Alternative Operationalizations of Ideology and Union Membership Sample Clauses

Alternative Operationalizations of Ideology and Union Membership. Second, one may be concerned about the robustness of the results in light of the coarseness of the partisanship variable. A more nuanced measure could change the relationship between politics and private prisons, so I checked the robustness of the results using an alternative operationalization of ideology. I used Shor and XxXxxxx (2011)’s measure of state legislative partisanship23, in which they use state roll call voting data to estimate state legislator ideology. Unfortunately, the data for the first stage loses over half of its observations, as the Shor and XxXxxxx (2011) data only begins in 1993. When I use this measure in place of Republican Control from Table 2.3, the results remain insignificant. Unfortunately, I cannot include this measure in place of the same measure in Table 2.2 because the data only begins in 1993, and many states had already dropped out of the event history dataset because they had already privatized. Nevertheless, it appears that using this more fine-grained measure does not substantively change the results. Finally, one may be concerned about the proxy I calculated to approximate the number of corrections officers that are unionized in each state year. I instead used the percent of the public work force that is unionized, from Xxxxxx and Macpherson (2003), in place of this variable and replicated Tables 2.2 and 2.3 in the appendix. The OLS results remain insignificant, as well as the variable for public union membership, but in the Xxx proportional hazards analysis, Republican Governor is now slightly 23I averaged the state House and Senate chamber measures of ideology. significant, indicating that states are more likely to privatize when they have Repub- lican governors. Additionally, as the percent of public union members increases, the likelihood of privatizing for the first time decreases. There are two potential inter- pretations of this result. First, perhaps the percent of public union members is a better proxy for the number of unionized corrections officers as the two variables are not highly correlated. Moreover, it could be the case that a stronger union presence overall is helpful at preventing states from privatizing, and not necessarily the pres- ence of unionized corrections officers only. Either way, overall union strength may be effective at preventing the adoption of private prisons initially, but is not effective at slowing the growth once it has begun.