Exploring Profiles of Responses for the Sources of Science Self-Efficacy Scale

• In Event: Evaluating the Psychometric Quality of Self-Efficacy Measures With Diverse Item-Analysis Methods

Abstract

Purpose and Theoretical Framework

In the domain of science, self-efficacy is related to middle and high school students’ motivation and achievement (Britner & Pajares, 2001, 2006; Pajares, Britner, & Valiante, 2000) and to college students' achievement and persistence in science-related majors and career choices (Andrew, 1998; Gwilliam & Betz, 2001; Lent, Brown, & Larkin, 1984; Luzzo, Hasper, Albert, Bibby, & Martinelli, 1999). Because of the documented importance of this self-belief, researchers have turned their attention recently to how self-efficacy is formed and developed.
Most research dealing with the sources of self-efficacy have suggested that mastery experiences are the most influential source, which is in line with Bandura’s (1997) own hypothesis. Most researchers have used a variable-centered approach to reach this conclusion, in which they regress self-efficacy on the four sources of self-efficacy simultaneously and examine the unique variance explained by each source. Other analytic tools might enable researchers to examine the origins of self-efficacy differently, however. It may be that patterns exist in how individuals come to develop their self-efficacy. For example, some students may derive their sense of efficacy primarily through mastery experiences. Others may derive their self-efficacy through a combination of two or more sources of self-efficacy. Bandura argued that individuals use various combinatory rules to weight and integrate various types of efficacy-relevant information. Most quantitative research methods used to date have not shed light on the combinatory effects of the sources of self-efficacy. Furthermore, as researchers have documented in the past, the ways in which students interpret efficacy-relevant information from these four sources may differ as a function of one’s demographic affiliations, such as gender, race/ethnicity, and grade level (Britner, 2007; Britner & Pajares, 2006; see also Usher & Pajares, 2008 for a review).
In the present study, we used a unique approach, by combining latent profile analysis and Rasch modeling to discern whether patterns exist in how students interpret the sources of self-efficacy in science and to examine whether group membership might explain these patterns. Our specific goals were as follows:

1. To explore the latent classes for the sources of science self-efficacy that might be revealed among middle and high school students using a unique method of analysis that incorporates latent profile analysis and Rasch Modeling.
2. To test the dimensionality of the sources of science self-efficacy scale across each latent class to evaluate whether the different sources of self-efficacy are tapping the same construct across different latent groups.
3. To explore whether each of the latent groups differed as a function of gender, race/ethnicity, and grade level.

Methods
We sampled 1225 students enrolled mostly in Grades 6, 9, and 10 (a small number of students were enrolled in Grades 11 and 12 because they had to re-take the 9th and 10th grade science class, which they had failed in the past).

Preliminary Results
Preliminary analyses have revealed that there are qualitatively different profiles grouped using the sources of self-efficacy, and that these profiles differ by gender and grade level.

Authors

• Jason A. Chen, The College of William and Mary

• Ellen L. Usher, University of Kentucky

• Jennifer Randall, University of Massachusetts