Hello, dear BUGS-users!
At the moment, I´m collaborating on a meta-analysis concerned with
effects of interventions aimed at pro-environmental behavior. We are
using a hierarchical model with normal likelihood and normal prior.
My problem is that I´m rather confused about the target quantity of the
model:
It seems to be common practice to model the distribution of effect sizes
and not of direct differences in pre-posttest proportions. We are
discussing the use of "cohen´s h", which has the following property. The
detectability of a difference in proportions seems to depend on the
specific location of the difference on the scale between 0 and 1.
"Cohen`s h" neutralizes this dependence and provides equal test power,
regardless of the differences position.
Phi = 2 arcsin(sqrt(P))
Cohen´s h = Phi_1 – Phi_2
(Compare: Cohen, J. (1988). Statistical power analysis for the
behavorial sciences. New York: Academic Press, pp.179)
Now my question is the following: Does it make sense to model the
distribution of "cohen´s h" instead of the pre-post differences?
I don´t see how a bayesian meta-analysis and considerations on
test-power go together and would be grateful for any help!
Best regards:
Gero
---
[log in to unmask]
http://www.uni-marburg.de/menschenfeindlichkeit/schwenk.htm
-------------------------------------------------------------------
This list is for discussion of modelling issues and the BUGS software.
For help with crashes and error messages, first mail [log in to unmask]
To mail the BUGS list, mail to [log in to unmask]
Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html
Please do not mail attachments to the list.
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list
|