Date of Award

8-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Human Centered Computing

Committee Chair/Advisor

Dr. Bart Knijnenburg

Committee Member

Dr. Eileen Kraemer

Committee Member

Dr. Jinkyung Katie Park

Committee Member

Dr. Nicole Bannister

Abstract

The computing sector's rapid growth has intensified demand for computing professionals, yet undergraduate computer science (CS) programs continue to lose substantial numbers of students, often within the first two years, before they reach degree completion. Attrition is not evenly distributed: women and other underrepresented minority (URM) students remain less likely to persist in CS despite broader increases in enrollment. These persistent disparities suggest the need to understand not only how to broaden entry into computing but also how students' experiences within programs shape retention over time. In this dissertation, I address this critical challenge by focusing on identifying inclusive strategies to improve the retention of URM students in computing. Across five coordinated studies, I trace retention-relevant experiences from early pipeline exposure to undergraduate persistence, identifying which psychological and contextual factors matter, how they differ across demographic groups and course levels, and how they may reinforce one another over time.

Study 1 established an upstream perspective by evaluating six classroom-based AI and cybersecurity literacy modules with middle school students, testing whether early exposure improves cybersecurity/AI knowledge, behavioral intentions, attitudes, and interest in computing pathways. Survey-based pretest--posttest analyses showed declines in knowledge, intentions, and some attitudes, with grade-linked differences in the patterns of change. However, students' post-module ratings and interviews conveyed a contrasting narrative of perceived learning, meaningful conceptual connections, and increased intentions toward safer online behavior. This divergence foregrounded two themes that recur across the dissertation: (a) retention-relevant development is multifaceted and can be missed by single indicators, and (b) learners' outcomes are shaped by context and implementation conditions. These insights motivated a shift from early exposure alone toward identifying the broader set of subjective factors that must be sustained and supported once students enter CS programs.

Study 2 translated that motivation into a diagnostic and measurement-focused undergraduate inquiry by identifying retention-relevant subjective factors and examining how they vary across demographic groups. Moving beyond the single-intervention context of Study 1, I developed and validated a set of societal-level retention scales across three survey studies. These yielded a stable core of constructs --- most consistently computing identity, self-efficacy/confidence, perceived preparedness relative to course demands (current experience), social support, and familiarity with future opportunities --- that can be used to study retention systematically. Importantly, the demographic patterns showed that women consistently reported lower identity, lower self-efficacy/confidence, and less favorable perceived preparedness. Further, early coursework emerged as a vulnerable period where attitudes and intentions were comparatively weaker. These results positioned the dissertation to move from "what factors matter" to "how do these factors combine and translate into persistence intentions as students progress through the CS sequence?"

Study 3 directly followed from Study 2's most persistent disparities by employing social cognitive career theory (SCCT) to test whether gender differences in grit and perceived programming experience relate to intention to complete the CS degree across course levels. The findings showed that women reported lower perceived programming experience across course levels. Also, grit's gender gap changed across the sequence: women began CS1 with higher grit, but by CS3 reported lower grit than men. These upstream differences mattered because grit and programming experience predicted self-efficacy, and self-efficacy shaped outcome expectations and attitudes, which in turn predicted intention to complete the degree. Study 3 thus provided an explanatory bridge from Study 2's documented resource gaps to the downstream divergence in persistence intentions observed by the later courses, revealing how course progression can amplify or reverse gender differences in key persistence-relevant constructs.

Study 4 draws on identity-based motivation and self-determination theory to model key retention factors --- computing identity, self-efficacy, adaptive help-seeking, and social support --- as an interdependent set of experiences that co-occur in qualitatively distinct motivational profiles among CS undergraduates. From the analysis, I identified four motivational and support profiles that separated students sharply on attitudes toward CS and intentions to persist. Specifically, the at-risk/needs-deprived profile showed the weakest outcomes, the thriving/needs-satisfied profile showed the strongest outcomes, and the self-reliant but unsupported profile demonstrated that high intention can coexist with low support when identity and confidence are strong. Crucially, Study 4 provided a person-centered complement to Study 2's mean differences and Study 3's pathways by showing how equity gaps can manifest as differential concentration of students, especially gender-minority students, in profiles characterized by jointly depleted resources rather than by deficits in any single factor.

Study 5 is a longitudinal extension that connects the dissertation's cross-sectional and course-level patterns (Studies 2--4) to the temporal logic needed for designing and timing retention supports. Anchored on Social Cognitive Theory, computing identity research, and help-seeking as academic self-regulation, the study examined the longitudinal stability and reciprocal relationships among computing knowledge, self-efficacy, attitude toward CS, computing identity, and help-seeking across consecutive CS courses, and whether these relationships held across gender. Four of the five constructs were significantly stable, with knowledge being the notable exception. Self-efficacy predicting knowledge was the only cross-lagged relationship replicated across both cohorts. Critically, the pattern of reciprocal influence shifted developmentally across transitions: knowledge played a stronger predictive role early in the sequence, feeding efficacy beliefs and computing identity, while computing identity emerged as the central driver of both self-efficacy and attitude in the advanced transition, consistent with identity consolidation as students progress. Gender analyses revealed meaningful baseline disadvantages in psychological factors among gender-minority students in the advanced cohort, though the primary longitudinal relationships remained robust after accounting for gender.

Together, the five studies advance an evidence-based, theory-grounded understanding of retention as an equity-critical developmental process, from early pathway formation to persistence through degree completion. The dissertation provides measurement infrastructure for studying retention with greater precision, identifies consistent demographic disparities in core retention resources, and integrates complementary explanatory and person-centered evidence about how these resources translate into persistence intentions. Critically, it does not stop at documenting patterns. It interrogates where findings converge and where they create theoretical friction, and it translates the accumulated evidence into concrete, stakeholder-specific guidance for instructors, advisors, program administrators, and researchers. By clarifying what matters, for whom, how these factors work together, and when they shift across the course sequence, this work strengthens the foundation for designing inclusive strategies that can more reliably support URM students' persistence and success in computing.

Author ORCID Identifier

0000-0001-8406-9467

Download

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.