Stakes that Stick: Persistent Effects of Retention Threat

• In Event: Retention’s Reach: Incentive and Peer Effects Beyond the Gate Grade

Abstract

Grade retention policies create incentives for students to make adequate progress ahead of evaluation. A small body of research suggests that the effects of retention policy extend beyond the 2–10% of students retained each year, as the prospect of retention shapes learning environments and outcomes for a wide population of students (Hong & Hong, 2021; Battistin & Schizerotto, 2019; Roderick et al., 2002).

 

This paper examines whether student achievement gains and behavior change influenced by pressure in “gate” grades persist. To investigate the threat of retention, we exploit the fact that students in Ohio can avoid retention at the end of Grade 3 by demonstrating proficiency on a test administered in the fall of Grade 3 and use a regression discontinuity design to estimate the effect of being subject to the threat of retention during third grade on outcomes. Specifically, we compare students just below the retention threshold in the fall of Grade 3, who face the risk of retention throughout the year, with those just above the threshold, who do not. Because students at risk of retention in Grade 3 are more likely to be retained than their already-promoted peers, we use a mediation analysis approach to identify the direct effect of the retention threat on student achievement and behavioral outcomes in Grades 4–8. As part of this analysis, we add new estimates of the effects of retention on retained students.

 

We find that effects of Grade 3 retention threat continue to improve achievement in Grades 4–8 and are sizable. Compared to effects of 0.106 SD on Grade 3 ELA achievement, effects of 0.068 SD persist in Grades 4 to 7, decreasing to 0.060 SD in Grade 8, before increasing to 0.104 SD in Grade 8. Math achievement effects that persist are larger than the effects of 0.045 SD detected in Grade 3. Grade 4 effects of 0.078 SD fade to about 0.050 SD in Grades 5 to 7 before increasing to 0.075 SD in Grade 8. All estimates are statistically significant at the 0.01 level. Students subject to retention threat in Grade 3 decrease absences and involvement in disciplinary incidents; effects in later grades follow the direction of these effects, though only statistically significant effects are detected in Grades 5, 7, and 8.

 

Additionally, we examine mechanisms through which positive achievement effects persist. We find that students who faced the threat of retention are, in Grades 4 through 8, placed in English Language Arts and Math classrooms with higher-achieving peers and fewer retained classmates. In Grades 4 and 5, they are also assigned to classrooms with lower levels of economic disadvantage, and by Grades 7 and 8, they are more likely to have higher value-added ELA teachers. Additionally, students who faced the threat of retention are more likely to be assigned remedial coursework in Grade 4, and more likely to take advanced math courses in Grades 7/8. Additional analysis will examine heterogeneity in effects and mechanisms by the extent of academic tracking within a school.

Author

• Sarah University - Maxwell Souders, American University
  Presenting Author