowlast()- string variables, you need to spell out a
-str- result, as you found.
Nick
[email protected]
Andrzej Niemierko
egen rowlast_for_string = rowlast(a b)
causes "type mismatch" error message for a and b string variables while:
egen str rowlast_for_string = rowlast(a b)
works perfectly fine.
On 8/29/09 11:05 AM, "Nick Cox" <[email protected]> wrote:
> Inspection of the code shows that -egen, rowlast()- is indifferent to
> variable type in so far as
>
> 1. The syntax specifies a varlist without restriction.
>
> 2. The test for missingness uses -missing()- which applies to numeric
> and string variables alike.
>
> -egen, rowlast()- is likely to fall over for a mix of numeric and
string
> variables, an unlikely but not impossible request.
>
> Nick
> [email protected]
>
> Andrzej Niemierko
>
> Thank you very much Joseph! Indeed, it does work. Somehow I forgot to
> include type (str) in the command line.
>
> On 8/29/09 4:50 AM, "Joseph Coveney" <[email protected]> wrote:
>
>> -egen . . ., rowlast(. . .)- works with string variables for me.
> What
>> problem
>> are you having?
>>
>> Joseph Coveney
>>
>> . clear *
>>
>> . set obs 2
>> obs was 0, now 2
>>
>> . generate str a = "A"
>>
>> . generate str b = "B"
>>
>> . replace a = "" in 1
>> (1 real change made)
>>
>> . replace b = "" in 2
>> (1 real change made)
>>
>> . egen str rowlast_for_string = rowlast(a b)
>>
>> . list, noobs
>>
>> +------------------+
>> | a b rowlas~g |
>> |------------------|
>> | B B |
>> | A A |
>> +------------------+
>
>> Andrzej Niemierko wrote:
>>
>> Is there an equivalent of egen function "rowlast(varlist)" for string
>> variables?
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
