Economic Perceptions,Presidential Approval,and Causality

Abstract

Citizens process economic information in ways that confirm their prior political beliefs. We therefore see stronger effects of presidential approval on economic perceptions than vice versa. Yet as economic conditions worsen, voters become heavily exposed to negative economic information and more likely to experience conflicting political and economic considerations. As conditions improve, identities and evaluations fall back into sync and political attitudes become more useful as an identity-confirming shortcut. I expect that the effect of economic perceptions on presidential approval grows stronger as economic conditions worsen and the effect of presidential approval on economic perceptions grows stronger as conditions improve. Using panel data from four American National Election Studies, I apply simultaneous equation extensions of the Anderson–Hsiao estimator to test the relationship between economic and political attitudes. Results show that the effect of economic perceptions on presidential approval was substantially stronger during the Great Recession than during the previous three panel studies.

Keywords

economic voting panel data presidential approval

How do macroeconomic conditions affect the relationship between presidential approval and economic perceptions? Two competing theories have emerged from the literature on economic voting, each painting a different picture of the relationship between political and economic opinions. The first, and more traditional, approach places economics before politics—Voters use subjective evaluations of the economy as the basis for updating political attitudes (Fiorina, 1978, 1981; Key, 1966; Kiewiet, 1983; Kinder & Kiewiet, 1979, 1981; Lewis-Beck & Stegmaier, 2000; Weinschenk, 2010). The second approach places politics before economics—Voters’ evaluations of the economy tend to be biased by preexisting political attitudes (Bartels, 2002; De Boef & Kellstedt, 2004; Evans & Anderson, 2006; Evans & Pickup, 2010; Gerber & Huber, 2010; Johnston, 2006; Tverdova, 2012; Wlezien, Franklin, & Twiggs, 1997). Grounded in the theoretical tradition of The American Voter (Campbell, Converse, Miller, & Stokes, 1960) and the psychological literature on motivated reasoning (Kunda, 1990; Lodge & Taber, 2000; Lord, Ross, & Lepper, 1979; Nir, 2011; Rousseau & Snehal, 1999), this latter approach has increasingly become the norm in the study of economic voting. Although these two approaches disagree over the direction of causality, each side has argued adamantly that the relationship moves in either one direction or the other. Until very recently, the field has mostly ignored the moderating role of the economic context for the causal relationship between economic perceptions and political attitudes.

In this study, I examine how the dynamic relationship between economic perceptions and presidential approval is moderated by the broader economic context. During economic recessions, the economy tends to dominate the airwaves as one of the most important issues for voters; pundits and politicians debate the causes of and potential solutions to economic problems, and the media’s attention to economic issues has a significant impact on voter behavior (Goidel & Langley, 1995; Harrington, 1989; Hetherington, 1996; Sanders & Gavin, 2004; Soroka, 2006). As a result, voters are more heavily exposed to negative economic information during economic downturns, which may conflict with their prior political attitudes—attitudes that subsequently become less useful as cognitive shortcuts for processing economic information (Lavine, Johnston, & Steenbergen, 2012). Under such circumstances, I expect the impact of economic perceptions on political attitudes to be the strongest. During better times when voters are less heavily exposed to negative economic information, political attitudes should serve as a more useful cognitive shortcut for evaluating the economy and confirming one’s prior beliefs. Thus, the effect of economic perceptions on political attitudes should be weakest during good economic times. If these expectations hold, then we can expect that attitudes toward the president will matter more for the formation of economic opinions during good economic times than during relatively worse economic times. If macroeconomic conditions can determine the direction of the relationship between economic perceptions and presidential approval, then we will gain a better understanding of when voters are likely to rely on political heuristics for evaluating the economy and when they are likely to hold incumbents accountable for economic performance. Such an understanding is important for scholars who try to explain the role of the economy in public opinion and election outcomes, as well as for political actors who try to anticipate the electoral behavior of voters.

I test this theory using the 1990-1991-1992, 1992-1994-1996, 2000-2002-2004, and 2008-2009-2010 ANES Panel Studies, which are each characterized by distinct economic trends. Using simultaneous equation extensions of dynamic panel models, I test the effect of presidential approval on economic perceptions and the effect of economic perceptions on presidential approval during each panel study. Although not directly modeling objective economic conditions, the distinct economic contexts in which the panel studies were conducted allow us to examine how the dynamic relationship between political and economic attitudes varies across economic contexts. The results suggest that the effects of economic perceptions on presidential approval were significantly stronger following the economic recession of 2008 than compared with more prosperous economic times. Surprisingly, however, the effect of presidential approval on economic perceptions also increased during the economic recession. The most likely explanation for this finding is that rather than presidential approval and economic perceptions having a symmetrical relationship, attitudes toward the president are always an influential source of political and economic attitudes even as the economy worsens and economic information becomes more readily available.

I begin in the next section by reviewing the existing literature on economic voting and how voters use motivated reasoning to simplify the formation of economic perceptions. The second section describes my theoretical expectations, outlining specific hypotheses regarding the direction of the relationship between economic perceptions and presidential approval. The third section describes the data and methods used in the analysis, which is presented in the fourth section. The final section discusses the academic and electoral implications of the findings, as well as potential avenues for further research.

Economic Perceptions and Political Attitudes

The relationship between economics and politics is one of the most well-documented and highly debated topics in the literature on voting behavior and public opinion.¹ An abundance of evidence suggests that the economy does have significant influences on political behavior. Some of the earliest links between the two were established at the aggregate level, with the health of the economy bearing a significant impact on the electoral success of incumbent presidential and congressional candidates (Bloom & Price, 1975; Fair, 1978; Goodman & Kramer, 1975; Kramer, 1971, 1983; Tufte, 1975).² On the individual level, research has shown that voters form political preferences based on evaluations of the national economy (Fiorina, 1981; Kiewiet, 1983; Kinder & Kiewiet, 1979, 1981; Lewis-Beck, Jacoby, Norpoth, & Weisberg, 2008; Nadeau & Lewis-Beck, 2001; Sniderman & Brody, 1977) and their own personal finances (Fiorina, 1978, 1981; Klorman, 1978; Lewis-Beck, 1985; Tufte, 1978).³ Comparativists have extended this line of research to show that these individual-level economic effects are not unique to single western countries (Duch & Stevenson, 2008; Lewis-Beck, 1988), although van der Brug, van der Eijk, and Franklin (2007) found that such effects vary with party competition and the clarity of economic responsibility. Undoubtedly, the economy does seem to matter not only for the vote but also for individual-level political attitudes. These traditional approaches to economic voting assume a causal relationship that moves from economics to politics, which is a key component for democratic accountability. Normatively, we expect voters to be able to evaluate economic performance and either reward or punish incumbent candidates accordingly.

