Abstract
Mixed results from evaluations of school reform suggest a need for evidence to explain why some models succeed while others fail. Addressing that need, this study uses structural equation modeling to estimate difference-in-differences models that examine mediating mechanisms for positive effects produced by Innovation Zone (iZone) reforms in Memphis, Tennessee. I find that iZone schools increased peer collaboration between teachers, which resulted in improved student achievement. Also, recruiting effective teachers led to a more positive learning environment and ultimately to improved student achievement. These results highlight peer collaboration, a positive learning environment, and the recruitment of effective educators as important practices that will likely facilitate improved school performance under future school reform plans.
Acknowledgments
The author wishes to thank Gary Henry, Ron Zimmer, Carolyn Heinrich, and Shaun Dougherty for their excellent feedback on this manuscript.
Open Research Statements
This manuscript was not required to disclose open research practices, as it was initially submitted before JREE mandating open research statements in April 2022.
Notes
1 The survey items are TNTP’s copyright and can be used for academic purposes only. Data sharing agreements with TNTP do not allow publication of all survey items.
2 Note that students are the unit of observation throughout this analysis.
3 Note that, in 2015–2016, Tennessee experienced complications from rolling out a new test. In response, the state decided not to report any scores form EOG exams. Therefore, I do not use test scores from that year in this analysis.
4 Values above 0.95 for CFI and TLI and below 0.08 for RMSEA were considered acceptable to good fit (Hair et al., Citation2010).
5 Note that estimates from models using weights for schools based on student enrollment are nearly identical to the primary results reported here.
6 Note that using sub-samples does not substantially affect statistical power because the number of schools with one year of post-turnaround data is the same as the number of schools with two or more years of data post-turnaround. Thus, the number of schools in each sub-sample are the same.