problem with reshape: no xij variables found

[Nader Mehri]

Hi,
I am trying to reshape the below dataset but I got an error: no xij variables found. Thanks for your help!
NM

reshape long mlog_*, i(age ) j(profile)

* Example generated by -dataex-. For more info, type help dataex
clear
input float(age mlog_HD mlog_HH mlog_HU mlog_UD mlog_UH mlog_UU)
 0 .004579183 .9107729 .08464794 .033043746  .2934907 .6734655
 1 .005324981 .9076573 .08701778 .036915746 .29491556 .6681687
 2 .006227782 .9040858 .08968644   .0413826  .2955122 .6631052
 3 .007324957  .900004 .09267098   .0465443 .29524764 .6582081
 4 .008663595 .8953469 .09598954  .05251758  .2940882 .6533942
 5 .010303146  .890036 .09966086  .05943817 .29199946 .6485624
 6   .0123188 .8839776 .10370362  .06746305 .28894618 .6435907
 7 .014805734 .8770588 .10813546  .07677244 .28489274 .6383348
 8 .017884482  .869144 .11297154   .0875713  .2798038 .6326249
 9 .021707566 .8600699  .1182225   .1000899  .2736456 .6262645
10 .026467614 .8496408 .12389161  .11458267  .2663878 .6190295
11 .032407008 .8376221 .12997088  .13132481  .2580061 .6106691
12  .03982897 .8237353 .13643572   .1506054 .24848618 .6009084
13  .04910954 .8076524 .14323808   .1727161  .2378284 .5894555
14  .06070913 .7889932 .15029763   .1979343  .2260538 .5760119
15  .07518136 .7673277 .15749097  .22649983  .2132108 .5602894
16  .09317504 .7421854 .16463958  .25858563  .1993825 .5420319
17  .11542299 .7130795  .1714976   .2942632 .18469305 .5210438
18  .14270909 .6795489 .17774205   .3334662  .1693125 .4972213
19   .1758036 .6412259  .1829705   .3759573 .15345787 .4705848
20   .2153585 .5979297  .1867118   .4213054 .13738889 .4413058
end

[Ali Atia]

Reshape long takes a stub as an argument rather than a list of existing variables, which is what mlog_* expands out to.

reshape long mlog_,i(age) j(profile,string)

[Nick Cox]

I think Ali Atia means

 
 reshape long mlog_,i(age) j(profile) string

[Ali Atia]

As far as I can tell, both variations do the same thing.

[Nick Cox]

Indeed; string appears to be an undocumented suboption of j(). I did not know that! Thanks!

(Or else where is it documented? How did you find out?)

[Ali Atia]

I don't recall actually consulting documentation when I first used the string suboption, so it's likely that I used it assuming it was allowed and was happily correct.

I think you could call it at least partially documented. The Advanced Syntax section of the reshape helpfile specifies that -reshape j- has a string option, and that "the basic syntax of reshape is implemented in terms of the advanced syntax, so you can mix basic and advanced syntaxes." So adding a string suboption to j() could reasonably be considered mixing basic and advanced syntaxes.