Critics of these traditional approaches have found an endogenous relationship between economics and politics. On the aggregate level, studies describe a political business cycle in which incumbent politicians pursue specific economic policies aimed at maximizing electoral support in upcoming elections, in the hope that voters will be myopic to the long-term effects of such policies (Hibbs, 1979; Nordhaus, 1975; Tufte, 1978). Other studies have found that the economic policies that meet the electoral interests of incumbent politicians vary across constituencies (Hibbs, 1977; Wright, 2012). Thus, economic policies are endogenous to the electoral interests of political actors.⁴ More recently, De Boef and Kellstedt (2004) show that aggregate levels of consumer confidence tend to be shaped by political factors and media coverage to a greater extent than actual changes in economic conditions. On the individual level, voters tend to perceive the economy more positively when they identify with the party of the president (Bartels, 2002). A major criticism of the traditional approach to economic voting is that such studies typically rely on cross-section analyses, which are inappropriate for making causal inferences about the dynamics of a relationship (Wlezien et al., 1997). Research based on longitudinal data has shown that economic perceptions tend to be biased by political attitudes, including presidential approval, partisanship, and vote intentions (Evans & Anderson, 2006; Evans & Pickup, 2010; Gerber & Huber, 2010; Johnston et al., 2005). Experimental research has also found evaluations of personal finances to be biased by preexisting political commitments (Sigelman, Sigelman, & Bullock, 1991), and more recent evidence suggests that economic opinions are shaped by both political predispositions and expectations of the political future (Ladner & Wlezien, 2007). In addition, the ability of voters to hold incumbents accountable for past performance is constrained by perceptual biases in retrospective decision making (Huber, Hill, & Lenz, 2012). These findings cast doubt on the traditional model of economic voting by questioning the assumption that voters are capable of correctly evaluating the economy and forming political attitudes accordingly.

The evidence for an endogenous relationship between economic perceptions and political attitudes is afforded strong theoretical support by social and psychological models of voting behavior, which tend to view political predispositions as the primary predictor of issue attitudes and electoral choices (Berelson, Lazarsfeld, & McPhee, 1954; Campbell et al., 1960; Green, Palmquist, & Schickler, 2002). It is from this theoretical background that arguments against traditional approaches to economic voting have emerged. The idea of partisanship as a stable attachment formed early in life casts doubt upon the ability or willingness of voters to update their political preferences based on evaluations of economic performance, as Key (1966) and Fiorina (1981) would suggest. Instead, the social and psychological models would argue that evaluations of economic performance depend on an individual’s identification with the incumbent party or other preexisting political attitudes. This approach to economic voting is afforded further support by psychological literature on motivated reasoning, which posits that individuals tend to accept information as true when it confirms their own prior beliefs while rejecting or counterarguing information that contradicts their prior beliefs (Kunda, 1990; Lord et al., 1979; Lodge & Taber, 2000; Nir, 2011; Rousseau & Snehal, 1999). Political psychologists have extended the motivated reasoning thesis to show a strong tendency for individuals to perceive the political and economic world in ways that confirm their prior political considerations (Enns, Kellstedt, & McAvoy, 2012; Enns & McAvoy, 2012; Gerber, Huber, & Washington, 2010; Goren, Federico, & Kittilson, 2009; Jerit & Barabas, 2012; Lebo & Cassino, 2007; Taber, Cann, & Kucsova, 2009).

Theoretical support for an endogenous relationship between economic perceptions and political attitudes is complimented by a fair amount of empirical support. In one of the early empirical studies of the endogenous relationship between economic and politics, Anderson (1995) found that the direction of the relationship between economic conditions and government support varied both over time and across countries. From 1980 to 1988, inflation rates improved under a Republican president. Yet, more than half of respondents to the 1988 ANES who identified with the Democratic Party believed that inflation had gotten worse over the previous 8 years, compared with only 13% of respondents who identified with the Republican Party (Bartels, 2002). Following the Democratic takeover of Congress in the 2006 midterm elections, voters diverged sharply in how they evaluated the national economy even though conditions had not objectively changed much at all (Gerber & Huber, 2010). In November 2004, about 55% of ANES respondents who identified strongly with the Republican Party evaluated the national economy as having improved over the previous year; only about 6% of strong Democrats gave the same positive evaluations. Similarly, only about 24% of respondents to the 1996 ANES who identified as strongly Republican evaluated the economy as having improved over the last year, compared with about 63% of strong Democrats. In each case, however, the health of the national economy had not actually changed substantially.⁵ Evidence from the 2000-2002-2004 ANES Panel Study has suggested that presidential approval, partisanship, and vote choices influence economic perceptions more strongly than they are shaped by economic perceptions (Evans & Pickup, 2010). These examples provide strong evidence that economic perceptions tend to be politically biased. They also reflect a growing consensus that the traditional conception of an economic voter providing democratic accountability through regular evaluations and Bayesian updating simply does not exist (Anderson, 2007).

