Predicting Student Retention via Expectancy Value Theory Using Data Gathered before the Semester Begins

Pamela Bilo Thomas; Campbell R. Bego; Arinan De Piemonte Dourado

Download Paper | Permalink

Conference: 2023 ASEE Annual Conference & Exposition
Location: Baltimore , Maryland
Publication Date: June 25, 2023
Start Date: June 25, 2023
End Date: June 28, 2023
Conference Session: First-Year Programs Division (FYP) - Technical Session 5: Supporting Success 2
Tagged Division: First-Year Programs Division (FYP)
Page Count: 15
DOI: 10.18260/1-2--43930
Permanent URL: https://strategy.asee.org/43930
Download Count: 146

Request a correction

Paper Authors

biography

Pamela Bilo Thomas University of Louisville

visit author page

Dr. Pamela Bilo Thomas is an assistant professor at the University of Louisville, where she teaches introductory programming languages courses in Python and C/C++. She has published in a variety of journals in conferences in her subject area of computational social science, and is interested in using data and machine learning techniques to understand human behavior.

visit author page

biography

Campbell R. Bego University of Louisville orcid.org/0000-0002-8125-3178

visit author page

Campbell Rightmyer Bego, PhD, PE is a cognitive science and engineering education researcher in the Department of Engineering Fundamentals at the University of Louisville's Speed School of Engineering. She studies engineering learning and engineering retention.

visit author page

biography

Arinan De Piemonte Dourado University of Louisville

visit author page

Arinan Dourado, Ph.D., is an Assistant Professor of Mechanical Engineering at the University of Louisville.
Prior to joining UofL, he worked as a Lecturer in his home country (Brazil) for three years, teaching and mentoring low-income, first-generation STEM students from rural communities. Additionally, Dr. Dourado worked as an instructor at the University of Central Florida for two years, primarily serving Hispanic first-generation students. Currently, his working on developing and applying machine learning/artificial intelligence tools to identify and suggest intervention actions to increase student retention and success.

visit author page

Download Paper | Permalink

Abstract

Educating the next generation of college students to be engineers is important for many reasons. From a societal perspective, engineers work together to solve problems and improve the quality of life for many people. For students, an engineering major unlocks the ability to get a job in a growing field with a myriad of opportunities. However, many engineering students do not make it through engineering school. At our university, less than 60 percent of engineering students graduate within six years. Researchers have discovered that many experiences and perceptions lead to attrition, including low grades, feelings of belonging uncertainty, imposter syndrome, financial issues, loss of interest in engineering, and other life stressors. Although it is possible to intervene on any one of these factors and hope to make a difference, designing interventions that target the most important contributors to student decision-making are likely to make the greatest improvement on retention rates. In addition, even though a vast amount of research has pointed to first-semester grades as being of extreme importance, second semester is oftentimes too late to intervene. It is therefore necessary to consider the student experience as a whole in the first semester and work towards identifying the most important causes for student attrition.

To that end, we will use the Situated Expectancy Value Theory (EVT) framework to consider the engineering student experience. SEVT proposes that student experiences lead to three important considerations: (1) their anticipated success in engineering, (2) how much they value an engineering degree, and (3) the perceived financial, emotional, and social cost of staying in engineering school. In this paper, we propose to use surveys deployed in the first week of school that measure the branches of SEVT to identify the most effective interventions for students. Along with their survey information, we will also use information about the student that we can gather from before they entered school, including standardized test scores, GPA, demographic data, and Pell eligibility. As a large public university that enrolled 25.2% racial minorities, 33.7% first generation students, and in which 37.7% of students were Pell-eligible as of Fall 2020, we present an opportunity to study this problem in our own student population.

To make these predictions, we will use an evolutionary neural network approach. The evolutionary approach improves upon itself to create the best possible predictive network, as it fine-tunes the correct number of layers and nodes that can make an optimum prediction. The algorithms will use retention data as output for the network and identify which of our students are most at risk, and which of the EVT facets is most appropriate for intervention. We plan to use the insights from our model to determine the most effective interventions for our student population. In future work, we plan to tailor our interventions to each student using explainable methods.

Citation
Format

Thomas, P. B., & Bego, C. R., & Dourado, A. D. P. (2023, June), Predicting Student Retention via Expectancy Value Theory Using Data Gathered before the Semester Begins Paper presented at 2023 ASEE Annual Conference & Exposition, Baltimore , Maryland. 10.18260/1-2--43930

TY - CPAPER
AB - Educating the next generation of college students to be engineers is important for many reasons. From a societal perspective, engineers work together to solve problems and improve the quality of life for many people. For students, an engineering major unlocks the ability to get a job in a growing field with a myriad of opportunities. However, many engineering students do not make it through engineering school. At our university, less than 60 percent of engineering students graduate within six years. Researchers have discovered that many experiences and perceptions lead to attrition, including low grades, feelings of belonging uncertainty, imposter syndrome, financial issues, loss of interest in engineering, and other life stressors. Although it is possible to intervene on any one of these factors and hope to make a difference, designing interventions that target the most important contributors to student decision-making are likely to make the greatest improvement on retention rates. In addition, even though a vast amount of research has pointed to first-semester grades as being of extreme importance, second semester is oftentimes too late to intervene. It is therefore necessary to consider the student experience as a whole in the first semester and work towards identifying the most important causes for student attrition.

To that end, we will use the Situated Expectancy Value Theory (EVT) framework to consider the engineering student experience. SEVT proposes that student experiences lead to three important considerations: (1) their anticipated success in engineering, (2) how much they value an engineering degree, and (3) the perceived financial, emotional, and social cost of staying in engineering school. In this paper, we propose to use surveys deployed in the first week of school that measure the branches of SEVT to identify the most effective interventions for students. Along with their survey information, we will also use information about the student that we can gather from before they entered school, including standardized test scores, GPA, demographic data, and Pell eligibility. As a large public university that enrolled 25.2% racial minorities, 33.7% first generation students, and in which 37.7% of students were Pell-eligible as of Fall 2020, we present an opportunity to study this problem in our own student population.

To make these predictions, we will use an evolutionary neural network approach. The evolutionary approach improves upon itself to create the best possible predictive network, as it fine-tunes the correct number of layers and nodes that can make an optimum prediction. The algorithms will use retention data as output for the network and identify which of our students are most at risk, and which of the EVT facets is most appropriate for intervention. We plan to use the insights from our model to determine the most effective interventions for our student population. In future work, we plan to tailor our interventions to each student using explainable methods.

AU - Pamela Bilo Thomas
AU - Campbell R. Bego
AU - Arinan De Piemonte Dourado
CY - Baltimore , Maryland
DA - 2023/06/25
PB - ASEE Conferences
TI - Predicting Student Retention via Expectancy Value Theory Using Data Gathered before the Semester Begins
UR - https://strategy.asee.org/43930
DO - 10.18260/1-2--43930
ER -