Chapter 6 Take-Home Message

• Measurement invariance is important

• Comparisons based on single questions without knowledge of the equivalence can be questioned

• Adapt the measurement testing procedure to the model you want to test

• Don’t think about measurement invariance only when you get the data:

Think about measurement invariance and actively include it during questionnaire design

• Remember MGCFA needs at least 3 indicators per latent variable to test for it.

• Compare distributions and look for substantial deviations of the equality constraints over groups for slopes (factor loadings) and intercepts

• Equivalent loadings (metric invariance) necessary for correlations, regressions

• Equivalent intercepts (scalar invariance) necessary for comparing latent means

• Comparing means and relationships between latent variables across countries does not have to require perfect invariance.