Abstract: The literature on economic voting — which relates the state of the economy to vote choice — operates under the assumption that voters can and do assign responsibility to candidates and parties on the ballot for economic conditions. Yet, the validity of this notion has not been fully explored to date. The few previous studies on this matter have only analyzed attribution of blame, but never of credit, for economic conditions. Another limitation of existing studies is that they have not been able to look simultaneously at all actors to whom voters may attribute responsibility. Using unique data from the 2012 NORC Presidential Election Survey conducted between September 24 and October 22, 2012, we overcome these limitations. This paper analyzes whether the effect of economic assessments on vote choice is mediated by assignment of responsibility for economic conditions. In particular, we find that respondents’ attribution of blame mediates the effects of negative economic conditions on vote choice, but attribution of credit has no such mediating effect. We also find that the full effect of negative economic conditions mediated by blame attributed to the incumbent President is strong, but non-existent for blame attributed to other actors. Finally, we find the full effect of negative economic conditions on vote choice is stronger than that of positive economic conditions.
Abstract: Economic voting is perhaps one of the most studied relationships in Political Science. The last 30 years of research on economic voting based on survey data have employed a question that is ill-suited to study the relationship between the economy and vote choice in a manner that would be consistent with the theory. Focusing on prospective sociotropic economic assessments (i.e. evaluations about the future state of the national economy), the paper dissects the question that is typically used in survey-based economic voting analyses and singles out two specific causes of measurement error: the phrasing and the response scale of the commonly used question. It also proposes a corrected survey instrument - based on survey experiments conducted in the US and Mexico - that consists of a battery of questions that condition on each candidate being in office, and utilizes response scales that disambiguate what respondents mean when they respond that the economy is the same as it is now.
Abstract: Measures of perceptions of the future state of the economy have fed a plethora of studies on the impact of economic conditions on vote choice. Yet, the phrasing of the standard economic prospective evaluation question makes it difficult to determine what information are individuals providing when they answer it. Specifically, when respondents are asked about the future state of the economy in pre-election surveys (while the winner of the election is unknown), it is unclear whether respondents provide an assessment of the future state of the economy under i) their preferred party; ii) under the likely winner; iii) under the current incumbent; or iv) under some other alternative. This paper seeks to explain what is contained in the answers to the standard prospective economic evaluation question. Evidence from the 2005 British Election Studies (BES) suggests that individuals answer this question with an assessment about the future state of the economy under their preferred party with some qualifications: the incumbent party if they prefer that party (regardless of whether they think it will win the election), or a challenger when they prefer that party and think it is also the likely winner of the election. No clear results are found for individuals who prefer a party that they think is unlikely to win. After the election, respondents converge answering about the future state of the economy under the party that won the election. These findings stress the need to replace the current question with a battery of questions that capture the future state of the economy if each one of the parties competing were to win the election. This information would provide the necessary information to test economic voting more adequately.
Abstract: This paper describes and links two conceptual ideas that have evolved in parallel in the fields of survey methodology and survey statistics. Such ideas have been empirically utilized by survey researchers with some disconnect until now. We argue that it is possible to align research strategies aimed to describe populations with strategies that seek to analyze them. Particularly, we unpack theoretical foundations that link the advantages of a survey methodology design for data collection, namely Planned Missingness (PM), with a single framework that takes advantage of a statistical approach known as Multiple Imputation (MI). When grounded on theory, models can be developed to go beyond data description and be used to test multivariate relationships. We discuss and illustrate this PM-MI pairing on an Election Day survey carried out in the 2006 Presidential election in Mexico. The exemplified advantages are applicable to broader settings where similar needs call for designs that cope with similar shortcomings.
"Economic Assessments in Times of Crisis: Are There Effects of External Shocks on Economic Perceptions?"