Increased perception of student misbehavior also has a small, positive association with the likelihood that a new teacher switches schools. Discussion SASS/TFS was designed to provide a cross-sectional snapshot of the teacher workforce. Since the data are not longitudinal in nature, the analysis cannot account for the nonrandom assignment of teachers to schools. While the analysis attempts to include a comprehensive set of controls to account for the social advantage and disadvantage of schools, it is impossible to include an exhaustive set of controls, and there is likely some amount of bias in the results. While the analysis makes a contribution to the turnover literature, the results should be interpreted with some caution. The descriptive results on teacher transitions reveal that teacher churning (switching schools) accounts for more of the annual teacher turnover than actual attrition. The nationally representative data used in this paper indicate that 1.5 times as many teachers move to a different school as opposed to leaving the field altogether. This has important implications for policy makers that seek policy levers that limit the amount of turnover in the field. As hypothesized, and demonstrated in this paper, the transition decisions made by movers are determined by different factors than the transition decisions of leavers. Once the effects of the perceived workplace conditions are allowed to vary by movers and leavers , a simple comparison of Tables 4.4 and 4.5 to Tables 4.6 and 4.7 demonstrates that the movers are driving nearly all of the statistically significant results. The logit models forced movers and leavers together in the dependent variable. Looking at the full sample of teachers, the marginal effect of administrative support is the same in the logit model as it is for the movers group in the multinomial logit model. A similar phenomenon occurs with the new teacher sample. The marginal effects of administrative support and behavioral climate are only slightly larger in the logit model ( − 0.012 and 0.004) than the effects for the movers group of the multinomial logit model ( − 0.009 and 69
0.002). Further, none of the perceived workplace conditions reaches statistical significance for either group of leavers in the multinomial logit models. While administrative support and behavioral climate demonstrate statistical relationships with the transition decisions, the marginal effects of these variables are substantively insignificant. Using the full sample movers group an example, the marginal effect of administrative support is roughly 1/16 th the marginal effect of union participation. Additionally, administrative support has an extremely small variance, from 0.10-0.36 across leavers , movers , and stayers , making it unlikely that this workplace condition provides much utility as a policy lever in mitigating teacher turnover.
- Summer '16
- ........., Math Education