[R] Greenhouse-Geisser epsilon for interaction term

[Samuele Carcagno]

Hi,

I'm running a repeated measure ANOVA to test the effects of the
within-subjects
factors "congr", "isi" and their interaction, on the variable "latencies"

aov(latencies ~ congr*isi + Error(subj/(congr*isi)),data=dats)

"congr" has 4 levels, while "isi" has "3" levels.

I would like to check for possible violations of the sphericity assumption,
and
following the example given by Baron and Li(2004), I was able to compute
Greenhouse-Geisser epsilon for the effect of "congr" and for the effect of
"isi". I calculated the variance/covariance matrix for each of these terms
by
averaging the "latencies" across the levels of the other factor,  but I
can't
figure out how to calculate the variance/covariance matrix for the
interaction
term. I've looked up several books and tutorials over the internet to find a
solution, but either they skip entirely over the actual calculations and
present
instead the output given by SPSS, or they are at a too advanced level for
me.

Any help would be really appreciated,


Samuele


References:

- Baron,J. & Li,Y.(2004).Notes on the use of R for psychology experiments
and
questionnaires. Available:
http://www.psych.upenn.edu/~baron/rpsych/rpsych.html