Predicting Student Success: The Development of Psychosocial Measures Using the College Student Inventory

Year: 2017

Author: Baier, Stefanie, Sawilowsky, Schlomo, Brockmeyer, Monica

Type of paper: Abstract refereed

High school academic success, standardized test scores, and college entry exams are used to narrow the pool of applicants, admit students, or determine if students should be accepted into developmental programs. A variety of variables are predictive of student success, including non-academic variables (Barneveld et al., 2012). Support services help students to have an increased sense of belonging (Strayhorn, 2012), and engaged learners are more likely to persist in college (Kuh, 2008). Psycho-social variables such as mentoring support, and increased self-efficacy have been found to help students' intent to persist (Baier, Markman, Pernice-Duca, 2016).
The 99 item College Student Inventory (CSI) was developed to study newly admitted college students. It was designed to help determine dropout proneness, academic stress, and predicted difficulty for interventions. It includes about a variety of psycho-social variables.
This presentation shows the process of developing subscales for students' self-efficacy, perceived mentoring support, social connectedness, and perceived student support needs from the CSI based on a student sample from an U.S. urban research university. For study 1, an exploratory factor analysis was employed for an academically vulnerable student sample (N=113) which yielded two reliable subscales for perceived mentoring (Cronbach Alpha = .73) and self-efficacy (CA = .58). For study 2, a first-time- in-any-college-student sample (N=1311) led to the development of a reliable subscale on perceived student support needs (CA = .89). Further analyses revealed group differences in sex (females having greater support needs than males, p = .02, d =.18), race/ethnicity (African American having greater student support needs than other ethnicities, p = .00), first-generation student status greater than other students (p = .00, d = .18), and socioeconomic status (students on Free and Reduced Lunch greater than students not on Free and Reduced Lunch, p = .00, d = .28).
Recommendations are given to indicate how university personnel can best use available data sources to develop reliable subscales for their unique and general university student population, as well as how to identify student subpopulations with higher support needs.