Design analysis help

[Rose, Fred]

Students in a research methods class designed a taste-test study comparing 3 brands of coffee.  The basic design is:

                   Cup
             A      B      C
Group 1:
Group 2:
Group 3:

Thus, participants in each group tasted 3 cups of coffee and rated each on a 1-5 point scale; thus “cup” is a repeated measure.  Groups differed in the labeling and contents of the cups (basically, students were interested in whether knowledge of the brand/type of coffee alters preferences), and so “Group” is a between groups measure.  What would be the “best” way to analyze these data?  I’m drawing a blank on how to compare the relative preferences across groups.  Is there an omnibus test to do this?  I’m thinking of MANOVA, but wanted some input to be sure since multivariate analysis isn't my strong suit (actually, I’m not sure that I have a strong suit, but that’s another story).

Thanks in advance,

Fred

--
Fredric E. Rose, Ph.D.
Assistant Professor of Psychology
Palomar College
760-744-1150 x2344
[hidden email]

[Maguin, Eugene]

Fred,

I would conceive of this as a repeated measures anova with cup as the within
factor (repeated measure) and brand/type as the between factor. Repeated
measures anova has stringent assumptions regarding the within-group
variance-covariance matrix (vcm) that can be checked. This would be a good
exercise for your class to work through the assumptions. You also could
analyze this as a multivariate anova (i.e., manova) but while manova allows
broader assumptions about the vcm, the price of those broader assumptions is
reduced power. Again, the class could work through this analysis as well.

Gene Maguin

[Swank, Paul R]

This is also easily handled with a mixed models approach, wehre there
are typically more options for the variance covariance matrix than in a
repeated measures ANOVA.


Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Gene Maguin
Sent: Tuesday, April 10, 2007 2:48 PM
To: [hidden email]
Subject: Re: Design analysis help

Fred,

I would conceive of this as a repeated measures anova with cup as the
within factor (repeated measure) and brand/type as the between factor.
Repeated measures anova has stringent assumptions regarding the
within-group variance-covariance matrix (vcm) that can be checked. This
would be a good exercise for your class to work through the assumptions.
You also could analyze this as a multivariate anova (i.e., manova) but
while manova allows broader assumptions about the vcm, the price of
those broader assumptions is reduced power. Again, the class could work
through this analysis as well.

Gene Maguin

[Rose, Fred]

Thanks to those who have responded so far.  I think just asking the question helped give me clarity.  Isn't this "merely" a 3x3 mixed design?  Thus, as Paul noted, a mixed model would do what I wanted.  Unless someone corrects me, I'll just slink back, somewhat embarrassed, that I didn't think of this sooner.

Fred

Fredric E. Rose, Ph.D.
Assistant Professor
Behavioral Sciences
Palomar College
1140 W Mission Rd.
San Marcos, CA 92069
760-744-1150 x2344



-----Original Message-----
From: SPSSX(r) Discussion on behalf of Swank, Paul R
Sent: Tue 4/10/2007 2:08 PM
To: [hidden email]
Subject:      Re: Design analysis help

This is also easily handled with a mixed models approach, wehre there
are typically more options for the variance covariance matrix than in a
repeated measures ANOVA.


Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of
Gene Maguin
Sent: Tuesday, April 10, 2007 2:48 PM
To: [hidden email]
Subject: Re: Design analysis help

Fred,

I would conceive of this as a repeated measures anova with cup as the
within factor (repeated measure) and brand/type as the between factor.
Repeated measures anova has stringent assumptions regarding the
within-group variance-covariance matrix (vcm) that can be checked. This
would be a good exercise for your class to work through the assumptions.
You also could analyze this as a multivariate anova (i.e., manova) but
while manova allows broader assumptions about the vcm, the price of
those broader assumptions is reduced power. Again, the class could work
through this analysis as well.

Gene Maguin