Re: st: horizontal box plots and log scale

[Nick Cox <njcoxstata@gmail.com>]

I am no longer very clear on your real problem or how it led to your  
first post.

But if the values given are like your data, which are relative risks  
for various situations or studies, then I would not reach for a box  
plot at all. I would want to see all the values and so would use - 
graph dot- or -stripplot- (SSC) with an -over()- option. Such commands  
allow text labels for each value.

When some values are very close together -devnplot- (SSC) can work  
better than -stripplot-. There is an example in my San Diego  
presentation, downloadable from the Stata website.

Nick

On 12 Sep 2012, at 22:58, Susanna Cramb  
<SusannaCramb@cancerqld.org.au> wrote:

> Thanks so much Nick for your very prompt reply.
>
> I had seen the FAQ... My sole reason for the transformation is to  
> obtain equally spaced ticks for the values 0.4, 0.5, 0.66, 0.8, 1.0,  
> 1.25, 1.5, 2.0, 2.5 (as it's using relative risk) so I didn't think  
> it was necessary/desirable to transform the data. (Incidentally, I  
> also tried -stripplot-, but had the same problem).
>
> However, if that is the only option, then obviously that will have  
> to do.
>
> Thanks again,
> Susanna
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner- 
> statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
> Sent: Wednesday, 12 September 2012 7:04 PM
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: horizontal box plots and log scale
>
> This is a FAQ. See
>
> FAQ     . . . . . . . . . . . . . . . . . . . Box plots and  
> logarithmic scales
>        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .   
> N. J. Cox
>        9/05    How can I best get box plots on logarithmic scales?
>                http://www.stata.com/support/faqs/graphics/boxandlog.html
>
> The essence of the problem is that -xsc(log)- just takes the graph  
> you would have otherwise and transforms it logarithmically. You must  
> transform the data first, draw the box plot, and then fix the axis  
> labels.
>
> -mylabels- (SSC) may help with the last.
>
> I think you are wrong about Stata version here: this is in principle  
> a problem with any version of Stata and a wired in boxplot command.
>
> Nick
>
> On Wed, Sep 12, 2012 at 9:53 AM, Susanna Cramb <SusannaCramb@cancerqld.org.au 
> > wrote:
>
> > I am having difficulties in generating horizontal box plots on the  
> > log scale.
> >
> > I am using Stata/SE 12.1 for Windows (32-bit), although I also have  
> > the problem with the same version under (64-bit) and also Stata/SE  
> > 11.2 for Windows (32-bit). I did not have this problem when using  
> > earlier versions of Stata.
> >
> > If I run the following example from the graph box help file, Stata  
> > generates the graph as expected:
> >
> > sysuse nlsw88, clear
> > graph hbox wage, over(ind, sort(1)) nooutside ytitle("") title 
> > ("Hourly
> > wage, 1988, woman aged 34-46", span) subtitle(" ") note( "Source:
> > 1988 data from NLS, U.S. Dept. of Labor, Bureau of Labor  
> > Statistics" ,
> > span)
> >
> > However, if I request it on the log scale, as in the following  
> > code, I obtain a horizontal box plot where all the categories are  
> > squished on top of each other at the very top of the graph, while  
> > the horizontal axis remains unchanged.
> >
> > sysuse nlsw88, clear
> > graph hbox wage, over(ind, sort(1)) nooutside ytitle("") title 
> > ("Hourly
> > wage, 1988, woman aged 34-46", span) subtitle(" ") note( "Source:
> > 1988 data from NLS, U.S. Dept. of Labor, Bureau of Labor  
> > Statistics" ,
> > span) yscale(log)
> >
> > Has anyone else had this problem, and is there any solution?
> *
*
*   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/