Examining Expectancy-Value Beliefs and Traces of Engagement in Science as a Function of Prior Knowledge

• In Event: Multimethod Approaches to Understanding STEM Outcomes: The Importance of Scientific Thinking, Motivation, and Engagement

Abstract

Objective & Theoretical Framework
Contemporary Expectancy-Value Theory (EVT) (Eccles et al., 1983) proposes that individuals’ success expectancies and task values directly predict academic performance and choice. Empirical research demonstrates that expectancies and values predict educational and career persistence, performance, and task choice, while perceived costs are maladaptive for academic success (Eccles & Wigfield, 2020; Flake et al. 2015; Jiang et al., 2018; Perez et al., 2014). While EVT also highlights less proximal factors for success like students’ prior knowledge and behaviors, few studies have examined EVT beliefs along with observable behaviors in authentic learning environments or accounted for the influence prior knowledge exerts on those factors. We examined the relations among EVT beliefs, digital traces of student digital resource use, and exam performance as well as the moderating role of prior knowledge in these relations.

Method & Data
Participants (N = 388) were students enrolled in a large introductory biology course at a southeastern US university. As a proxy for expectancies, self-efficacy was assessed using items (a>.7) designed according to principles outlined by Bandura (2006) while values and cost scales (all a>.7) were adapted from Perez et al. (2014). Pretests were administered via a learning management system (LMS) during the first week of classes and included a prior knowledge measure to distinguish students with high (top 2 quintiles, N=138) and low (bottom 2 quintiles, N=172) prior knowledge. Time-stamped logs of student activity were grouped to reflect use of digital resources representing frequency of engagement in out of class activities, optional work, course management strategies, guided reading questions, and exam preparation.
Outcomes included four course exams. Students were allowed to drop their lowest exam score, typically the first, which served as a calibration opportunity for students to adjust their expectations[1] while the fourth exam was cumulative and could not be dropped. Thus, we first conducted a hierarchical regression to examine the variance in achievement explained by prior knowledge, EVT beliefs, and behaviors (Table 1), then conducted regressions for initial, unit, and summative exams with the overall sample and high and low levels of prior knowledge subgroups (Table 2).

Results & Significance
Values and psychological cost predicted achievement of learners overall and in the low prior knowledge group; student perceptions of self-efficacy, values, and costs predicted little variance in achievement among high prior knowledge learners. Digital engagement behaviors predicted performance on all exams for all groups and explained more variance among high prior knowledge learners than in overall and low prior knowledge samples. Low prior knowledge learners’ engagement behaviors explained more variance on final exams than on prior exams. Self-efficacy, values and costs appear to have stronger relations with performance among students with lower prior knowledge, and behavioral engagement further contributes to all models explaining academic achievement.

Authors

• Michael Berro, University of North Carolina - Chapel Hill

• Matthew L. Bernacki, University of North Carolina Chapel Hill

• Tony Perez, Old Dominion University

• Robert D Plumley, University of North Carolina - Chapel Hill

• Kelly Hogan, University of North Carolina - Chapel Hill

• Mara Evans, UNC Department of Biology

• Kathleen M. Gates, University of North Carolina - Chapel Hill

• Abigail Panter, University of North Carolina - Chapel Hill

• Jeff A. Greene, University of North Carolina - Chapel Hill