2x2x2x2 design

[m***@gmail.com]

Hi,

I am looking for any ideas on how to analyze the data coming from a
2x2x2x2 factorial design experiment. It is a within-subjects experiment
where 4 factors were varied with 2 levels each (2x2x2x2) and
measurements of 4 different responses were measured. I am interested in
finding out if there are interactions among the factors while
determining the measures individualy.

I am an engineer with little knowledge in statistics. I would
appreciate any help or any reference to good references (papers or
books) on the subjects. I think I might be looking at the wrong places
because the books barely touch the multivariate case with more than one
level.

Minerva

[Ray Koopman]

Post by m***@gmail.com
Hi,
I am looking for any ideas on how to analyze the data coming from a
2x2x2x2 factorial design experiment. It is a within-subjects experiment
where 4 factors were varied with 2 levels each (2x2x2x2) and
measurements of 4 different responses were measured. I am interested in
finding out if there are interactions among the factors while
determining the measures individualy.
I am an engineer with little knowledge in statistics. I would
appreciate any help or any reference to good references (papers or
books) on the subjects. I think I might be looking at the wrong places
because the books barely touch the multivariate case with more than one
level.
Minerva

Because the design is 2^k, each anova effect has only one degree of
freedom and can be represented by a single linear contrast of the 16
conditions. So for each of the 15 sources of variance (4 main effects,
6 two-way interactins, 4 three-way interactions, 1 three-way
interaction), you can generate an n x 4 (cases x measures) matrix of
contrast scores. For each such matrix, the null hypothesis is that all
the column means are zero, which is the simplest of all multivariate
hypotheses.

[f***@gmail.com]

Post by Ray Koopman

Post by m***@gmail.com
Hi,
I am looking for any ideas on how to analyze the data coming from a
2x2x2x2 factorial design experiment. It is a within-subjects experiment
where 4 factors were varied with 2 levels each (2x2x2x2) and
measurements of 4 different responses were measured. I am interested in
finding out if there are interactions among the factors while
determining the measures individualy.
I am an engineer with little knowledge in statistics. I would
appreciate any help or any reference to good references (papers or
books) on the subjects. I think I might be looking at the wrong places
because the books barely touch the multivariate case with more than one
level.
Minerva

Because the design is 2^k, each anova effect has only one degree of
freedom and can be represented by a single linear contrast of the 16
conditions. So for each of the 15 sources of variance (4 main effects,
6 two-way interactins, 4 three-way interactions, 1 three-way
interaction), you can generate an n x 4 (cases x measures) matrix of
contrast scores. For each such matrix, the null hypothesis is that all
the column means are zero, which is the simplest of all multivariate
hypotheses.

how about posting the syntax for that?
thank you

[David Marso]

You do realize you are replying to a message from 2005?
--

Post by f***@gmail.com
how about posting the syntax for that?
thank you

[c***@gmail.com]

Funny disappearence!
Anyway....thanks you, David, for your clear suggestions.
They were so useful to me.