Re: st: * mark indicating between group significans in box plots

[Lars Folkestad <lfolkestad@health.sdu.dk>]

Thank you for that answer Paul.
I'm in medicine. 

I must say i really value the fact that this forum is not only giving
stata code
But also tips to whats relevant - and how to make more sense of the
presented data. 
I agree with both of you that there are better ways to present paired data
than 
A simple box plot with a * marking for significant.

My study uses matched controls, and the matching has been done on known
confounding
Variables - thus the paired statistics.

A twoway scatter (over age) is also being presented in my
Paper to address the distribution and range question in more detail.

Again thank you both for your excellent comments.
lars

Den 11/11/11 11.55 skrev "Seed, Paul" <paul.seed@kcl.ac.uk>:

>As Lars has paired data, he might want to consider
>a ladder plot.  This will show the linkage between
>his observations, on which his test is based.
>
>************* example program *************
>clear
>sysuse auto
>gen id = _n
>rename turn t1
>rename trunk t2
>keep t1 t2 id
>reshape long t, i(id) j(number)
>twoway (connected t number ,  text(1.5 50 "an interesting graph") )
>************* end example *************
>
>And no, I don't think Nick was rude. I have similar discussions
>all the time.  It no doubt depends on discipline, and Lars
>does not give us a context to understand what he is trying to do,
>but in medicine and now in biomedical science there has been
>an increasing emphasis on estimating the size of effects,
>with confidence intervals and exact p-values; and
>away from mere "yes/no" thinking of the kind encouraged by stars.
>
>BW
>
>Paul T Seed, Senior Lecturer in Medical Statistics,
>Division of Women's Health, King's College London
>Women's Health Academic Centre KHP
>020 7188 3642.
>
>"I see no reason to address the comments of your anonymous expert
>... I prefer to publish the paper elsewhere" - Albert Einstein
>
>
>
>
>Den 10/11/11 13.19 skrev "Nick Cox" <n.j.cox@durham.ac.uk>:
>
>>I don't think anyone would have thought you rude here. Me, possibly....
>>
>>Here is an oblique answer.
>>
>>sysuse auto
>>graph box turn trunk, text(55 50 "an interesting graph") ysc(r(. 60))
>>
>>Nick 
>>n.j.cox@durham.ac.uk
>>
>>Lars Folkestad
>>
>>It seems that some of my text has disappeared here.
>>
>>Im using a: Wilcoxon signed ranks test - as my data is paired and not
>>normally distributed.
>>
>>I did not mean to be rude.
>>
>>I respect your decision not to answer, but seeing that this the starring
>>is a much used and
>>Often asked for by reviewers in my field I hope someone who knows the
>>answer to how to star mark
>>A boxplot will answer.
>>
>>Den 10/11/11 12.50 skrev "Nick Cox" <njcoxstata@gmail.com>:
>>
>>>I will show my hand, as it were.
>>>
>>>I've often seen people test a difference between means and show a box
>>>plot as the graphical equivalent. I'd argue that a plot of means with
>>>confidence intervals would be directly related, although even then
>>>there are pitfalls.
>>>
>>>Of  course, that may not be what you are doing. If your test were say
>>>a Mann-Whitney-Wilcoxon test, a box plot makes more sense.
>>>
>>>Even so, a -dotplot- or -stripplot- (SSC) shows much detail than a box
>>>plot for two groups. A -qqplot- is even better.
>>>
>>>I personally dislike starring as a primitive practice and draw the
>>>line at explaining how to do it but there are plenty of people to
>>>think otherwise.
>>
>>Den 10/11/11 10.31 skrev "Lars Folkestad" <lfolkestad@health.sdu.dk>:
>>
>>>>>Sorry, thought it was a graphics question and did thus not mention the
>>>>>statistics
>>
>>Den 10/11/11 10.05 skrev "Nick Cox" <njcoxstata@gmail.com>:
>>
>>>>>>Various thoughts, but I'll mention only one: What's the underlying
>>>>>>test?
>>
>>On Thu, Nov 10, 2011 at 8:49 AM, Lars Folkestad
>>
>>>>>>> Im trying to indicate between group significant in a box plot using
>>>>>>>
>>>>>>>
>>>>>>>        . graph box y1 y2
>>>>>>>
>>>>>>> Now i want to indicate that the between y1 and y2 there is a
>>>>>>>between
>>>>>>>group
>>>>>>> significants using the standard * for p<0.05 and ** for p<0.001
>>>>>>> I can do it in the editor, but if there is a code i could put in my
>>>>>>>do
>>>>>>> file, that would be the easiest.
>>>>>>>
>>>>>>> Iv been fiddling with the yvar option in something like this:
>>>>>>>
>>>>>>> . graph box y1 y2, box(1, lcolor(navy) yvar(re("*")) box(2,
>>>>>>>fcolor(green)
>>>>>>> lcolor(green))
>
>*
>*   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/


*
*   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/