Re: st: Suppress unused values of categorical var in graph? (Stata 11)

[Nick Cox <njcoxstata@gmail.com>]

I can't find a -rew- on SSC.

Nick

On Sun, Aug 14, 2011 at 8:37 PM, Kaulisch, Marc
<kaulisch@forschungsinfo.de> wrote:
> In order to keep the same order of the categorical variables on the graph as I would get with -ciplot- (from ssc) I add another line to the code, using the -revv- command (from ssc):
>
> egen group = group(for_eu), label
> revv group, gen(groupa)
> twoway rcap ub lb groupa, hori || scatter groupa mean, ///
> legend(off) ytitle("") ylabel(1/5, valuelabel angle(0) nogrid)
>
> Marc
>
> -----Ursprüngliche Nachricht-----
> Von: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Kaulisch, Marc
> Gesendet: Sonntag, 14. August 2011 21:23
> An: statalist@hsphsun2.harvard.edu
> Betreff: AW: st: Suppress unused values of categorical var in graph? (Stata 11)
>
> Nick,
> Stupid me, haven't thought about that solution because I used "egen group" only with two or more vars... Your solution is very handsome.
> Of course, my problem is self-inflicted but in some cases it is useful that values of categorical variables are not equally spaced.
>
> Marc
>
> -----Ursprüngliche Nachricht-----
> Von: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von Nick Cox
> Gesendet: Sonntag, 14. August 2011 17:53
> An: statalist@hsphsun2.harvard.edu
> Betreff: Re: st: Suppress unused values of categorical var in graph? (Stata 11)
>
> Marc's problem could be regarded as self-inflicted. He creates a variable with values 1, 4, 6, 8, 12 and then complains that the values when supplied as one axis of a -twoway graph- are not plotted with equal spacing.
>
> More seriously: Marc's problem is certainly that -twoway- has no notion of a categorical variable. However, two more lines of code produce equal spacing:
>
> egen group = group(for_eu), label
> twoway rcap ub lb group, hori || scatter group mean, ///
> legend(off) ytitle("") ylabel(1/5, valuelabel angle(0) nogrid)
>
> I'll look into the problem with -ciplot- reported by Marc.
>
> Nick
>
> On Sun, Aug 14, 2011 at 1:02 PM, Kaulisch, Marc <kaulisch@forschungsinfo.de> wrote:
>> I came across a problem with -ciplot- (from ssc) because option
>> ytitle("") does not suppress the ytitle. As Nick Cox (author of
>> ciplot) mentioned earlier that he does not touch ciplot anymore and
>> one should use the statsby strategy (see Stata Journal 10, 1,
>> p.143-151). I started to follow his advise but come to a point of
>> dissatification and maybe some one can help me (still Stata 11 user).
>>
>> I have a categorical variable with categories ranging from 8 to 44
>> where some values in this range are not attached to any categories.
>> Drawing a graph means that on the axis of this variable there is a lot of "unused"
>> space. Is there a way to reduce this?  Is there a way to tell Stata
>> that this is a categorical variable (factor variables are not allowed...)?
>> (Of course a recode of the cat var would do the trick but why should
>> this be the only option?)
>>
>> My example code with the auto dataset
>>
>> sysuse auto
>> recode foreign (0=1) (1=4), gen(for_eu) replace for_eu = 6 if
>> regexm(make, "Renault") replace for_eu = 8 if regexm(make, "VW")
>> replace for_eu = 12 if regexm(make, "Audi") statsby mean=r(mean)
>> ub=r(ub) lb=r(lb) N=r(N), clear by(for_eu): ci weight levelsof for_eu,
>> local(fe) twoway rcap ub lb for_eu, hori || scatter for_eu mean, ///
>> legend(off) ytitle("") ylabel(`fe', valuelabel angle(0) nogrid)
>>

*
*   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/