Social Science Statistics Tutor?

D

DefenceDoom

New Member
#1
#1
Hi Everyone!
I'm doing a doctorate in psychology and am about to defend my dissertation, which is establishing the psychometric properties of a questionnaire about body image.
I did a lot of stats for my dissertation, but honestly did not always understand exactly what I was doing or why I was doing it. I regret rushing, and I think I submitted prematurely. However, my defence is in less than a month and there is no going back now!
I would really like to hire a tutor to help me out a few times over the next few months. Someone with a good knowledge of social science statistics and psychometrics. I particularly need help understand discriminant function analysis, but other things as well, even some very basic things. Most of all, I need someone who can explain things in a patient, clear, uncomplicated manner.
Any leads?
Thanks!
 
T

the42up

New Member
#2
#2
I am certain you can find someone to help. I can say right away that my first area of concern is that you are using discriminant function analysis. I would also be curious as to why you are using this over logistic regression. Data sets in the real world rarely meet all the assumptions to make DFA tick (though when they do it is the superior technique to LogReg).

I also would ask why you are not doing an exploratory factor analysis and confirmatory factor analysis followed by moderator invariance testing if you wanted to know the psychometric properties of an instrument you are developing.

I would say that since you are going to use DFA (since you have already submitted) the most important aspect seems to be justifying it to your committee over logistic regression. If you have someone whose expertise is measurement, stats, or methods they might ask you that question.
 
D

DefenceDoom

New Member
#3
#3
Thanks so much for your reply. Do I ever regret submitting before getting this all sorted!

I did review the assumptions of DFA and seemed to meet them. I do plan to discuss this with my stats committee member this afternoon, though.

The reason I didn't do factor analysis is because my measure was too small (11 items). Based on my internal consistency analysis, there was some indication that there may be a couple factors at play, but I only had one question assessing one of the factors. So if I had my time back, I would have started with a much longer questionnaire and whittled it down with factor analysis.

If that makes sense....please feel free to tell me if it doesn't!
 
You must log in or register to reply here.