3.4.3.2. Underlying structure. As with the data from
the video-based vignettes and the Common Sense
Questionnaire, a CFA was used to test the fit of a
single-factor model to the data. The fit of this model was
marginal (χ2
(90) = 244.42, p =.00; CFI =.92;
RMSEA = .07, 90% CI =.06–.08). The loadings of the
vignettes on the latent factor ranged between .53 and
.70, with the exception of one vignette that had a loading
of .38. The majority of the commonalities for the
vignettes ranged between .28 and .49, indicating that the
vignettes are reasonable measures of the underlying
construct of practical abilities as reflected in the
acquisition and use of general, college-related tacit
knowledge. The vignette with the relatively low loading,
whose commonality was .14, appears to be an
exception.
Examination of the content of this vignette reveals
that participants were asked to indicate their preference
for particular school-related activities, rather than rate
the quality of the activities as a strategy for achieving a
particular goal. As tacit knowledge is applied toward
achieving a particular goal (adapting to, shaping, or
selecting the environment), this vignette did not capture
a key aspect of using practical abilities. The relatively
poor measurement of the construct by this vignette
suggested that it should be removed from further
analyses. The same CFA when fit to the data with this
R.J. Sternberg / Intelligence 34 (2006) 321–350 337
variable excluded had acceptable fit (χ2
(77) = 205.814
p=.00; CFI =.92; RMSEA =.07, 90% CI =.06–.08).
The results of this analysis justified the formation of a
single composite composed of 14 items for further
analyses, representing practical abilities as reflected in
the acquisition of general, college-related tacit knowledge
captured across vignettes. This composite was
formed by taking the unit-weighted average of the
Mahalanobis distances across all 14 vignettes.