Exploratory Measurement Invariance: A New Method Based on Item Response Theory

Methods of assessing measurement invariance can be time consuming and laborious. In this study, we show that a large sample typical of organizational research can be divided into sub-samples across which measurement invariance is not present. However, attempts to use this sub-group membership to predict the results of measurement invariance analyses across demographic groups did not prove useful. Implications for the broader conceptualization of measurement invariance are discussed. 1Test items are said to be free of differential functioning (i.e. are invariant) if they are related to the constructs they measure in the same way across multiple conditions or groups of people. A great deal of recent research has attempted to identify group boundaries across which instruments do not function equivalently (Vandenberg, 2002). However, progress in this area has been slow as comparisons of differential functioning are typically quite laborious, usually involving comparisons between only two groups or conditions at a time. Because it is difficult or impossible to anticipate whether a specific group boundary will be a source of differential functioning, researchers have had little choice but to test for differential functioning across all group differences. Previous research has examined differences by sex (Powers & Jones, 1984)