Item Details

abstract

Can the content of students’ responses to the “saying-is-believing” exercise they complete as part of social-belonging interventions help us understand “for whom” and “when” such an intervention can be successful? What’s more, can natural language methods be used to understand that content? To investigate these questions, this thesis uses responses from a multisite randomized controlled trial of such an intervention given to high-school students taking their first AP course. I begin by providing a descriptive account of the high-level topics common across student responses. Those topics are discovered by fitting is analyzed using latent dirichlet allocation (LDA), an unsupervised topic-modeling procedure. Next, the numerical representation of the topics was used in predictive models testing the two following questions: (1) are students’ race and social class predictive of response content? and (2) does that content broadly relate to academic outcomes or to intervention effects on academic outcomes? The results show that students wrote meaningfully about their experiences in AP and that natural language methods can be used to capture that meaning. Moreover, what students wrote about was predictive of treatment effects on academic outcomes and was predicted by students’ race and social class. Taken together these findings suggest that what students write about can enhance our understanding of “for whom” and “when” a social-belonging intervention can be successful.

subject

Advanced Placement

Belonging

High-school

Natural Language Processing

Psychological Intervention

Topic Modeling

contributor

Gaudiello, Eric C (author)

Brady, Shannon T. (committee chair)

Cole, Veronica T. (committee member)

Dahill-Brown, Sara (committee member)

date

2020-08-28T08:35:25Z (accessioned)

2020-08-28T08:35:25Z (available)

2020 (issued)

degree

Psychology (discipline)

identifier

http://hdl.handle.net/10339/96951 (uri)

language

en (iso)

publisher

Wake Forest University

title

“FOR WHOM AND WHEN?”: USING NATURAL LANGUAGE PROCESSING TO UNDERSTAND WHAT STUDENTS WRITE ABOUT IN SOCIAL-BELONGING INTERVENTIONS AND WHETHER THE CONTENT OF THEIR WRITING CAN PREDICT INTERVENTION BENEFITS

type

Thesis