Despite the overwhelming evidence for motivated reasoning in the processing of political and economic information, scholars have also established mechanisms through which the use of motivated reasoning can be diminished. For example, when individuals are exposed consistently enough to information that contradicts their own beliefs, they eventually reach a tipping point at which their own beliefs begin to shift (Redlawsk, Civettini, & Emmerson, 2010). Similarly, when an individual’s party identification temporarily conflicts with evaluations of the party’s performance, he or she becomes more likely to rely on objective information than partisan cues when making political decisions (Lavine et al., 2012). Recent experimental evidence suggests that perceptual biases among partisans may be diminished when individuals possess specific information about a policy or issue (Bullock, 2011), or when offered incentives for giving “correct” responses (Bullock, Gerber, Hill, & Huber, 2013). Moreover, the broader information environment has been shown to moderate partisan disagreement over economic facts—when available economic facts are unanimously positive or negative, disagreement is diminished (Parker-Stephen, 2013). Psychologists have also argued that individuals possess a sufficiency threshold, or a desired level of subjective confidence in a decision—when the actual level of confidence in a decision falls below this sufficiency threshold, the resulting confidence gap can lead the individual to seek out more objective information than the cognitive heuristics she would normally rely on (Lavine et al., 2012; Payne, Bettman, & Johnson, 1993).⁶ How might these moderating factors for the use of motivated reasoning be extended to the study of economic voting?

Scholars of economic voting have only recently begun to address how the use of political shortcuts for the formation of economic perceptions might vary over time. Recent evidence from the U.K. suggests that economic perceptions are most likely to have an exogenous effect on government approval when economic conditions are relatively healthy (Chzhen, Evans, & Pickup, 2013). Other findings confirm that partisans are more likely to agree on economic realities during times of crisis (Parker-Stephen, 2013). Aside from these findings, very little attention has been paid to how aggregate economic conditions might moderate the individual-level relationship between economic perceptions and political attitudes. I extend the conditional motivated reasoning arguments described above to show that not only are voters less likely to rely on political shortcuts during bad economic times but also that economic perceptions during such times can actually lead voters to update their own political attitudes. I elaborate on this theory in the following section.

Accounting for the Macroeconomic Context

I argue that the degree to which individuals use perceptions of the economy to update their attitudes toward the president are moderated by the macroeconomic context. The theoretical underpinnings of this expectation are grounded in recent psychological literature that demonstrates the circumstances under which individuals become less likely to rely on motivated reasoning. Psychologists describe three main goals that guide how individuals process information and make decisions: efficiency, accuracy, and belief perseverance (Lavine et al., 2012). First, individuals want to be efficient—that is, they want to expend as little cognitive effort as possible when processing information and making decisions. Second, they want to be accurate—while preserving as much cognitive energy as possible, they also desire to accurately process information. Third, individuals have an innate desire to process information in ways that confirm their own prior beliefs. How individuals process information and translate it into decision outcomes depends on how they prioritize these three goals. Although partisanship is often the most easily accessible shortcut for processing political information in ways that confirm prior beliefs (Jacobson, 2010; Jerit & Barabas, 2012; Kim, Taber, & Lodge, 2010; Lebo & Cassino, 2007; Nir, 2011; Rahn, 1993; Taber et al., 2009; Taber & Lodge, 2006; Westen, Blagov, Harenski, Kilts, & Hamann, 2006), the tendency to do so depends on the congruence of an individual’s partisan identification and partisan evaluations (Lavine et al., 2012). Lavine and his colleagues argue that when partisans become temporarily unsatisfied with the performance of their party’s leadership, they turn away from the efficiency and belief perseverance provided by partisan cues and look instead for objective information that will promote the goal of accuracy. When partisan identities and evaluations conflict, a confidence gap emerges between the desired level of confidence in a decision and the actual level of confidence provided by the partisan cue. As a result, the individual places less weight on the goal of efficiency and more weight on the goal of accuracy. Doing so requires turning away from partisan cues.

I extend this line of reasoning to argue that as economic conditions worsen, voters become more likely to experience similar conflicts between their partisan identities and their evaluations of government performance. Consequently, they should become more likely to turn away from their political predispositions and seek out other sources of economic information when forming evaluations of the economy. This tendency is strengthened by the increased attention politicians and pundits pay to the economy during a recession. Studies have shown that individuals tend to respond more strongly to negative information than positive information (Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001; Soroka, 2006; Taylor, 1991), and the media tends to devote greater attention to economic issues during bad economic times than during good economic times (Goidel & Langley, 1995; Harrington, 1989; Sanders & Gavin, 2004). Moreover, previous literature shows that such media coverage of economic issues has a significant impact on perceptions of the economy (Goidel & Langley, 1995; Hetherington, 1996; Sanders & Gavin, 2004). We also know that when the information environment consists of mostly positive or mostly negative economic signals, partisan disagreement on economic issues tends to be diminished (Parker-Stephen, 2013). As a result, the times when voters experience conflicts between partisan identities and performance evaluations are the same times when the information environment consists of predominantly negative economic facts. The acquisition of new negative economic information (whether through voluntary or involuntary exposure), by diminishing political disagreement over economic facts, should make economic perceptions more likely to have an exogenous impact on attitudes toward the president. As such, I also expect presidential approval to matter less for economic perceptions when the economy is performing poorly.

I hypothesize that the effect of economic perceptions on presidential approval will be stronger during bad economic times than during more prosperous times. During an economic recession, individuals become heavily exposed to negative economic information. This leads to a disparity between political identities and evaluations of government performance (Lavine et al., 2012), which makes political cues less useful as information shortcuts. Under such conditions, I expect individuals to be more likely to form perceptions of the economy and subsequently update their attitudes toward the president. Thus, we should see the strongest effects of economic perceptions on presidential approval during an economic recession. During more stable economic times, individuals are less heavily exposed to negative economic information and political identities and performance evaluations are more likely to fall into sync with one another. With a lack of economic information, less of an incentive to seek out such information, and a greater incentive to confirm their own prior beliefs, individuals should become more likely to rely on political cues when evaluating the economy. As a result, we should see weaker effects of economic perceptions on presidential approval during good economic times than during bad economic times. If this is the case, we can also expect the effect of presidential approval on economic perceptions to be greater during good economic times than during bad economic times.

