help with anova

[Christine]

Hello,

I want to look at differences between groups (among African Americans and
Caucasians) for 5 continuous variables.  Can I use a one way ANOVA with the
Bonferonni post hoc test for this?

Thanks

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Swank, Paul R]

As long as the continuous variables behave nicely but I would not use Bonferroni here as it is too conservative to use for all possible tests. You might want to research the Benjamini-Hochberg test.

Dr. Paul R. Swank,
Professor
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Christine
Sent: Friday, July 15, 2011 10:24 AM
To: [hidden email]
Subject: help with anova

Hello,

I want to look at differences between groups (among African Americans and
Caucasians) for 5 continuous variables.  Can I use a one way ANOVA with the
Bonferonni post hoc test for this?

Thanks

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Bruce Weaver]

I have a couple questions.

1. What are the sample sizes?

2. Do you have 5 univariate questions, or one multivariate question?  I.e., do you want to test for differences on each of the 5 variables separately, or do you want one comparison of the groups on a weighted combination of the 5 variables via MANOVA?  

Christine wrote

Hello,

I want to look at differences between groups (among African Americans and
Caucasians) for 5 continuous variables.  Can I use a one way ANOVA with the
Bonferonni post hoc test for this?

Thanks

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Rich Ulrich]

Something is mis-stated here.  That would be "a repeated measures ANOVA"
(if they are all scaled with the same range and in the same direction,
reflecting one latent variable), or "5 one-way ANOVAs" (i.e., t-tests).

For the Bonferonni post-hoc, that's just a matter of using 1% instead of
5%  for the nominal cutoff, when starting with 5 t-tests at the 5% level.

However, as Bruce suggests, if you are interested in a *pattern* of
differences, you might want a MANOVA.  For two groups, that is a simple
discriminant function.

For five variables, I would usually have one overall hypothesis in mind,
so I would either make my overall test by repeated measures if that
describes the hypothesis; or on a single variable so that the test has 1 d.f.  

That "single variable" is sometimes one variable from the set, like using
"Relapse"  instead of several rating scales; or I create a composite score
from the five.  - In between those ideas.... factor analysis might tell me
whether I should create two composites, or proceed with one.

--
Rich Ulrich

> Date: Fri, 15 Jul 2011 11:24:08 -0400

> From: [hidden email]
> Subject: help with anova
> To: [hidden email]
>
> Hello,
>
> I want to look at differences between groups (among African Americans and
> Caucasians) for 5 continuous variables. Can I use a one way ANOVA with the
> Bonferonni post hoc test for this?
>