Convergent validity is the overlap between alternative measures
that are intended to measure the same construct but have
different sources of undesired variation (Judd et al., 1991). In other
words, if several observed indicators are utilized to measure a
theoretical construct (i.e., latent variable), those observed indicators
should share a good deal of variance (converge together). In
estimating convergent validity for structural equation modeling
studies, examining the standardized confirmatory factor analysis
(CFA) parameters’ estimated pattern coefficient is one method
often used (Marsh and Grayson, 1995). Convergent validity can be
assessed from the measurement model by determining whether
each indicator’s estimated pattern coefficient on its posited
underlying construct factor is significant (Anderson and Gerbing,
1988). That is, if the values in the off diagonal are large, convergent
validity is achieved. Statistically significant large factor loadings
indicate convergent validity.