Data

Ideally, testing these expectations requires a panel survey that spans a long enough time period to capture sufficient variation in objective economic conditions, and which includes the appropriate survey items. Unfortunately, such data are rare. An alternative approach that can indirectly capture the moderating role of the economic context is to analyze multiple small-T panels that take place under significantly different economic conditions and then compare the results. If my hypotheses hold, then a panel of respondents during bad economic times should yield a stronger effect of economic perceptions on presidential approval than a panel of respondents during good economic times. Although this approach does not directly model changes in objective economic conditions, it provides a good starting point for understanding how the relationship between economic perceptions and presidential approval is moderated by the macroeconomic context.

In the analyses that follow, I examine the reciprocal relationship between economic perceptions and presidential approval using data from four ANES Panel Studies: the 1990-1991-1992 Panel Study, the 1992-1994-1996 Panel Study, the 2000-2002-2004 Panel Study, and the 2008-2009-2010 Panel Study.⁷ During the first three panel studies, representing control groups, the national economy was mostly stable and performing relatively well (in comparison with the most recent panel study). The fourth panel study, representing a treatment group, took place during the Great Recession between 2008 and 2010. Even the 1990-1991-1992 economic recession was relatively mild when compared with the more recent recession. The financial crisis during the 2008-2009-2010 study had an unprecedented influence on the opinions and behavior of voters, parties, and candidates.

While the economy was by no means the only contextual factor that changed between the panels, the drastic economic downturn in 2008 sharply distinguishes the treatment panel from the three control panels. It should be noted that the panels might also be distinguished by a range of other contextual factors, including variation in the party of the incumbent president, variation in relevant foreign policy issues, and most obviously, variation in the personal characteristics of presidents Bush, Clinton, Bush, Jr., and Obama. Nonetheless, the economy was undoubtedly one of the most important issues facing voters and candidates going into the 2008 presidential election, and that salience resonated throughout the remainder of the decade. Economic issues during the 2008-2009-2010 Panel Study influenced the political agenda to a greater degree than any period dating back arguably as far as the Great Depression.

Figure 1 shows the nation’s monthly seasonally adjusted unemployment rate, with the months of each panel wave indicated with vertical dashed lines. During the 1990-1991-1992 Panel Study, the national unemployment rate fluctuated between roughly 6% and 8%, dropping as low as about 5% toward the end of the 1992-1994-1996 Panel Study. During the 2000-2002-2004 Panel Study, unemployment fluctuated between about 4% and 6%. Trends began to change dramatically following the 2008 financial crisis—unemployment rose from about 5% to as high as 10% during the 2008-2009-2010 Panel Study. It should be noted that although unemployment is often used as a popular measure of economic conditions, the trend shown in Figure 1 is not exclusive to the single indicator. A similar trend is reflected in Figure 2, which shows a sharp decline in the University of Michigan’s Index of Consumer Sentiment during the 2008-2009-2010 Panel Study. Thus, although contextual factors besides the economy undoubtedly changed between the panel studies, the national economy was one of the clearest characteristics distinguishing the context of the 2008-2009-2010 Panel Study from the previous three studies.

Figure 1.

National unemployment by month, 1990-2010.

Figure 2.

Index of consumer sentiment, 1990-2010.

In all four panel surveys, economic perceptions are measured using respondents’ retrospective evaluations of the national economy.⁸ Respondents were first asked, “Thinking about the economy in the country as a whole, would you say that over the past year the nation’s economy has gotten better, stayed the same, or gotten worse?” A follow-up question then asked, “Would you say much better [worse] or somewhat better [worse]?” Responses were then coded as follows: 1 = much worse, 2 = somewhat worse, 3 = about the same, 4 = somewhat better, 5 = much better. Presidential approval was measured by asking respondents, “Do you approve or disapprove of the way [president] is handling his job as president?” A follow-up question then asked, “Do you approve [disapprove] strongly or not strongly?” Responses were then coded as follows: 1 = disapprove strongly, 2 = disapprove not strongly, 3 = approve not strongly, 4 = approve strongly. In the 2008-2009-2010 Panel Study, a neutral category (neither approve nor disapprove) was included between “disapprove not strongly” and “approve not strongly,” creating a 5-point scale instead of a 4-point scale. Because approval is coded differently in the most recent survey, caution should be exercised when comparing coefficients across panel studies. To facilitate interpretation, both economic perceptions and presidential approval are standardized in each study to have a mean of zero and a standard deviation of one.

In most analyses of economic voting, we would also need to control for partisanship and basic demographic factors. However, because I employ a differencing approach (described below), observations on any variable that do not change across panel waves are dropped from the analysis. Demographic factors such as education and income can safely be excluded from the analysis. Partisanship can also be safely excluded because it tends to remain fairly stable over time—Its inclusion in the following analysis would result in a drastic decrease in sample size, whereas its exclusion from the analysis does not raise major implications for the substantive findings. In the analyses that follow, missing observations due to panel attrition and nonresponses are dropped via listwise deletion.⁹

Model Specification

In the following analysis, I model presidential approval as a function of economic perceptions and economic perceptions as a function of presidential approval. For simplicity’s sake, I begin by describing models of approval as a function of economic perceptions before moving into the multiple equation framework. The traditional economic voting model (with exogenous controls excluded for space purposes) is typically specified as

A P P R O V E_{i} = β_{0} + β_{1} E C O N_{i} + ε_{i},

for $i = 1, \dots, N$ individuals, where $A P P R O V E_{i}$ is an individual’s approval of the president’s job performance, $E C O N_{i}$ is an individual’s retrospective evaluation of the national economy, $β_{0}$ is an intercept, and $ε_{i}$ is an unobserved disturbance term.

