Putting ‘political’ back in political trust: an IRT test of the unidimensionality and cross-national equivalence of political trust measures

Much research intro political trust—its causes, correlates and trends—builds on the twin assumptions that trust in a wide range of political institutions is ultimately an expression of (1) a singular and (2) a cross-nationally equivalent underlying attitude. Yet, the widespread assumptions of unidimensionality and cross-national equivalence of political trust is at odds with the dominant conceptual understanding of political trust as a relational concept, driven by subjects, objects, and their interplay. This paper employs Rasch modelling as a direct, strict test of unidimensionality, equivalence and item hierarchy. We test the fit of the Rasch model on political trust items in seven widely used, cross-national surveys (World Values Survey, Afrobarometer, Arabbarometer, Asian Barometer, Eurobarometer, European Social Survey, and Latinobarometro), covering 161 national surveys in 119 countries across the globe. We find that the unidimensional specification of the Rasch model does not fit the standard political trust question batteries. Political trust is not cross-nationally equivalent; trust in specific political institutions is more than a mere indicator of an underlying attitude. This conclusion does not impede cross-national research into political trust; rather it illustrates the need for consistent robustness checks across a range of objects of political trust. Our findings open up new venues for substantive research questions on specific objects of political trust and their relationships. © 2019, The Author(s).