You may have to transform the percentages, so you
might as well consider a logistic analysis. If you
do, don't trust the Wald statistics -- use the
deviance to make a decision.
Sky wrote:
>
> I read your follow-up after replying to your first question. To answer this
> one: you can't do an Anova on the raw scores on accuracy, because it's a
> dichotomous variable. One of the assumptions in analysis of variance is that
> your DV is measured on interval or ratio level. So in this case you HAVE to
> use the sumscores/percentages/means on your 25 measurements of accuracy. Now
> your DV can have 26 different values and it may be seen as a continuous
> variable.
> Sky
>
> Jim Kroger heeft geschreven in bericht ...
> >May I ask a followup? I think this may get better at the heart of my
> >query. Suppose that instead of temperature, I use a measure of vision: if
> >they correctly identify a letter a distance away they get a "1," otherwise
> >a "0." I wish to use accuracy as the DV. I can calculate the percent of
> >trials in each cell obtaining a "1" and use the resulting four percentages
> >for each subject as the ANOVA data. Or, should I do an ANOVA on the raw 0s
> >and 1s themselves? My intuition tells me the two are not the same, but I
> >can not say why. (if necessary, a larger number can represent correct)
> >
> >Thank you,
> >Jim
--
Bob Wheeler --- (Reply to: [EMAIL PROTECTED])
ECHIP, Inc.
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
