Survey questions about party competence: Insights from cognitive interviews q

Voter assessments of party competence have become a key explanation of electoral decision-making. However, there are at least three important aspects to understanding responses to questions on issue-speci ﬁ c party competence: comprehension dif ﬁ culties; a lack of well-formed attitudes and relevant information; and the use of response heuristics. We used 20 cognitive interviews carried out in Austria in 2011 to test competence questions. The interviews show us how respondents explain their responses. We ﬁ nd evidence that many people (1) may hold only weak opinions and have little information on issue-speci ﬁ c party competence and (2) may make use of distinct but related concepts, particularly salience and position, when answering questions about competence. We provide recommendations for researchers and survey designers based on our ﬁ ndings


Introduction
Voter perceptions of party competence on specific issues have become an important way of explaining electoral choices and election outcomes.Voters are said to assess a party's or politician's ability to handle or deal effectively with a political issue (Belluci, 2006;Green and Jennings, 2012a;Walgrave et al., 2012).Those parties and politicians deemed to be particularly competent are then endorsed and rewarded in the voting booth (e.g., Stokes, 1963;Fiorina, 1981;Mondak, 1995;Fournier et al., 2003;Belluci, 2006;Green and Jennings, 2012a,b;Lenz, 2012). 2  Competence is also part of how parties compete: over time, parties develop reputations for competence on certain issues, and such parties have been said to 'own' and campaign particularly heavily those issues (Budge and Farlie, 1983;Petrocik, 1996;van der Brug, 2004;Belanger and Meguid, 2008;Green and Hobolt, 2008;Clark, 2009;Walgrave et al., 2012).
Given the perceived importance of competence to vote choice and party competition, measuring voter assessments of party competence on key issues has become commonplace in election surveys.For example, the most recent surveys carried out by the American, British, Canadian, European and Irish election studies all include detailed questions on perceived issue-specific party competence.However, we know very little about how people answer these questions.When asked to assess the competence of a party or a politician on a specific issue, how do participants in a survey arrive at a response?
Based on existing research on individual attitudes and survey response strategies, we suggest that there are at least three important aspects to understanding responses to questions on issue-specific party competence.Each of these may influence measurement quality as they might affect the responses recorded in the survey.
First, differences in responses may arise not from differences in opinion but from differences in comprehension, so what respondents think the terms used in the question mean (Tourangeau, 1984;Tourangeau and Rasinski, 1988;Tourangeau et al., 2009).Second, respondents may lack well-formed attitudes or relevant information on issue-specific party competence (Converse, 1964;Krosnick, 1988;Zaller and Feldman, 1992;Chong, 1993;Green and Jennings, 2012a).Yet, they may still give a response to the question (Converse, 1964), often by using satisficing strategies, that is, by giving a response that is good enough rather than as accurate as possible (Krosnick, 1991).Third, responses to competence questions may as a result be characterized by the use of heuristics, which are one type of satisficing strategy.Three particular useful heuristics for issue-specific competence are: competence on other issues (Green and Jennings, 2012a); the importance of the issue to the party and the position it takes; and party attachment in general (Rahn et al., 1994).
Identifying measurement issues regarding opinions on issue-specific party competence is important because we need to know what considerations and opinions these responses are likely to reflect.An awareness of potential problems will also help researchers to improve questions on party competence in future surveys.Moreover, problems with the measurement of competence may affect the validity of conclusions based on analyses that make use of these questions (Goerres and Prinzen, 2012).For example, if answers to party competence questions are formulated based on general party affect, then we must be careful in claiming a unidirectional causal link from competence to vote choice (Johns, 2010;Evans and Chzhen, 2013).
In this paper, we use the results from 20 cognitive interviews used to pre-test a series of competence questions for the Austrian National Election Study (AUTNES).We find that some voters only have weakly pre-formed attitudes and limited information on issue-specific party competence and make use of related concepts such as salience and position to provide an answer.In the conclusion, we suggest some potential ways of addressing these measurement issues in issue-specific party competence questions.
We begin by describing in detail the potential ways that voters can answer questions on issue-specific competence.Next, we present the cognitive interviewing method.Then, we assess the results of our research and conclude with recommendations for researchers and survey designers.

Answering questions on issue-specific party competence
Questions about issue-specific party competence can be seen as questions about attitudes, and research into survey response has theorized that answering such questions proceeds in four steps: comprehension, retrieval, judgment and response (Tourangeau, 1984;Tourangeau and Rasinski, 1988;Tourangeau et al., 2009). 3Our study focuses on the first three steps of this process.

Comprehension
First, respondents may differ in what they understand by the words used in the question.Surveys use different words to get respondents to provide issue-specific competence assessments.For example, they are asked which party is best at 'handling' an issue (British Election Study) or at 'dealing with' a problem (American National Election Study).Personal definitions of these words and terms may vary across respondents.If this is the case, this introduces potential measurement error into voter assessments of competence.As a result, we cannot be sure if differences in responses are due to differences in opinion or just differences in understanding.

Retrieval and judgment
After understanding the question, respondents need to arrive at a judgment regarding issue-specific competence.The memory model and the on-line model of information processing provide two different perspectives on how this occurs.The memory model suggests that respondents store information in their memory; this includes various kinds of considerations, beliefs, feelings or impressions (Hastie and Park, 1986;Zaller and Feldman, 1992).To answer attitude questions, respondents then have to retrieve this information from their memory.Retrieval also means evaluation and selection: the more appropriate and accessible a consideration is, the more likely it will be used to provide a response (Zaller and Feldman, 1992;Tourangeau et al., 2009).A respondent's judgment is then how she combines these considerations in order to reach an overall conclusion.This also means that responses may lack stability if respondents have multiple considerations on the issue (Zaller and Feldman, 1992;Tourangeau et al., 2009: p. 181).For example, Chong (1993) used cognitive interviews to show that, on the issue of civil liberties, people first gave top-of-the-head answers if they were not particularly familiar with the issue.However, they often changed their opinion after thinking about the question at greater length (see also Hochschild, 1981).In sum, in the memory model respondents need not have a pre-formed opinion on issuespecific party competence, but they do need to have some information from which to sample in order to form such a judgment.