Finally,although analyzing adoption and use separately
appears to represent the most appropriate modeling
approach, for the reasons cited earlier (Allison, 1984), we
also considered an alternative form of modelling the effects
of CEO influence and firm performance on LTIP adoption
and use. We created a dependent variable called "adoption
and non-use of LTIPs" and conducted a discrete-time event
history analysis on the 6314 firm-years of data. The results
of this analysis were consistent with those reported here:
CEO influence was positively related to the likelihood of
adopting and not using LTIPs for the same three influence
measures (t-statistics ranged from 2.18 to 3.54). Also, both
measures of prior performance were significantly and
negatively related to the likelihood of adopting but not using
LTIPs (t-statistics ranged from 2.07 to 3.28).
DISCUSSION
The results of this study suggest that there is a separation