The construct validity of AC dimensions is a well-worn research area
over the past 30 years. Sackett (2012) likened the current state of the AC
literature to a projective personality measure, where the interpretation of
the extant literature reflects the researchers’ underlying assumptions. A
substantial source of confusion in the AC literature and elsewhere is the
frequent failure of the CTCM model, the most accurate analytic tool for
evaluating construct validity, to converge to an admissible solution. Our
results indicate that the inability to support CFA models that include dimensions
should not be interpreted that ACs do not measure dimensions.
Instead, when the number of indicators was increased, allowing for a full
set of model comparisons, we found clear factor analytic evidence for
the presence of dimensions. On the other hand, regardless of how ACs
are scored, exercises also appear to play a key role. Thus, these findings
point to the importance of considering both exercises and dimensions as
substantive variables in AC research and practice. More generally, these
findings illustrate a useful approach to increase the likelihood of admissible
CTCM models when evaluating construct validity in organizational
research.