Investigating the construct validities of the Learning Styles Inventory and the Learning Styles Profile: A latent trait structural equation model
Date of Award
Doctor of Philosophy (Ph.D.)
First Committee Member
James D. McKinney, Committee Chair
The purposes of this study are to (1) determine whether the Learning Styles Inventory (LSI) and the Learning Styles Profile (LSP) instruments measure distinct common dimensions of learning styles as purported by researchers in the field of learning styles theory; (2) determine whether the covariance structures of the LSI and the LSP can be confirmed from data collected in a sample of middle school students; (3) determine if the confirmed latent-variables in the LSI and LSP are independent or related; and (4) determine whether the confirmed factors of learning styles measures have significant correlations with students' achievement scores and teacher-rated academic competence.Relatively few studies were reported that address construct and predictive validities and reliability of learning styles measures. Furthermore, a large body of the literature on learning styles focused on the treatment effects of learning styles programs on educational variables without solid psychometric and constructual foundations of learning styles measures. Therefore, there is a need to determine the validity of learning styles measures in order to adequately assess if significant relationships exist between learning styles variables and other educational variables before the implementation of learning styles in education can be justified scientifically.A sample (N = 1100) was drawn randomly from clusters of Social Studies classes (n = 45) in five middle schools within two public school systems in the state of Florida. A structural equation modeling (Joreskog, 1978; Thompson, 1998; Kerlinger, 1992) was used to examine the factor structures of LSI and LSP. First, factor analyses on the LSI and LSP data sets were conducted to examine whether significant factor solutions existed. Second, covariance structure analysis was used to identify and confirm the factor structures of both instruments that defined the latent variables of learning styles. Third, using the constructs derived from the covariance analysis of the LSI and LSP, a multiple regression technique was used to determine whether the learning styles factors have significant correlations to academic achievement (Stanford Achievement Test scores) and teachability of students as defined by functional behaviors (Classroom Behavior Inventory scores).Findings from the study revealed clearly defined factor structures for the LSI (18 factors at the item level and 4 factors at the scale level) and LSP (11 factors at the item level and 4 factors at the scale level). These factors were somewhat different from those reported by the test developers. Nonetheless, confirmatory factor analysis of the measurement models of the LSI and LSP produced three subscale models that were confirmed to be an adequate fit of the sample data (cognitive, instructional, and environmental preferences). The covariation of the confirmed measurement models of both instruments proved to be a significant structural model that specified and validated the latent-trait of learning styles in a sample data of middle school students. Significant relationships were found between learning styles variables and reading comprehension, mathematics problem solving, and academic competency. Such findings provide empirical support to justify further research on educational practices based on learning styles theory and assessment procedures.
Education, Tests and Measurements; Education, Educational Psychology; Psychology, Psychometrics
Roberts, Owen Anthony, "Investigating the construct validities of the Learning Styles Inventory and the Learning Styles Profile: A latent trait structural equation model" (1999). Dissertations from ProQuest. 3775.