[R] anova repeated measure interpretation

[Robert Espesser]

Dear all,

I have tried to use R for the repeated measures
experiment design in a phonetic study, and in the
resulting forms I met some problems which perturb the
final interpretation of the results.

I will explain the experiment design first:
the same 7 subjects were answering a question about 25
linguistic stimuli; the stimuli were the same
utterances which were processed in 3 different
ways (3 conditions), ie each subject listened 25*3 stimuli.
I would like to test the
effect of condition and of the stimulus on the
subjects' performance.
My dependent variable is Nboundaries .
I think there is:
one random effect :subject .
2 within-subject fixed effects: stimulus(25 levels)
  and condition (3 levels) .
Following the explanations of J. Baron,
(Notes onthe use of R  for psychology...),
I have run:


 > boundaries.mod<- aov (Nboundaries ~

condition*stimulus +Error(sujet/(condition+stimulus)))

 > summary (boundaries.mod)


Error: sujet
           Df  Sum Sq Mean Sq F value Pr(>F)
condition  1  0.0044  0.0044  0.0011 0.9743
Residuals  5 19.1133  3.8227

Error: sujet:condition
           Df  Sum Sq Mean Sq F value  Pr(>F)
condition  2  6.3134  3.1567  1.7119 0.22528
stimulus   1 11.5549 11.5549  6.2663 0.02934 *  (???)
Residuals 11 20.2838  1.8440
---

Error: sujet:stimulus
                     Df  Sum Sq Mean Sq F value Pr(>F)

stimulus            24 290.522  12.105 20.3986 <2e-16 ***
condition:stimulus   1   0.065   0.065  0.1101 0.7405    (???)
Residuals          143  84.860   0.593

---

Error: Within
                     Df  Sum Sq Mean Sq F value
Pr(>F)
condition:stimulus  48  40.984   0.854  1.9854 0.0003245 ***
Residuals          287 123.429   0.430




I understand that after removing the effect due to individual subject, 
the factor condition has no effect (F: 1.7119).
and that the factor stimulus has a strong  effect (F:20.3986).

But what are the  additional lines
(I have inserted the question marks at the end of these lines),
which are not present in the examples  found in the
litterature (Baron, puzzle example) ?
They seem to appear when the both involved  factors have more than 2 levels.
Consequently, I am not sure
about the rightness and the interpretation of this model.

I will really appreciate your help.


-- 
Robert Espesser
Laboratoire Parole et Langage  UMR 6057, CNRS
29 Av. Robert Schuman  13621 AIX    (FRANCE)
Tel: +33 (0)4 42 95 36 26   Fax: +33 (0)4 42 59 50 96
http://www.lpl.univ-aix.fr/~espesser
mailto:Robert.Espesser at lpl.univ-aix.fr