Document Type



Doctor of Philosophy



Date of Defense


Graduate Advisor

Kathleen Haywood, Ph.D.


Clark Hickman, Ed.D.

Shawn Woodhouse, Ph.D.

Jennifer McCluskey, Ph.D.


Students who fail a course early in their college careers are at risk of leaving college before completion of their programs. Little is known about this high-risk group and why some members of this group complete their programs while others do not. This research addressed this gap in the college-level persistence literature by focusing first on differences between completers and non-completers, then on differences between students who fail a course early in their college careers but become completers or non-completers. The goal was to understand the protective value of individual traits beyond the prediction of risk based on membership to a high-risk group.

The National Education Longitudinal Study (NELS: 88/2000) database was employed and yielded a rejection of the first two Null hypotheses in this study. It was hypothesized that differences in self-concept, social supports, or locus of control scores exist between completers and non-completers of post-secondary programs. It was also hypothesized that there are gender differences in self-concept, social supports, or locus of control scores between completers and non-completers. Further, these hypotheses were proposed to hold for both four-year degree or higher and two-year certificate/degree or higher students.

The Null hypothesis was not rejected for the subset of participants who earned failing grades. No statistically significant differences were seen for self-concept, social supports, or locus of control scores or gender between students who failed a course and subsequently persisted to completion of a post-secondary program and students who earned failing grades but did not persist to completion. The number of students in the database with data for failing grades was low and was an obstacle to the research. This meant that had the Null hypothesis been proffered for the students who did not fail courses, it would also be rejected for all composite variables for the overall group and by gender.

The actionable data in the results came in the form of gender differences, though not specifically for participants with course failure data. Gender differences were significant for all Locus of Control and Self-Concept composite variables, as well as the second family composition variable (collected on a later date).

OCLC Number