What would be thwe correct statistical procedure for this design?
From: Guido F. Gebauer (guido_at_gfgebauer.de)
Date: 11/16/04
- Next message: Will: "Independent random variables versus non correlated variables"
- Previous message: Aleks Jakulin: "Re: Journal ref for applied Bayesian Analysis"
- Messages sorted by: [ date ] [ thread ]
Date: 16 Nov 2004 01:28:25 -0800
Hi,
I would be most grateful if anybody could recommend a statistical
procedure for the following research design. Sorry for bothering
mathematicians with the low knowledge of a psycholiogiust:)
I am interested in examining moderating effects of personality
variables on the ability of certain (assumed to be continuous or rank
ordered) criteria of credibility of witness statements to distinguish
between true and false statements.
In order to do that I will ask a number of participants to tell two
stories, one of the being true and the other being false. So each
participants will tell two stories. In addition participants will
solve several tests of cognitive ability and fill in a perosnality
questionaire based ont he Big Five. By text analysis, for each of the
2 X number of participants stories 15 criteria of credibility will be
assessed and rated on a scale from 0-5. Interrater-Agreement will be
calculated.
The aim of the study is then (1) to examine the capability of the
different criteria of credibility to distinguish between true and
false stories (that has been done several times already), (2) to
examine whether individual measures of personality and cognitive
ability influence the capacity of the criteria of credibility to
distinguish between true and false statements (e.g. it may be that in
participants with high verbal creativity the criterion of "unusual
details" does not differentiate between true and false statements, or
that in subjects with high "openess for feelings" the criterion
"report of own emotional experience) does not differentiate anymore
between true and false statements).
I do not want simply to divide the participants in high and low groups
for the respective personality and cognitive variable, but I want to
employ the personality and cognitive varibales as continuous
variables, the criteria of crediiblity may be treated as continuous or
rank ordered varibales.
The design is a within-subject-design since each subject is telling
two stories. The statistical procedure then should be able to tell me
how well the individuel criteria of credibility distinguish between
true and false stories and whether this is influenced in any way by
the continuous personality and ability variables.
I want first to make an overall statistical assessment (all variables
of credibility X all varibales of personality/ability on capacity to
distinguish between true and falsse stories. Secondly I want to
determine for each individual variable of credibility how its capacity
to distinguish between true and false stories is influenced by
variables of personality and ability.
In order to determine sample size in advance I also would need to
calculate how much subjects I need in order to have a good chance to
identifiy a moderate effect as significant.
Can anybody give me some advice on the statistical procedure I habve
to employ and any software capable of doing that? Thank´s a lot!
Guido
- Next message: Will: "Independent random variables versus non correlated variables"
- Previous message: Aleks Jakulin: "Re: Journal ref for applied Bayesian Analysis"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