The first problem with this traditional economic voting model is its limitation to a single time period. Cross-sectional designs are often inadequate for making causal inferences because of their inability to track change and stability over time. We know from a long line of literature that political attitudes tend to remain fairly stable over time (Bartels, 2000; Campbell et al., 1960; Green et al., 2002), and can thus expect that prior presidential approval will be a strong predictor of current approval.¹⁰ In the model described above, prior presidential approval is unobserved and thus contained within the disturbance term, $ε_{i}$ . By not modeling past values of approval, the results would reflect biased estimates of the relationship between approval and economic perceptions. To control for the stability of presidential approval, we need two or more observations of the same individuals at different points in time. Panel data allow for the specification of dynamic models, such as

A P P R O V E_{i, t} = γ_{0} + γ_{1} A P P R O V E_{i, t - 1} + γ_{2} E C O N_{i, t} + δ_{i, t},

for $i = 1, \dots, N$ individuals and $t = 1, \dots, T$ time periods, where $T \geq 2$ , $γ_{1}$ is an estimate of the stability of presidential approval from one panel wave to the next, and $δ_{i, t}$ is an unobserved disturbance term. Although controlling for the stability of presidential approval produces a more efficient estimate of its relationship with economic perceptions, another problem is introduced with the use of panel data. In particular, the decomposition of the error term, $δ_{i, t}$ results in two components:

δ_{i, t} = λ_{i} + σ_{i, t},

where $λ_{i}$ is an unobserved individual-specific effect and $σ_{i, t}$ is the remaining idiosyncratic disturbance that varies over time and across individuals.¹¹ The individual fixed effect ( $λ_{i}$ ) includes any unobserved factors that vary across individuals that might influence a given individual’s approval of the president’s job performance. By leaving these individual effects to the error term, the coefficient for economic perceptions will most likely be biased. In some cases, this bias can be corrected using the within transformation, or fixed effects estimation, which removes the individual-specific effects and the intercept by demeaning the equation (Angrist & Pischke, 2009; Finkel, 1995; Hsiao, 2003; Wooldridge, 2006) so that

A P P R O V E_{i, t}^{*} = γ_{1} A P P R O V E_{i, t - 1}^{*} + γ_{2} E C O N_{i, t}^{*} + δ_{i, t}^{*},

where * denotes deviations from individual means. Unfortunately, a new problem arises when we include a lag of the dependent variable $A P P R O V E_{i, t - 1}$ with fixed effects. Nickell (1981) has demonstrated that the fixed effects estimator is inconsistent for dynamic panel models with fixed $T$ even as $N \to \infty$

The small-T large-N problem that plagues many dynamic panel models has been the subject of extensive debate, and the literature on how to handle biases in such models has not been conclusive. An alternative approach to removing the individual-specific effect is to difference the equation, so that,¹²

Δ A P P R O V E_{i, t} = β_{1} Δ A P P R O V E_{i, t - 1} + β_{2} Δ E C O N_{i, t} + Δ σ_{i, t} .

Even still, a new source of bias arises from Equation 5—in particular, the lagged-differenced outcome is now correlated with the differenced disturbance term through their respective components, $A P P R O V E_{i, t - 1}$ and $σ_{i, t - 1}$ . One approach to breaking the correlation between the lagged-differenced outcome and the differenced error term is to instrument the lagged-differenced outcome with some larger lag of the outcome (or a larger lag of the differenced outcome) that is not correlated with $σ_{i, t}$ or $σ_{i, t - 1}$ . The Anderson–Hsiao estimator (Anderson & Hsiao, 1981) corrects for this bias by using the second lag of the outcome as an instrument for the lagged-differenced outcome. As $T$ increases, more instruments become available but, as is the case with many survey panel studies, $T$ tends to remain small even as $N$ increases. In the cases of the following analyses, only three waves of panel data are available—thus, for the Anderson–Hsiao estimator, the only available instrument for $Δ A P P R O V E_{t - 1}$ is $A P P R O V E_{t - 2}$ . Other scholars have argued that a single lag serves as a weak instrument by not utilizing all the information available in a longitudinal data structure, and turn instead to a generalized method of moments (GMM) framework (Arellano & Bond, 1991; Arellano & Bover, 1995; Blundell & Bond, 1998). Although these approaches take advantage of specified moment conditions, they have also been shown in some cases to result in biased and inefficient estimates (Kiviet, 1995). In the cases of the following analyses, GMM estimators and the Anderson–Hsiao estimator did not produce substantively different results. Although the coefficients varied slightly between the two methods, the overall conclusions were the same. For simplicity purposes, only the Anderson–Hsiao estimates are presented here.¹³

The Anderson–Hsiao approach results in the following two-stage approach¹⁴:

\hat{Δ A P P R O V E_{i, t - 1}} = π_{0} + π_{1} A P P R O V E_{i, t - 2} + υ_{i, t},

and

Δ A P P R O V E_{i, t} = β_{1} \hat{Δ A P P R O V E_{i, t - 1}} + β_{2} Δ E C O N_{i, t} + Δ σ_{i, t},

where $\hat{Δ A P P R O V E_{i, t - 1}}$ in Equation 6 consists of the predicted values from the first-stage equation. Ideally, we would want to include additional lags of the outcome, such as $A P P R O V E_{t - 3}$ and $A P P R O V E_{t - 4}$ , or additional lags of the differenced outcome, such as $Δ A P P R O V E_{t - 2}$ and $Δ A P P R O V E_{t - 3}$ , as instrumental variables, but with only three waves of panel data, such instruments are simply not available. Nonetheless, because the GMM estimator has also been shown to have some inefficiencies (Kiviet, 1995) and the results between the GMM estimator and the Anderson–Hsiao estimator are not substantively different, we can still have confidence in the use of a single lag as an instrumental variable.

Equation 6 provides a more reliable estimate of the causal effect of economic perceptions on presidential approval than fixed effects estimation or ordinary least squares (OLS). We can test the reciprocal of this relationship using the same approach with economic perceptions as the outcome variable and presidential approval as the explanatory variable, so that

\hat{Δ E C O N_{i, t - 1}} = π_{2} + π_{3} E C O N_{i, t - 2} + ε_{i, t},

and

Δ E C O N_{i, t} = β_{3} \hat{Δ E C O N_{i, t - 1}} + β_{4} Δ A P P R O V E_{i, t} + Δ ϑ_{i, t},

where $\hat{Δ E C O N_{i, t - 1}}$ consists of the predicted values from the first-stage Equation (9).

