Data Analysis

Means and standard deviations were calculated for each question
by academic field of study, and analyses of variance were run for each
question to assess differences among the six fields. The various
ratings were correlated within questionnaire sections. For example,
within the section on reasoning-skills, the ratings of frequency-and
importance were correlated; within the section on reasoning errors,
the ratings of frequency and seriousness were correlated. -
Finally, within each section (and for each kind of rating), the
data were factor analyzed to effect some reduction in the large number
of questions. A principal axis factoring, with squared multiple
correlations as the initial estimates of communalities, was used to
determine the number of factors to be retained for each section,
according to both the magnitude of the eigenvalues and the breaks in
their size. (Our inclination was to err on the side of retaining too
many factors at this exploratory stage.) Various numbers of factors
were then rotated according to the varimax criterion. Although other
oblique rotations could have been used also, it was felt that-the
detection of uncorrelated factors would best serve the objectives of
further test development.
The Sample
A total of 165 chairpersons (65% of those contacted) nominated a
total of 297 faculty members, of whom 255 (86%) returned usable
questionnaires. The response rates across fields were generally
comparable.
Full professors constituted a slight majority of the responding
sample (51%) ; associate professors made up the next largest proportion
(34%). About 13% were assistant professors, and the remaining small
proportion were deans, associate deans, or lecturers.