Methodological Artifacts in Measures of Political Efficacy and Trust: A Multiple Correspondence Analysis

Many authors report a positive relationship of education andpolitical interest with political efficacy and trust, but itis well known that both of the former are associated with responsestyles, such as a tendency to "strongly agree." Since they arerelated to both a substantive concept (political efficacy andtrust), and to methodological effects (agreement bias and atendency to give non-substantive responses) it is importantto assess whether the substantive relationship is due to methodologicalartifacts. Applying multiple correspondence analysis to the1984 Canadian National Election Study, we will discuss a methodwhich allows to test a set of items for measurement effectssuch as ordinality and response sets. In the given example,ordinality of the political efficacy and trust items could beconfirmed only for politically interested respondents. For respondentswith low political interest, there is clear evidence of a responseset resulting in a tendency to "strongly agree" regardless ofthe direction of the items. Taken together, these findings callinto question the substantive relationships reported in theliterature.