LTIP adoption and use were analyzed separately, as
recommended by Allison (1984: 43) for the analysis of
multiple events when the causal determinants of each event
are distinct and logically sequential, as adoption and use are.
We used event history analysis to model the occurrence of
LTIP adoption, the first event, and for the reduced set of
adopting firms, we modelled the causal process that
determined LTIP use, the second event. The hypotheses on
LTIP adoption were tested using continuous-time event
history analysis with time-varying covariates (Allison,1 984;
Yamaguchi,1 991). Event history analysis is appropriate when
the data is longitudinal and the phenomenon of interest is a
discrete event, as is LTIP adoption. Since specific dates of
adoption were available and adoption was observed over a
relatively long time interval,minimizing the number of tied
events, a continuous-time,p roportional hazards model was
used (Cox, 1972; Yamaguchi, 1991). The Cox model takes
the following form:
LTIP adoption and use were analyzed separately, asrecommended by Allison (1984: 43) for the analysis ofmultiple events when the causal determinants of each eventare distinct and logically sequential, as adoption and use are.We used event history analysis to model the occurrence ofLTIP adoption, the first event, and for the reduced set ofadopting firms, we modelled the causal process thatdetermined LTIP use, the second event. The hypotheses onLTIP adoption were tested using continuous-time eventhistory analysis with time-varying covariates (Allison,1 984;Yamaguchi,1 991). Event history analysis is appropriate whenthe data is longitudinal and the phenomenon of interest is adiscrete event, as is LTIP adoption. Since specific dates ofadoption were available and adoption was observed over arelatively long time interval,minimizing the number of tiedevents, a continuous-time,p roportional hazards model wasused (Cox, 1972; Yamaguchi, 1991). The Cox model takesthe following form:
การแปล กรุณารอสักครู่..