Problems with repeated-measures ANOVA (three within-subjects factors)

[Kristin Bott]

Hullo folks,

I'm working with a faculty/student team in Psychology who would like to run a repeated-measures ANOVA on some EEG data. Despite parroting the examples in Stata documentation/posted elsewhere, I continue to encounter bse() and bseunit() errors when I set up their analysis -- hoping you all can help. (Using Stata/SE 13.1 for Mac.)

Design:
10 participants (participant)
avg amplitude of readings taken of brain activity (avgAmp)
measured at 10 electrodes on each participant (electrodes)
in response to stimuli classified as y/n for two factors (awareness, relevance) -- so each subject will see something relevant of which they are aware, relevant of which they are not aware, irrelevant thing of which they are aware, irrelevant thing of which they are not aware...

Following this example for no between-subjects and two within-subjects factors -- http://www.stata.com/support/faqs/st...anova/#half713

-- I came up with the following for no between-subjects factors and three within-subjects factors --

anova avgAmp participant electrode / participant#electrode relevance / participant#relevance awareness / participant#awareness awareness#electrode awareness#relevance relevance#electrode awareness#relevance#electrode, repeated(electrode relevance awareness)

...which returns bseunit() error -- even if I specify bseunit(participant) at the end of the command.

The faculty advisor can run the analysis as he wants in Statistica, so in addition to getting the analysis running, I have a metric for what the numbers "should" be.

Thoughts appreciated; thanks so much.

cheers -
-kristin

Instructional Technologist / Reed College / Portland OR

Data structure / example
participant / electrode / relevance / awareness / avgAmp
1, 1, 1, 1, 10
1, 1, 1, 0, 14
1, 1, 0, 1, 7
1, 1, 0, 0, 15
1, 2, 1, 1, 8
1, 2, 1, 0, 10
... etc

[Joseph Coveney]

Try something like that shown in the attachment (plain-text output copied from the Results window). Start at the section labeled "Begin here". I tried several ways to imbed the results in the post proper, but I couldn't get ANOVA table to display properly no matter what I did (block code, monospace font, you name it), and so I had to attach it, sorry.

[Joseph Coveney]

Arrgh! A minor thing, but I put the participant random effect in the wrong location. (Also, you don't actually need the bse() anymore, too.) Take a look at the corrected-code output attached here. The ANOVA model begins at the section of output with the same title.