Analyses
We examined the discriminant validity of our different constructs - the line manager’s
enactment of HR practices, the line manager’s relations-oriented leadership behaviour and the
HR department’s service quality - using confirmatory factor analysis. We developed and
tested three models (Bentler & Bonett, 1980): 1) a model in which the indicators of the
different constructs loaded on a single factor, 2) a model in which the indicators loaded on
their hypothesized factors (with the index for the line manager’s enactment of HR practices
being a single indicator factor), and 3) a model in which the indicators of the two line
management constructs loaded on a single factor, and the indicators of the HR department
construct loaded on a second factor. The results, given in Table 4, show that the three factor
model provides a significantly better fit to the data than the other two models. The goodnessof-fit
indices of this model meet the conventional standards for good model fit (Browne &
Cudeck, 1993; Hu & Bentler, 1998)