and Bentler's suggested cut-off values for acceptable model fit (i.e., .98
vs. .95 for both the TLI and CFI). The middle row in Table 2 highlights
the traditional model's absolute and relative fit statistics.
As might be expected, the elaborated model (with its additional
predictive pathways) showed slight improvements in relative
goodness of fit (see the bottom row of Table 2). In a chi-square
difference test between the two models, the elaborated model fit the
data significantly better than the original model (ΔΧ
=43.82,
Δdf =4, p=.0000). In terms of spillover effects, however, only the
path between organizational commitment and safety at work proved
to be statistically significant (β=.21, t =6.39). Safety climate did not
impact either of the commitment-related outcomes.
4. Discussion
4.1. Key Findings
The conceptual model proposed in this study was confirmed; that
is, all of the hypothesized relationships were statistically significant
and in the expected directions. In particular, the organization's
occupational safety and health polices and programs (OSHP&P)
impacted both safety climate and organizational commitment, and
as hypothesized, perceived organizational support (POS) partially
mediated both of these effects. Safety climate and commitment, in
turn, were linked to their respective traditional outcomes. That is,
safety climate was related to perceived safety at work and selfreported
work accidents, while organizational commitment was
related to withdrawal behaviors (turnover intention; absenteeism,
and tardiness) and employee vitality. The elaborated model provided
only limited evidence of spillover effects. Organizational commitment
did affect perceived safety at work but not work accidents. Safety
climate failed to impact either of the organizational commitment
outcomes (i.e., employee withdrawal or vitality). The implications of
these findings are discussed below.