"The Role of Non-Cognitive Variables in Identifying Community College Students in Need of Targeted Supports"

Authors: Loris Fagioli (Irvine Valley College), Rachel Baker and Gabe Avakian Orona 

Abstract

​Measuring students’ self-reported non-academic aptitudes and beliefs—skills, traits, or knowledge that are not inherently academic, such as self-efficacy, grit, and conscientiousness—is one area that has been gaining popularity over the past few decades. Researchers and practitioners have argued that academic data, and placement tests in particular, may not adequately capture students’ aptitude and potential for success. In this study, we examine if the use of self-reported non-cognitive beliefs and aptitudes could improve the targeting of support in community colleges. We rely on a unique data set in which almost 3,000 incoming students at two community colleges were asked a battery of questions regarding their perceptions of their own non-academic aptitudes and beliefs. We follow these students for two years to track their academic progress and success. In addition to information on course enrollment, success, and persistence, we also have data on the students’ use of on-campus resources. Together, these data allow us to ask and answer the following research questions:

1. Can the use of non-academic measures improve predictions of academic success and persistence in community college over the rich set of academic background and test score variables to which schools usually have access?
2. Do behavioral mediators, such as visiting a counselor or tutor, or the number of units attempted in the first-term, explain the relationship between the non-academic measures and outcomes?

This paper offers three unique contributions over past work: (1) we focus our study on a diverse group of students enrolled in community colleges and examine to what extent a wide range of non-academic measures could improve predictions of academic failure across groups of students; (2) our detailed data allow us to examine both longer term outcomes and the more proximal outcomes that can help us understand the patterns that lead to these longer term academic successes and failures; and (3) we capitalize on a large and diverse sample and have access to the students’ full academic record from the schools (rather than self-reported outcomes).

To answer these research questions, we fit a series of regression models, first including all seven non-academic measures and then including our two composite factors, to determine if NCVs are jointly predictive of important academic and persistence outcomes beyond predictions based on past academic achievement. We do this by predicting post-secondary outcomes (GPA, persistence, and completion) using the rich set of demographic and academic variables that are often available to college placement offices, including high school GPA and placement test scores. We then add a set of non-academic measures (or the factors, depending on the model) and observe if this new set of information adds meaningfully to our predictions.

NCVs tend to be significant predictors of community college outcomes; however, the practical utility of these measures is questionable. Provided the desire to best serve a growing and diversified student body, community colleges do not gain much in their efforts to more accurately assess student abilities by using NCVs. Future research should confirm these results with studies in additional settings.