graphing results of post-hoc pairwise comparisons

[Des Grieux]

Hello fellow SPSS users!

I have searched for this information everywhere but could not find anything of much help and I'd appreciate any advice.

How do you show the results of post-hoc pairwise comparisons (i.e. significant group differences) on bar charts with means, like in the example chart attached below? Do I need an R plugin for this or is this function accessible directly in SPSS? If not, what is the procedure for using the R plugin?

I'd be grateful for all your help. Thanks in advance!

[Rich Ulrich]

The way that I know of showing a concise summary of results of post-hoc
tests uses a list of the ordered means.  For your chart, I see (approximately)

1 2 8 5 4 6 3 7      as the increasing order of the means, when numbered 1-8.

Then you underline (1 2 8 5) to show that they "do not differ";  and you
underline (5 4 6)  to show that they "do not differ".  Other contrasts "differ".

SPSS Oneway used to provide a version of this for independent means (not
for repeated measures).  (Does Oneway still exist?  features available?)

The main body of theory for post-hocs is devoted to the example of independent
means.  Adapting that thought and theory to correlated means is still an iffy
business, whether you find a reference or do-it-yourself.  KISS ought to apply.


Your chart is very hard to read, and it would be harder to read if it were complete.

If your post-hocs are computed as separate t-tests because of unequal errors,
then you need the table-matrix of contrasts, 8x8, which I think SPSS provided
for Oneway ANOVA.  Separate, paired-t's are often recommended for repeated
errors owing to the difference in errors that frequently results from varying
correlations.  That might also want a matrix, if you consider all the contrasts to
be interesting,

I don't think that the option you ask for is something needed.
--
Rich Ulrich

> Date: Sun, 10 Feb 2013 05:27:40 -0800

> From: [hidden email]
> Subject: graphing results of post-hoc pairwise comparisons
> To: [hidden email]
>
> Hello fellow SPSS users!
>
> I have searched for this information everywhere but could not find anything
> of much help and I'd appreciate any advice.
>
> How do you show the results of post-hoc pairwise comparisons (i.e.
> significant group differences) on bar charts with means, like in the example
> chart attached below? Do I need an R plugin for this or is this function
> accessible directly in SPSS? If not, what is the procedure for using the R
> plugin?
>
> I'd be grateful for all your help. Thanks in advance!
> <http://spssx-discussion.1045642.n5.nabble.com/file/n5718016/post-hoc.gif>
>
>

...

[John F Hall]

Help > Command Syntax Reference

 

ONEWAY is on page 1441: PLOT MEANS is on page 1467.

 

 

John F Hall (Mr)

[retired academic survey researcher]

 

Email:     [hidden email]

Website: www.surveyresearch.weebly.com

 

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Rich Ulrich
Sent: 10 February 2013 22:35
To: [hidden email]
Subject: Re: graphing results of post-hoc pairwise comparisons

 

The way that I know of showing a concise summary of results of post-hoc
tests uses a list of the ordered means.  For your chart, I see (approximately)

1 2 8 5 4 6 3 7      as the increasing order of the means, when numbered 1-8.

Then you underline (1 2 8 5) to show that they "do not differ";  and you
underline (5 4 6)  to show that they "do not differ".  Other contrasts "differ".

SPSS Oneway used to provide a version of this for independent means (not
for repeated measures).  (Does Oneway still exist?  features available?)

The main body of theory for post-hocs is devoted to the example of independent
means.  Adapting that thought and theory to correlated means is still an iffy
business, whether you find a reference or do-it-yourself.  KISS ought to apply.


Your chart is very hard to read, and it would be harder to read if it were complete.

If your post-hocs are computed as separate t-tests because of unequal errors,
then you need the table-matrix of contrasts, 8x8, which I think SPSS provided
for Oneway ANOVA.  Separate, paired-t's are often recommended for repeated
errors owing to the difference in errors that frequently results from varying
correlations.  That might also want a matrix, if you consider all the contrasts to
be interesting,

I don't think that the option you ask for is something needed.
--
Rich Ulrich

> Date: Sun, 10 Feb 2013 05:27:40 -0800

> From: [hidden email]
> Subject: graphing results of post-hoc pairwise comparisons
> To: [hidden email]
>
> Hello fellow SPSS users!
>
> I have searched for this information everywhere but could not find anything
> of much help and I'd appreciate any advice.
>
> How do you show the results of post-hoc pairwise comparisons (i.e.
> significant group differences) on bar charts with means, like in the example
> chart attached below? Do I need an R plugin for this or is this function
> accessible directly in SPSS? If not, what is the procedure for using the R
> plugin?
>
> I'd be grateful for all your help. Thanks in advance!
> <http://spssx-discussion.1045642.n5.nabble.com/file/n5718016/post-hoc.gif>
>
>

...