Don’t Rely on the Mean: Investigating Students’ Polytomous Belongingness Scores Using Explanatory Item Response Models

• In Event: Saturday Roundtable Session 1:15 pm
  In Roundtable Session: Innovative Methodologies in Item Response Theory (Table 20)

Abstract

Assessment of students’ social and emotional experiences in higher education has almost exclusively been evaluated using Classical Test Theory. Our study offers guidance on how Item Response Theory (IRT), specifically explanatory IRT, can be used to assess student experiences. Using the Multi-Institutional Study of Leadership dataset, this study evaluates the Sense of Belonging scale by comparing the psychometric properties produced by linear regression, the generalized partial credit model (GPCM), and explanatory item response models (EIRM). We found that there are significant differences in how results can be interpreted using these various methods and advocate for greater use of Item Response Theory, especially, EIRM in analyzing student affairs surveys.

Author

• Pearl Lo, University of Maryland