Title:

Testing the longitudinal invariance of the bifactor model using the example of The Perceived Peer Integration Questionnaire

Creator:

Grygiel, Paweł

Publisher:

Educational Research Institute

Date:

2016

Abstract:

In longitudinal data, changes in constructs over time can only be sensibly interpreted if the measured variables are assumed to be invariant across time. This article uses the empirical example of The Perceived Peer Integration Questionnaire (PPI) and three rounds of the nationwide study on School conditions of education effectiveness (N = 4349) to illustrate the use of structural equation modeling to systematically test the measurement invariance of the bifactor model across time. The results prove that the PPI questionnaire is a reliable tool; it is substantially one-dimensional, with a bifactor structure, a longitudinally invariant measurement: configural, metric and scalar, but not strict. We can therefore assume that even though the factor patterns, size of factor loadings and thresholds do not differ significantly in successive editions of the study, the level of reliability of the measurement cannot be considered invariant over time. A lower level of measurement reliability was recorded for grade 3 than for grades 5 or 6. The article also shows the consequences of ignoring the assumptions relating to longitudinal invariance on the results of the statistical analysis.

Bibliographic citation:

Edukacja. 2016, 2 (137)

Subject and Keywords:

sociology ; longitudinal invariance ; bifactor model ; peer relationships ; reliability ; developmental changes

Coverage:

21st century

Language:

pol

Rights:

Creative Commons BY 4.0

Resource Type:

article

Format:

application/pdf

ISSN:

0239-6858