Academic journal article Academy of Educational Leadership Journal

Measurement Invariance of Ten Constructs of Pre-Matriculation Freshman Attitudes to College Student Attrition

Academic journal article Academy of Educational Leadership Journal

Measurement Invariance of Ten Constructs of Pre-Matriculation Freshman Attitudes to College Student Attrition

Article excerpt


This study evaluated the measurement equivalence/invariance (ME/I) of a ten-construct factorial measurement model of college student attrition based on self-reported attitudes of entering college freshmen. Establishing metric invariance is the first step towards validation of constructs relevant for an early warning system to prevent college student attrition. The study uses a survey administered annually from 1995 to 1999. This allowed students from the last cohort either to graduate or to drop out by 2004. It discusses ten constructs that resemble the Cabrera, Nora, and Castaneda (1993) model of first-second year of college student persistence. The Cabrera et al. (1993) model claims that college persistence is affected by external factors or student background variables, and endogenous factors such as academic integration, social integration, institutional commitment, goal commitment, academic performance, and intent to persist. We included additional background measures of high-school attitudes towards academics. We also introduced measures of social integration, political interests and concern-for-the-disadvantaged because these were suitable to the type of college from which we drew our sample. Results of the measurement invariance tests revealed full metric invariance and validity for seven of the ten constructs of our model. We discuss implications for future research.


What pre-matriculation attitudes could be used to build a consistent early warning system to prevent college dropout? That is, what constructs measured with pre-matriculation surveys could anticipate college student attrition with the same degree of measurement equivalence from year to year? A main stream of the college retention literature focuses on validation of predictive models of potential dropout students using first-to-second year college students surveys Astin, 1975; Bean, 1982; Cabrera, Castaneda, Nora, & Hengstler, 1992; Cabrera, Nora, & Castaneda, 1993; Tinto, 1975). Recent research emphasizes the evaluation of pre-matriculation surveys to establish early warning systems aiming to detect potential dropout students (Beck & Davidson, 2001; Porter & Umbach, 2006). This study contributes to the latter research stream by assessing the measurement invariance of constructs used in an early warning system of college dropouts.

Evidence of a highly predictive model of student attrition has been presented in previous research (Glynn, Sauer & Miller, 2003). We used the Glynn et al. (2003) measurement instrument that included items taken from the Cooperative Institutional Research Program (CIRP) Survey plus items from in-house survey instruments utilized internally in the past. Recent published research has provided preliminary indications that the factor structure is highly consistent over time having established evidence configural invariance(Glynn, Sauer, & Miller, 2005), yet no formal statistical tests were used to evaluate the metric or scalar invariance of the factor structure.

The purpose of this study is directed toward the first-step in the model estimation process, namely, to conduct statistical tests of measurement equivalence/invariance (ME/I) of the factor structure to verify invariance across attrition cohorts and over time. The rationale for determining time invariance is that the constructs used in the model must consistently measure the same latent characteristic in order for the construct to have validity as a measure of what it purports to represent. The rationale for the cohort invariance is to provide a basis for testing mean differences of constructs across attrition cohorts.

We begin by reviewing the retention model literature, as this is the basis for incorporating the constructs that are included in any valid model of student retention. We next review the literature on measurement invariance then proceed to a discussion of recent findings in the literature regarding the application of our freshman survey data to empirical application of a model of student retention. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.