Group Differences in Regression Intercepts: Implications for Factorial Invariance.

Area of Science:

• Psychometrics
• Educational Measurement
• Psychological Testing

Background:

• Differential prediction studies often analyze group differences in regression slopes or intercepts.
• Measurement equivalence is crucial for valid cross-group comparisons.
• The implications of intercept differences without slope differences for measurement equivalence are not fully understood.

Purpose of the Study:

• To investigate the implications of regression intercept differences for measurement equivalence when regression slopes are equal across groups.
• To provide theoretical conditions under which intercept differences can coexist with factorial invariance.
• To offer testable procedures for identifying such conditions in empirical data.

Main Methods:

• Developed two theorems defining conditions for intercept differences under factorial invariance.
• Utilized multiple-group confirmatory factor analysis (CFA) to test these conditions.
• Applied the proposed test procedures to real-world datasets.

Main Results:

• Demonstrated that intercept differences do not necessarily imply measurement bias when slopes are equivalent.
• Provided theorems specifying conditions for maintaining factorial invariance despite intercept variations.
• Illustrated the practical application of these tests using empirical data.

Conclusions:

• Intercept differences alone do not invalidate measurement equivalence if slopes are equal.
• The theorems and methods offer a framework for re-evaluating findings in differential prediction studies.
• These insights contribute to a more nuanced understanding of measurement invariance and bias detection.