Confirmatory Factor analysis (CFA) was used
to validate measures of constructs for
developing the measurement. CFA is a more
effective method for assessing unidimensionality
than exploratory factor analysis, coefficient
alpha, and item-to-total correlation. The purpose
is to ensure unidimensionality of the multipleitem
constructs or low item-to-constructs and
to eliminate unreliable items (Bollen, 1989).
Items that loaded on multiple constructs were
deleted from the model prior to testing. The
measures underlying the constructs are given
in Table 1. The measurement model was
analyzed using the SAS program and CALIS
procedure. An adequate fit was achieved for
the measurement model. The chi-square
to df freedom ratio = 1.823, the Bentler’s
Comparative Fit Index (CFI) = 0.9218,
Bentler and Bonett’s Non-normed Fit Index
(NNFI) = 0.9115, all of the t-statistics for the
indicator variables were greater than 2.576,
significant at p < 0.01, and no standard errors
were near zero. The confirmatory factor
analysis resulted in the elimination of a few
individual items (V6 and V7) because of low
factor loadings or high residuals. This result
was not surprising because many of the
survey items had been developed specifically
for the study, and other items had been
adapted from other literature streams.