|Date: ||Mon, 4 Nov 1996 10:19:59 -0500|
|Reply-To: ||"William B. Ware" <wbware@EMAIL.UNC.EDU>|
|Sender: ||"SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>|
|From: ||"William B. Ware" <wbware@EMAIL.UNC.EDU>|
|Subject: ||Re: Simple mixed design|
On Mon, 4 Nov 1996, Jennifer Marshall wrote:
> I am doing a simple mixed design with one between subjects factor and one
> within subjects factor.I have gotten an error term for the between
> subjects factor and for the interaction of the within and between subject
> factors. I do not understand where the error term comes from.
> Could someone please explain the error term AND where it comes from?
Here is the way that I conceptualize the design, as I understand you.
Let's call the between-factor A and the within-factor B. You have
observations nested within the A factor, but crossed with the B factor.
You should have two error terms: a between and a within. The "between"
error term is simply the pooled error term within each level of A. That
is, there are cases within A1, cases within A2, etc. It works just like a
oneway design. The "within" error term serves for both the B effect and
the A*B effect.
Think of the "within" term this way. Within each level of A, you have a
randomized block design; cases by B. Within each of those designs, you
have a B effect, a subjects effect, and a residual. The overall "within
subjects" error term is the pooled residuals...
In testing B and A*B, you need to attend to some assumptions. First, that
you have homogeneity of residuals, and that the pooled variance-covariance
matrix is "symmetric." You might want to look at Roger Kirk's book on
William B. Ware, Professor and Chair Educational Psychology
CB# 3500 EMAIL: firstname.lastname@example.org
University of North Carolina PHONE: (919)-966-5266
Chapel Hill, NC 27599-3500 FAX: (919)-962-1533