Using Moderated Nonlinear Factor Analysis (MNLFA) to Test and Adjust for Measurement Non-Invariance in the Student-Teacher Relationship Scale

• In Event: Measurement Invariance

Abstract

The Student-Teacher Relationship Scale (STRS; Pianta, 2001) is a frequently employed measure of teacher-student relationship quality. The present study examined the extent to which the STRS is invariant over time and demographic characteristics. Using longitudinal data (K-G6) from 1,140 students, we used moderated nonlinear factor analysis to test for measurement invariance in item intercepts and factor loadings. Results revealed that four of seven items assessing conflict were noninvariant (i.e., biased), with items functioning differentially across age, sex, and maternal education level. Similarly, five of eight items assessing closeness were noninvariant, with items functioning differentially across age, sex, maternal education level, and family income-to-needs ratio. Findings highlight the importance of testing and correcting for measurement noninvariance in the STRS.

Authors

• Sophia Magro, University of Minnesota
  Presenting Author

• Meriah DeJoseph, Stanford University
  Non-Presenting Author

• Glenn Roisman
  Non-Presenting Author