Hello all,
I have a quick question that I was hoping that one with superior
statistical skills to me might be able to help with.
I've recently been doing a study that attempts to compare mathematical
cognition with general cognition. This involves doing various
experiments with a mathematical population and a non-mathematical
population.
I've given two versions of a task to people (one version to each
person, within each population) that returns nominal data. With this
particular task it is common to find that those doing version 2 perform
significantly better than those doing version 1. Clearly, within
populations this is calculated with a chi-squared test and Cramer's V.
I want to say that, in some sense, the effect size gives a measure of
how much bias there is within the population.
When I do this I find that both mathematicians and the general
population perform significantly better (p<0.001) on version 2, but
that the effect sizes are very much different. For the mathematicians
it is 0.61, but for the general pop it is 0.79.
Is there a way of testing whether these effect sizes are significantly
different? Is this even a meaningful question?
Any help with this would be very much appreciated,
Matthew
-----------------------------------------------
Matthew Inglis
WS202a, Institute of Education
University of Warwick
02476 523899 (internal x23899)
-----------------------------------------------
|