st: RE: Re: reshape ,string has its limitations

[baum <baum@bc.edu>]

Alejandro,

Alejandro,

It is indeed the case that the second reshape will fail in that instance to 
do anything sensible, since it finds multiple words rather than unique 
string values in the 'j' variable.

I think the answer to this lies in Nick Cox' latest 'Speaking Stata' 
column. You need to use 'encode' to change 'var' into a numeric variable 
with its string values as value labels. This will then do the second 
reshape properly. I do not have time now to experiment with it, but when 
you have done that second reshape, you will have variables that have 
numeric tags. At that point you can apply sensible variable names. The real 
issue here is that a name like 'Total factor productivity' is not a legal 
Stata variable name; 'TFP' or 'TotalFactorProductivity' would be.

Thanks
Kit

--On Tuesday, September 17, 2002 11:52 -0400 "Riano, Alejandro" 
<ALEJANDROR@iadb.org> wrote:


> Hi, Kit your procedure for my reshape problem works almost perfect, the
> thing is that my descriptor usually is a very long string such as "Total
> factor Productivity" and when i do
>
> . reshape wide d,i(cty year) j(var) string
>
> i'll lose my variables column as i get:
>
>
> (note: j = Total factor productivity Volume of non-oil imports)
> (note: no data for descriptor == Total)
> (note: no data for descriptor == factor)
> (note: no data for descriptor == productivity)
>
> I ended up with only country year columns.
>
> The second problem is that the number of series is very large (about 170)
> and most of them are not available for all countries, which make coding
> very messy. I'd really appreciate your help on this,
>
> thank you
>
> ALejandro
> -----Original Message-----
> From: baum [mailto:baum@bc.edu]
> Sent: Tuesday, September 17, 2002 8:28 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: Re: reshape
>
>
>
> --On Tuesday, September 17, 2002 2:33 -0400 Alejandro wrote:
>
> > Hi, i have the a database like this:
> >
> > country         descriptor      year1970 year1971 ... year2000
> > Argentina          var1                x70          x71              x00
> > Argentina          var2                y70          y71              y00
> > Bolivia               var1                x70         x71
> > x00 Bolivia               var2                y70         y71
> > y00 .
> > .
> > .
> >
> > and so on... i want to know if exists a way in which i could reshape this
> > wide dataset into a long one that would look like this:
> >
> > country         year           var1           var2
> > Argentina       1970            x70            y70
> > Argentina       1971            x71            y71
> >     .
> >     .
> >     .
> >  do you know how could i do the reshape of the two variables jointly ? if
> > i use the <reshape> command i'd have to do it one by one.
>
> . list
>
>            cty        var     d1970     d1971     d1972
>   1.       arg       var1      1970      1971      1972
>   2.       arg       var2      2070      2071      2072
>   3.       bol       var1      2170      2171      2172
>   4.       bol       var2      2270      2271      2272
>
> . reshape long d ,i(cty var)
>
> . rename _j year
>
> . reshape wide d,i(cty year) j(var) string
>
>
> . list
>
>            cty       year     dvar1     dvar2
>   1.       arg       1970      1970      2070
>   2.       arg       1971      1971      2071
>   3.       arg       1972      1972      2072
>   4.       bol       1970      2170      2270
>   5.       bol       1971      2171      2271
>   6.       bol       1972      2172      2272
>
>
> There is always a way to do it with reshape. The difficulty is figuring
> out  how!
>
> Kit
>
> *
> *   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/

*
*   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/