Re: bonferroni correction to a mann whitney u test?

[Alan Mead]

PSPP implements a subset of SPSS syntax and it sounds like PSPP does not accept that syntax. I don't see this option in the manual: https://www.gnu.org/software/pspp/manual/pspp.pdf

As a work-around, you could run the M-W test multiple times and apply a Bonferroni-corrected alpha, right?

If you have 24 groups to compare, don't you have 24*23/2 = 276 paired comparisons to make?

If controlling the Type I error at the expense of enormously magnifying the Type II error is your goal, Bonferroni is a good way to go.  OTOH, if you care about the Type II error rate, I'd recommend that you do the comparisons without any correction and then calculate how many are significant to the number expected by chance.  You said that "all" of the comparisons are significant; if that's literally true, then comparing the 5% expected by chance to 100% observed outcomes, I'd conclude that virtually all of your groups are statistically significantly different. This doesn't sound like a situation which requires complex statistical decision-making.

-Alan

On 10/28/2016 1:03 AM, Jennifer L A Shaw wrote:

Hi,
I was wondering if anybody can help me.
My data failed the levenes test (ANOVA) so I applied the Kruskal Wallis non-parametric test. There were significant differences, so now I want to know where those differences lie by doing the Mann Whitney U test. I have 24 groups to compare with 3 reps in each group. Therefore, I would ideally like to apply a bonferroni correction to my Mann whitney test to avoid a type 1 error.

I have written my syntax like this:

NPAR TESTS
/MANN-WHITNEY = TotalCu BY byPlot (1, 5)
/POSTHOC={BONFERRONI, ALPHA ([0.0020833])}

but I get an error message like this:

.17-23: error: Syntax error at 'POSTHOC': expecting end of command.
/POSTHOC

Does anybody know what it is that I am doing wrong?

Can you apply a Bonferroni correction to Mann whitney in PSPP? If yes, how? If not, is there an alternative option?

I have tried doing mann whitney for each and every pair of groups without the bonferroni correction and everything comes out significant. When I know that it shouldn't

Jen

_______________________________________________
Pspp-users mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/pspp-users

-- 

Alan D. Mead, Ph.D.
President, Talent Algorithms Inc.

science + technology = better workers

+815.588.3846 (Office)
+267.334.4143 (Mobile)

http://www.alanmead.org

I've... seen things you people wouldn't believe...
functions on fire in a copy of Orion.
I watched C-Sharp glitter in the dark near a programmable gate.
All those moments will be lost in time, like Ruby... on... Rails... Time for Pi.

          --"The Register" user Alister, applying the famous 
            "Blade Runner" speech to software development