Designing response scales with multi-trait-multi-method experiments

Abstract

ABSTRACT Split-ballot multi-trait-multi-method experiments are used to evaluate the quality of measurement of different response scales of survey items gauging “evaluation of government services” and “political trust.” The response scales differ by agree/disagree scales, item-specific scales, total number of categories, and the presence of fixed reference points on their constructing extreme items. The Center for Opinion Studies of the Catalan government in Spain conducted the survey in 2011. The best response scale depends on the complexity of the topic and of the formulation of the question. For simple topics and formulations, the format of the response scale has no effect on the quality of measurement.