UC Irvine

Recruiting and retaining high-quality teachers and diversifying the teaching force are pressing needs in many school districts. This dissertation aims to understand the policies and programs that address these needs during the hiring process and early years of teacher employment. Study 1 examines the causal effect of a state-level mentoring program for beginning teachers on teacher turnover. The analysis suggests that this mentoring program affects the retention of novice teachers within schools, districts, and the state in different ways. While the programs have a positive impact on novice teacher retention at the state-level, it has no significant impact on school-level retention and may contribute to an increased likelihood of teachers moving to other school districts. In addition, the results suggest that participation in a mentoring program may have greater benefits for teachers of color and those working in less disadvantaged school districts. Study 2 examines, for those who participated in the mentoring program, whether the better matching of mentors and mentees resulted in better retention outcomes. Evidence suggests that novice teacher retention is significantly improved when mentors have prior experience at the novice teacher’s school site. The benefit of school site matches is particularly pronounced in rural, medium-size, and high-poverty districts. Finally, Study 3 analyzes short essay samples written by teacher candidates to understand whether essay characteristics are associated with applicant background, hiring outcomes, and their on-the-job outcomes. Results suggest that there are a number of significant relationships between essay attributes and applicants’ personal and professional backgrounds. Controlling for applicant characteristics, essay attributes identified by machine learning techniques are significantly correlated with essay scores assessed by district recruiters. Essay scores, as assessed by human resource professionals, significantly predict teacher hiring status and on-the-job outcomes. Essay topics, detected by machine learning techniques, are associated with teacher hiring status and value-added in math but have less predictive power for retention and value-added in ELA.