More Than Yes and No: Predicting the Magnitude of Non-Invariance Between Countries from Systematic Features

Measurement Invariance has long been the cornerstone of cross-cultural comparisons. Nevertheless, over time a research tradition has developed in which invariance tests are applied with the stated end goal of finding invariance between measures and an implicit view that non-invariance is a barrier to cross-cultural research. In the current paper we aim to challenge this view and urge researchers to consider non-invariance critically not as barrier, but as opportunity for cross-cultural research. Specifically, we show how invariance effect sizes of items can be used to understand psychometric distances between countries and formulate novel hypotheses on cultural differences. Using a previously published dataset on the cross-cultural comparability of subjective happiness from 59 countries, we show how invariance effect sizes can be used to detect problematic items and variables which shape the psychometric similarity of countries. Focusing on item differences, we showed that negatively worded items are performing markedly worse in cross-cultural comparisons and that this effect is exacerbated if countries are linguistically distant. Additionally, we showed that country level variables such as GDP or environmental factors such as temperature can be used to cluster similarities in psychometric functioning, creating novel possibilities to systematize sources of non-invariance on a granular level