Measurement Invariance and the Number of Scale Points in Multi-Item Instruments

Multi-item inventories are widely used in management research to measure latent constructs. Respondents indicate their answers on an agree-disagree scale in response to questions or statements. How many scale points are appropriate in such multi-item instruments is a much debated question. Most of previous research looks at identifying an optimal scale that maximize the reliability as measured by inter-item correlation . However there is little research reported on whether the measurement model and scale characteristics vary with the number of scale points used. In this study, multiple scale configurations using different number of response categories (scale points) are compared using a quasi-experimental research to test whether measurement invariance can be assumed. The results indicate that there is non-invariance between alternate forms as defined by different scale points.This is an important learning both for replication of research as well as in comparison or meta analysis involving several studies with different number of scale points.Researchers must carefully consider this change in scale charactaristics and loadings before they adapt existing inventories with modification in the number if scale points.