data analysis question

[Kristine Oppelstrup]

Hi,

I'm wondering if anyone would be willing to give me
some advice on how to analyze data that I've
collected. Below is a description - rather lengthy, I
know and apologize :)

If this description is not clear enough, I will be
more than happy to try to clarify.

The main focus of the study was to investigate whether
there are differences in performance evaluations based
on the sexual orientation of the target, and if there
is an interaction of this possible effect with gender,
i.e. if the effect is different for gay men and
lesbians.

It was a laboratory experiment, where undergraduate
students rated fictitious target persons on their job
performance. Each student rated 6 target persons: one
gay man, one lesbian, two straight men, and two
straight women. The specific characteristics of each
of there 6 people were described in his or her
“personnel file”. On-the-job-actions of each person
were described in fictitious scenarios, and the
students/raters were asked to provide job performance
ratings on a number of dimensions. Altogether, there
were 6 different scenarios, i.e. descriptions of job
actions; two representing a low job performance, two
representing a medium job performance and two
representing a high job performance. Thus, there were
all in all 6 fictitious “personnel files” and all in
all 6 different on-the-job-actions scenarios. Each
“personnel file” was combined with an on-the-job-
actions scenario, but they were combined (more of
less) randomly. Thus, since each rater rated the job
performance of 6 people based on the descriptions of
these peoples on-the-job-actions, each rater saw all
the existing “personnel files” and all the existing
on-the-job-actions scenarios, but combined
differently.

The main research question is whether there is any
difference in job performance ratings based on sexual
orientation and gender. The problem is that the design
is a 2*2*3 design with 2 levels of sexual orientation
(heterosexual and homosexual), 2 genders (male and
female) and 3 described levels of on-the-job-actions
(low, medium, high). Therefore, there are 12 cells
(2*2*3). However, each participant only rated (at the
most) 6 of these possible combinations, so for each
rater, there are at least 6 empty cells. Moreover,
these 6 empty cells are spread out in a haphazard,
non-systematic way. (There is no cell that is empty
for all raters.) Another possible combination is that
many of the raters rated only 5 of the possible 12
cells, i.e. one of the combinations occurred twice for
that rater. For example a rater may have rated two
heterosexual females that performed at a described low
level of on-the-job-actions. This occurred because the
combinations (“personnel files” with on-the-job-
actions scenarios) were combined randomly. 40% of the
316 raters rated only 5 of the possible 12
combinations.

I would really appreciate help with this.

Thanks for your time!
Kristine


      ____________________________________________________________________________________
Park yourself in front of a world of choices in alternative vehicles. Visit the Yahoo! Auto Green Center.
http://autos.yahoo.com/green_center/