examined five indices and the chi-square/degrees of freedom (χ2Zdf) indicator. In
particular we used the Tucker-Lewis Index (TLI; Tucker & Lewis, 1973), the Goodness-
of-Fit Index (GFI; Raykov and Marcoulides, 2000) and the Comparative Fit Index (CFI;
Bentler, 1990), in which values higher than 0.90 indicate a model with a good fit, and the
Root Mean Square Error of Approximation (RMSEA; Hu and Betler, 1999), in which
values less than 0.06 indicate a model with a good fit. In addition, a parsimonious index
was used, the Parsimonious Normed-Fit Index (PNFI; Mulaik, James, Van Alstine,
Bennet, Lind, & Stilwell, 1989) in which values above 0.80, usually indicate models with
good fit. Caution should be taken in the interpretation of fit indices when a large pool of
observed items is being analyzed, as in this case many parameter estimates will be
constrained to zero when simple factor structure is hypothesized. As O’Connor, Colder
and Hawk (2004) note, with a large number of constrains, fit indices (e.g., CFI) are more
likely to reflect a poor fit, which can be attributed to a large number of trivial
discrepancies between the observed and model implied covariance matrices. The χ2Zdf,
which adjusts for the sample size, is believed to be a better indicator of the model fit in
this situation. Generally a χ2Zdf less than 3.0 is considered good.
The results indicate that there was a rather acceptable good fit with the theoretical
framework of the four-factor model. More specifically, the factor structure of the
applicant sample fits the data well according to different goodness-of-fit indices (χ2(521,
N = 420) = 1252,618; χ2Zdf = 2.40; CFI = 0.85; TLI = 0.84; GFI = 0.83; RMSEA = 0.058;
PNFI = 0.72).
12