Because we are estimating a causal effect that moves in both directions, we also need to allow for the possibility that the error terms from Equations 6 and 8 are correlated. We cannot statistically compare the estimates in the two equations if they are not calculated simultaneously. The seemingly unrelated regression (SUR) framework estimates two or more equations simultaneously with errors correlated across equations, allowing us to reliably compare the direction of the relationship between economic perceptions and presidential approval across multiple panel studies.¹⁵ Figure 3 illustrates the structural equation setup used in the following analysis. By comparing the estimates of $β_{2}$ in Equation 6 and $β_{4}$ in Equation 8 across each of the panel studies, we can test the hypotheses described above. If economic perceptions matter more for presidential approval during bad economic times than during relatively better economic times, then $β_{2}$ (the effect of economic perceptions on presidential approval) should be larger during the 2008-2009-2010 Panel Study than during the previous three panel studies. If presidential approval matters more for the formation of economic perceptions during good economic times than during relatively worse economic times, then $β_{4}$ (the effect of presidential approval on economic perceptions) should be larger during the first three panel studies than during the 2008-2009-2010 Panel Study.

Figure 3.

Structural equation model.

Analysis and Findings

Table 1 shows standardized SUR extensions of the Anderson–Hsiao estimator, with presidential approval modeled as a function of economic perceptions in one equation and economic perceptions modeled as a function of presidential approval in another equation for each of the four panel studies. As described above, the lagged differences of the dependent variable in each equation are instrumented by their second lags. If my hypotheses hold true, then the effect of economic perceptions on presidential approval should be greater during the 2008-2009-2010 Panel Study than during the previous three panel studies, and the effect of presidential approval on economic perceptions should be greater during the first three panel studies than during the 2008-2009-2010 Panel Study.

Table 1.

SUR Estimates of Presidential Approval and Economic Perceptions.

Y = ∆Approval _t	1990-91-92	1992-94-96	2000-02-04	2008-09-10
^{∆Presidential Approval} _t-1	0.15*	0.02	0.10***	3.35***
	(0.07)	(0.03)	(0.02)	(0.20)
∆Economic Perceptiont _t	0.23***	0.30***	0.20***	0.87***
	(0.02)	(0.04)	(0.03)	(0.03)
Constant	−0.85***	0.32***	−0.43***	−0.26**
	(0.04)	(0.04)	(0.03)	(0.07)
R ²	0.01	0.01	0.04	0.63
Y = Perception _t
∆Economic Perception _t-1	0.19**	0.09	0.02	0.23**
	(0.07)	(0.05)	(0.03)	(0.07)
∆Presidential Approval _t	0.38***	0.36***	0.33***	0.62***
	(0.04)	(0.05)	(0.04)	(0.03)
Constant	−0.02	−0.07	0.82***	1.18***
	(0.05)	(0.07)	(0.05)	(0.07)
R ²	.01	.01	.004	.37
Observations	956	541	752	523

Note. Entries are standardized SUR extensions of the Anderson–Hsiao estimator with standard errors in parentheses. In the first equation of each model (Y = ∆Approval_t), ^{∆Presidential Approval}_t-1 is instrumented with ^{∆Presidential Approval}_t-2; in the second equations of each model (Y = Perception_t), ∆Economic Perception_t-1 is instrumented with ∆Economic Perception_t-2. SUR = seemingly unrelated regression.

p < .05. **p < .01. ***p < .001.

The coefficients for ∆Economic Perception_t and ∆Presidential Approval_t are both statistically significant across all four panel studies. In the first three panel studies, which represent relatively stable economic times, the coefficient for ∆Presidential Approval_t is larger in magnitude than the coefficient for ∆Economic Perception_t, indicating a perceptual bias in economic opinions during good economic times. This is not surprising, given the relative lack of negative economic information being sent to the public. In the 1990-1991-1992 Panel Study, a 1-standard deviation difference in presidential approval from one panel wave to the next was positively associated with about a 0.38-standard deviation difference in economic perceptions, whereas a 1-standard deviation difference in economic perceptions was positively associated with only about a 0.23-standard deviation difference in presidential approval. This difference in effect size was slightly smaller in the 1992-1994-1996 Panel Study, in which a 1-standard deviation difference in presidential approval was associated with about a 0.36-standard deviation difference in economic perceptions and a 1-standard deviation difference in economic perceptions was associated with about a 0.30-standard deviation difference in presidential approval. In the 2000-2002-2004 Panel Study, a 1-standard deviation difference in presidential approval was associated with about a 0.33-standard deviation difference in economic perceptions, whereas a one-unit difference in economic perceptions was only associated with about a 0.20-standard deviation difference in presidential approval. Thus, the effect of presidential approval on economic perceptions was stronger than the effect of economic perceptions on approval by 0.15 points, 0.06 points, and 0.13 points in the 1990-1991-1992, 1992-1994-1996, and 2000-2002-2004 Panel Studies, respectively.

The most substantively interesting results are found for the 2008-2009-2010 Panel Study. From the outset of the Great Recession, the effect of economic perceptions on presidential approval increased dramatically compared with the previous three panel studies. A 1-standard deviation increase in economic perceptions was associated with about a 0.87-standard deviation increase in presidential approval—an average difference of almost 0.63 standard deviations more than the previous studies. This suggests that, as expected, the tendency for individuals to evaluate the economy and update their attitudes toward the president grew significantly following the 2008 financial crisis. Quite surprisingly, and contrary to my expectations, the effect of presidential approval on economic perceptions also increased dramatically. A 1-standard deviation increase in presidential approval was associated with about a 0.62-standard deviation increase in economic perceptions. Once again, this difference was substantially higher than the previous panel studies—just over one fourth of a standard deviation higher on average than the previous three panel studies.

