Teitelbaum
Partners in Production or Crime?
29
industrial disputes between management and workers since management and/or workers decided
to terminate the employer-employee relationship. Since the exact point at which the dispute
became a closure is unknown, disputes were deemed to have ended closure following one year of
work stoppage. In other words, the cutoff for the maximum number of days a dispute could last
was set at 365. No work stoppages in the sample that lasted longer than one year resulted in a
negotiated settlement.
The model on dispute duration analyzes the effects of same set of independent and
control variables as the logistic model of industrial violence presented above. I present results of
the model in terms of hazard ratios. A hazard ratio is a summary of the difference between two
survival curves and represents the reduction or increase in the risk of “failure” associated with a
unit change of a given predictor. In this analysis, “failure” is the termination of a dispute
through negotiated settlement. Variables with hazard ratios less than 1 increase the risk of
failure (decrease the survival time of a dispute) and variables with ratios greater than 1 decrease
the risk of failure (increase the survival time of a dispute).
As with the logistic model of dispute violence, since the Major Party and Control
variables form the interaction term, we cannot interpret them directly. Instead, the hazard ratios
for these two variables represent the change in risk of failure associated with each predictor
when the other is equal to zero. Thus, if Control is 0 and all other predictors remain constant, the
presence of an incorporated union (scored “1” for Major Party) reduces the rate of negotiated
settlement by (100% - 25.5%) = 74. 5 percent. Similarly, if Major Party is equal to zero, and all
other predictors are held constant, a one-unit change in Control reduces the rate of negotiated
settlement by (100% - 49.4%) = 51.6%.
Convention