Recognizing the significance of social interactions in shaping human behavior and development, policies and programs often rely on peers and social relationships as mechanisms for inducing positive change. Yet, even in randomized control trials, social spillover can make an effective program appear ineffective, and measuring peer effects poses identification and other challenges, like what social relationships matter and what qualities of peers matter. In this dissertation, I draw from multiple disciplinary perspectives and use multiple analytic approaches to better understand the nature of peer effects and social spillover in the context of two school programs: a college summer bridge program and a mentor-based school engagement program. In Chapter 1, I use quasi-experimental methods to examine whether the academic or social qualities of classmates that students encounter at the start of college matter more in determining students’ college outcomes. I identify peer effects in a setting where college administrators used limited information to assign students to a cohort for a required three-week summer bridge course, which allows me to isolate exogenous variation in classmates, conditional on this information. I find that classmates’ social qualities (as measured by their predicted likelihood of joining a fraternity or sorority), but not their academic qualities (as measured by their predicted first-year GPA), influence students’ subsequent academic outcomes. Additionally, I find evidence that male students exposed to more social classmates respond by persisting and taking easier courses, whereas female students respond by earning lower first-year GPAs. In the next chapters, I add to our understanding of how social spillover works in schools by analyzing mixed methods data from a multilevel field experiment of Check & Connect (C&C) where mentor services were randomized at student, grade, and school levels. In Chapter 2, I document large spillover effects on eighth grade students’ high school choice behaviors where the indirect and direct effects of the program are similar in size. When I explicitly test for spillover, I find that students with only indirect exposure to the program apply to 30 percent more schools and are 25 percent more likely to be accepted to one more choice school than students with no exposure to the program. Yet, this quantitative approach does little to uncover how spillover occurred. In Chapter 3, I draw on qualitative interviews with C&C mentors to explore how spillover from mentors to students outside their caseload could happen to better understand what components of the program may contribute to spillover. I find that mentors describe unstructured and unplanned interactions with non-caseload students (often when non-caseload and caseload students are together) and describe becoming a part of the school community in ways that could influence the school environment and resources available to a broad set of students. Together, this research adds to our understanding of how social ties with peers and other adults in school contribute to program effectiveness and influence the outcomes and behaviors of students, particularly during key school transitions. Findings could inform the design of field experiments as well as policies and programs that aim to improve students’ outcomes in part through social interactions.

Creator

• Turner, Mary Clair Clair

DOI

• https://doi.org/10.21985/n2-8f8x-ff42

Subject

• Education policy
• Economics

Language

• en

Alternate Identifier

• etdadmin_upload_780212
• http://dissertations.umi.com/northwestern:15387

Date created

• 2020-01-01

Resource type

• Dissertation

Rights statement

• In Copyright