AEFP 45th Annual Conference

Toward a Meaningful Impact through Research, Policy & Practice

March 19-21, 2020

Does beginning teacher mentoring impact their retention?

Yujia Liu, University of California, Irvine,

Beginning teachers and teachers of color are typically retained at lower rates compared to their counterparts, which generates large costs for states and also harms students’ learning (Guarino, Santibañez, & Daley, 2006). Mentoring programs are provided to help accelerate new teachers’ professional growth and to increase their willingness to stay in the teaching positions. Current research on the impact of mentoring programs shows mixed results. Some studies find positive impacts on teacher retention (Henke, 2000) but some find that the impact on teacher retention is not significant (Glazerman et al., 2006, 2008, 2010).
This study investigated the impacts of a beginning teacher mentoring program implemented in certain districts in one of the western states in the United States. The beginning teacher mentoring program started from the 2008-2009 school year. It aims to improve the beginning teacher retention rate, increase the diversity of the teaching workforce, and boost teacher effectiveness. The program targets beginning teachers who have less than two years of teaching experience. Each year districts apply for grants from the state department of education and receive funding for a one-year mentoring program.
We used a difference-in-difference design with district by year fixed-effects models to account for staggered and rotating funding for this program across school districts. Specifically, we have three research questions: (1) How does participating in the mentoring program impact school districts’ one-year teacher retention rate of beginning teachers and beginning teachers of color? (2) Does the impact of the program vary by district level teacher diversity? (3) Does the program impact teacher retention in years two through five?
The state department of education provided administrative records of teachers (2007-2017), including their demographic characteristics, experience, average wages, working assignment, etc. Teacher employment records are used to compute the retention rate. A list of districts who received a mentoring program grant (2009-2016) was obtained from the state department of education.
Out of more than 200 districts, we included only those districts that had more than five teachers in any years between 2008 to 2016, leaving a sample of 186 districts. We also evaluated a sub-sample of high teacher diversity districts (N=18) in which the highest percentage of minority teachers in any of the years was above the national average in 2016 (19.9%).
Our key outcome variables are the retention rates for beginning teachers and beginning teachers of color. We calculated the district level one-year retention rate as the percentage of teachers who taught in the same districts the following year. Similarly, we calculated the retention rate for two-year, three-year, four-year, and five-year. Our key independent variable is the participation status of the in each year from 2009 to 2016. We have the school year of 2007- 2008 as our control year.
To control for covariances, we included district level teacher and student characteristics. Teacher characteristics of each school district include the percentage of females, the percentage of minority, average age, average experience, and average salary. Student characteristics of each school district include student enrollment, the percentage of students eligible for free and reduced lunch, the percentage of females, the percentage of minority, and four-year graduation rates. There is no significant difference on these variables among districts who ever took the mentoring program and who never took the program.
Prior reports comparing retention rates across districts with and without the program in a given year showed some positive effects. However, our study comparing districts with and without programs across years indicates no significant differences in one-year retention rates of either beginning teachers or beginning teachers of color. We only found significant differences on the retention rate of beginning teachers of color among relatively high teacher diversity districts when controlling for all teacher and student characteristics. The program seems to have an impact on the three-year teacher retention rate, but the results are not consistent for two-year, four-year, or five-year retention rates.
Limitations of the research include a lack of data about which teacher individual participated in the program and the fact that teachers’ decisions of retains are affected by more variables than those observed in our study.



It would be interesting to pursue your results with a qualitative follow-up of interviews with teachers for whom the mentoring did and did not work. As a BTSA mentor, I had very different experiences with my various beginning teachers for a wide range of reasons. It may be enlightening to dig in and determining some more latent variables that may be influencing the process.

Thank you, Allison, for sharing your experiences! This is a great point and I do believe qualitative follow-up interviews will help us understand more about the mechanisms!

Hello, that's a really interesting study! If you get detailed information about who gets the mentoring and how the programs are operated, it will provide more meaningful policy implication. I am just wondering whether there is systematic mechanism on applying and getting the program fund. For instance, if districts which have large portion of beginning teachers due to undesirable working conditions (e.g., high poverty level) are more likely to get funding for the program, the effect on attrition may be upwardly biased. I am also curious whether the treated and control groups showed paralleled trends in teacher attrition before the treatment group gets the program funding, which is one of assumption for causal interpretation. You may utilize event study model for this. Again, thank you for sharing your paper. Please let me know if you have any further questions or comments through

Thanks, Minseok! I appreciate the helpful suggestions! I tested the observables of districts that did or did not get funds and there's no significant difference. But I will look for more information on how the funding process was actually operationalized to rule out the potential biases. Also, thanks for the suggestions on the event study model, will test the paralleled trends. Very appreciated:)

This is really interesting work, thank you for sharing! Across Maryland, our school systems have all different approaches to mentoring, and here at the state we're always looking for guidance for them on how to structure their programs. Do you have any ideas about the fidelity of implementation or program quality? Your findings might be affected by whether teachers are actually receiving the mentoring that the program entails. -Dara Shaw (

Add new comment