Hi
On Sun, 16 Sep 2001 [EMAIL PROTECTED] wrote:
> What kind of statistical technique would you use with a study (for a class)
> consisting of 10 survey questions using a Likert Scale and two demographic
> questions (gender and age) in which one male and five females subjects
> responded? With such a low number of subjects to work with I'm not sure what
> method to suggest. This is not my area of expertise and I have someone
> asking me for suggestions? Thanks. Nina
In principle the best approach would be to: (a) calculate the
average of the 10 questions (assuming that it was designed to
measure a single construct), and (b) regress the resulting
average on the effect code for gender (-1, +1), age (ideally
calculated as a deviation score, i.e., age - Mean for age), and
the interaction or product of these two codes. With so few
subjects, however, this leads to problems. The df for the
denominator (i.e., the error term) will be n-p-1 = 5-3-1 = 1.
Just as a class demonstration, one might get away with regressing
scores on age alone. To a large extent, the answer also depends
on what point the instructor wanted to make with the example
(e.g., illustrate simple or multiple regression, t-test for
means, ...). With more subjects one could also test the
assumption that the measure was consistent (e.g., Cronbach's
alpha or split half).
Best wishes
Jim
============================================================================
James M. Clark (204) 786-9757
Department of Psychology (204) 774-4134 Fax
University of Winnipeg 4L05D
Winnipeg, Manitoba R3B 2E9 [EMAIL PROTECTED]
CANADA http://www.uwinnipeg.ca/~clark
============================================================================