One potential explanation for the growth in the effect of presidential approval on economic perceptions during the 2008-2009-2010 Panel Study is that even as the economy worsens, the relationship between economic perceptions and presidential approval might be moderated differently for strong partisans, weak partisans, and independents. It is plausible that presidential approval remains a stable predictor of political and economic attitudes even as the economic context changes and exposure to negative economic information increases,¹⁶ and perhaps becomes an even stronger predictor for individuals with stronger partisan attachments. These findings might be different if we looked at job performance evaluations for less visible political figures, such as Cabinet members, individual members of Congress, or political figures at the local or state level. Moreover, if we consider the notion that the least and most politically aware are the most prone to rely on partisan shortcuts when evaluating issues (Zaller, 1992), then the moderating effect of the economic context for the relationship between presidential approval and economic perceptions might also vary as a function of both partisan strength and political awareness. Nonetheless, these theories fall outside the scope of this study. The most substantively relevant finding in Table 1 is that the impact of economic perceptions on presidential approval increased substantially as the national economy fell into recession during the 2008-2009-2010 Panel Study.

To further illustrate the findings from Table 1, Figure 4 plots the expected standard deviation changes in presidential approval based on specified standard deviation changes in economic perceptions from one panel wave to the next, as well as expected standard deviation changes in economic perceptions based on specified standard deviation changes in presidential approval. The vertical bars in each graph represent the upper and lower bounds of the 95% confidence intervals, and the dashed horizontal line indicates no change in the outcome variable between panel waves. Thus, the values at which the point estimates meet the 0 line indicate no net effect of changes in the X variable on changes in the Y variable, whereas the further away a point estimate falls from the 0 line indicates stronger effects. However, it should be noted that the expected values plotted in Figure 4 each holds the instrumented lagged-differenced outcome ( $\hat{Δ A p p r o v a l_{t - 1}}$ and $\hat{Δ P e r c e p t i o n_{t - 1}}$ ) constant at its mean value, while manipulating only the differenced explanatory variable. As such, the points at which the estimates cross the zero expected change line are less meaningful than the slopes of the expected values. These graphs allow us to visually compare the effects of changes in economic perceptions on changes in presidential approval with the effects of changes in presidential approval on changes in economic perceptions.

Figure 4.

Expected changes in presidential approval and economic perceptions.

Reflective of the findings presented in Table 1, the panels of Figure 4 for the 1990-1991-1992, 1992-1994-1996, and 2000-2002-2004 studies indicate relatively similar relationships between changes in economic perceptions and presidential approval. In each of the three panels, the effect of changes in economic perceptions on presidential approval is somewhat weaker than the effect of presidential approval on economic perceptions. In the 2008-2009-2010 Panel Study, we see a sharp increase in the slope of the expected changes in presidential approval based on specified changes in economic perceptions, as well as the slope of the expected changes in economic perceptions based on specified changes in presidential approval. As hypothesized, the largest increase in effect size is in regard to the effect of economic perceptions on presidential approval during the latter panel study. Overall, the findings presented here lend support to the expectation that the effect of economic perceptions on presidential approval is substantially stronger during bad economic times than during relatively better economic times. Surprisingly, however, these findings do not support the expectation that attitudes toward the president matter more for the formation of economic perceptions during good economic times than during bad economic times.

Conclusion

The findings presented here show a substantial difference in the relationship between economic perceptions and presidential approval during time periods characterized by distinct macroeconomic contexts. In particular, these findings show that the effect of economic perceptions on presidential approval became much stronger as economic conditions worsened during the Great Recession of 2008. Somewhat surprisingly, and contrary to the findings of Chzhen et al. (2013), the effect of presidential approval on economic perceptions also became much stronger during the recession. It should be noted that Chzhen et al.’s analysis was conducted in the context of the U.K. political system and focused on approval of the Labour Party rather than a single political figure. In the British context, because visibility and responsibility are distributed throughout the party rather than focused on a single individual, it might be the case that the Labour Party serves as a weaker source of political attitudes for U.K. voters than the President serves for U.S. voters. If this is the case, then it makes sense that in the U.K. system government approval has a weaker impact on economic perceptions as the economy worsens, whereas in the U.S. context, presidential approval retains its impact on economic perceptions even as the economy worsens. The disparate findings between these two studies could also simply be the result of different measurements of political attitudes—one measured as approval of an entire party and the other measured as approval of the president’s job performance—or simply that institutional differences between the U.K. and U.S. political systems have instilled somewhat different psychological mechanisms in the minds of U.K. and U.S. voters.

A number of other potential explanations exist for the rejection of my hypothesis that the impact of presidential approval on economic perceptions should be greater during good economic times than bad economic times. First, it might be the case that the macroeconomic context moderates the relationship between political and economic attitudes differently depending on the strength and direction of an individual’s partisanship. Being such a highly visible political figure, attitudes toward the president might always be an influential source of political and economic attitudes, particularly for individuals with strong partisan attachments. Second, it might be the case that the ability to effectively manipulate new economic information as it becomes increasingly available during an economic recession, at the expense of identity-confirming political heuristics, might depend on a sufficient level of political sophistication. While both are theoretically interesting, each of these ideas falls outside the scope of the current study. Nonetheless, the puzzle introduced by this finding does not necessarily diminish the value of the other findings. The fact that the effect of economic perceptions on presidential approval varies over time, and in particular across distinct economic contexts, raises important implications for the study of economic voting. As scholars debate whether economic perceptions lead voters to update their political attitudes or whether economic perceptions are biased by preexisting political attitudes, they need to take a closer look at the broader political and economic context in which these relationships exist. The findings presented here show undoubtedly that the nature of this relationship is not as stable as previous research has seemed to suggest. A better understanding of when voters are most likely to use their perceptions of the economy to update their attitudes toward incumbent political figures will shed new light on the circumstances under which candidates and parties are the most likely to be held accountable for economic performance.

Footnotes

Acknowledgements

The author thanks Heather Ondercin, Mike Henderson, and Chuck Smith, as well as the editor and three anonymous reviewers for helpful comments and suggestions.

Authors’ Note

A previous version of this article was presented at the 2014 annual meeting of the Midwest Political Science Association in Chicago, Illinois.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Bradley Dickerson is a PhD Candidate in the Department of Political Science at the University of Mississippi. His research interests include public opinion, political psychology, and political methodology.

References

Anderson

C. J.

(1995). Blaming the government: Citizens and the economy in five European democracies. Armonk, NY: M.E. Sharpe.

Anderson

C. J.

(2007). The end of economic voting? Contingency dilemmas and the limits of democratic accountability. Annual Review of Political Science, 10, 271-296.

Anderson

Hsiao

(1981). Estimation of dynamic panel models with error components. Journal of the American Statistical Association, 76, 589-606.

Angrist

J. D.

Pischke

J.-S.

(2009). Mostly harmless econometrics: An empiricist’s companion. Princeton, NJ: Princeton University Press.

Arcelus

Meltzer

A. H.

(1975). The effect of aggregate economic variables on congressional elections. American Political Science Review, 69, 1232-1239.

Arellano

Bond

(1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies, 58, 277-297.

Arellano

Bover

(1995). Another look at the instrumental variable estimation of error-component models. Journal of Econometrics, 68, 29-51.

Bartels

L. M.

(2000). Partisanship and voting behavior, 1952-1996. American Journal of Political Science, 44, 35-50.

Bartels

L. M.

(2002). Beyond the running tally: Partisan biases in political perceptions. Political Behavior, 24, 117-150.

10.

Baum

C. F.

(2006). An introduction to modern econometrics using Stata. College Station, TX: StataCorp.

11.

Baumeister

R. F.

Bratslavsky

Finkenauer

Vohs

K. D.

(2001). Bad is stronger than good. Review of General Psychology, 5, 323-370.

12.

Berelson

B. R.

Lazarsfeld

P. F.

McPhee

W. N.

(1954). Voting: A study of opinion formation in a presidential election campaign. Chicago, IL: University of Chicago Press.

13.

Bloom

H. S.

Price

H. D.

(1975). Voter response to short-run economic conditions: The asymmetric effect of prosperity and recession. American Political Science Review, 69, 1240-1254.

14.

Blundell

Bond

(1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87, 115-143.

15.

Brody

R. A.

Sniderman

P. M.

(1977). From life space to polling place: The relevance of personal concerns for voting behavior. British Journal of Political Science, 7, 337-360.

16.

Bullock

J. G.

(2011). Elite influence on public opinion in an informed electorate. American Political Science Review, 105, 496-515.

17.

Bullock

J. G.

Gerber

A. S.

Hill

S. J.

Huber

G. A.

(2013). Partisan bias in factual beliefs about politics (No. w19080). Cambridge, MA: National Bureau of Economic Research.

18.

Campbell

Converse

P. E.

Miller

W. E.

Stokes

D. E.

(1960). The American voter. New York, NY: Wiley.

19.

Chaiken

Liberman

Eagly

A. H.

(1989). Heuristic and systematic information processing within and beyond the persuasion context. In Uleman

J. S.

Bargh

J. A.

(Eds.), Unintended thought (pp. 212-252). New York, NY: Guilford Press.

20.

Chzhen

Evans

Pickup

(2013). When do economic perceptions matter for party approval? Examining the endogeneity of economic perceptions before and during the economic downturn. Political Behavior, 36, 291-313.

21.

De Boef

S. L.

Kellstedt

P. M

. (2004). The political (and economic) origins of consumer confidence. American Journal of Political Science, 48, 633-649.

22.

Duch

R. M.

Stevenson

R. T.

(2008). The economic vote: How political and economic institutions condition election results. New York, NY: Cambridge University Press.

23.

Enns

P. K.

Kellstedt

P. M.

McAvoy

G. E.

(2012). The consequences of partisanship in economic perceptions. Public Opinion Quarterly, 76, 287-310.

24.

Enns

P. K.

McAvoy

G. E.

(2012). The role of partisanship in aggregate opinion. Political Behavior, 34, 627-651.

25.

Erikson

R. S.

MacKuen

M. B.

Stimson

J. A.

(2002). The macro polity. Cambridge, UK: Cambridge University Press.

26.

Evans

Anderson

(2006). The political conditioning of economic perceptions. Journal of Politics, 68, 194-207.

27.

Evans

Pickup

(2010). Reversing the causal arrow: The political conditioning of economic perceptions in the 2000-2004 U.S. presidential election cycle. Journal of Politics, 72, 1236-1251.

28.

Fair

R. C.

(1978). The effects of economic events on votes for president. Review of Economics and Statistics, 60, 159-173.

29.

Feldman

(1982). Economic self-interest and political behavior. American Journal of Political Science, 26, 446-466.

30.

Finkel

S. E.

(1995). Causal analysis with panel data. Thousand Oaks, CA: Sage University Press.

31.

Fiorina

M. P.

(1978). Economic retrospective voting in American national elections: A micro-analysis. American Journal of Political Science, 22, 426-443.

32.

Fiorina

M. P.

(1981). Retrospective voting in American national elections. New Haven, CT: Yale University Press.

33.

Gerber

A. S.

Huber

G. A.

(2010). Partisanship, political control, and economic assessments. American Journal of Political Science, 54, 153-173.

34.

Gerber

A. S.

Huber

G. A.

Washington

(2010). Party affiliation, partisanship, and political beliefs: A field experiment. American Political Science Review, 104, 720-744.

35.

Goidel

R. K.

Langley

R. E.

(1995). Media coverage of the economy and aggregate economic evaluations: Uncovering evidence of indirect media effects. Political Research Quarterly, 48, 313-328.

36.

Gomez

B. T.

Wilson

J. M.

(2001). Political sophistication and economic voting in the American electorate: A theory of heterogeneous attribution. American Journal of Political Science, 45, 899-914.

37.

Goodman

Kramer

G. H.

(1975). Comments on Arcelus and Meltzer: The effect of aggregate economic conditions on congressional elections. American Political Science Review, 69, 1255-1265.

38.

Goren

Federico

C. M.

Kittilson

M. C.

(2009). Source cues, partisan identities, and political value expression. American Journal of Political Science, 53, 805-820.

39.

Green

Palmquist

Schickler

(2002). Partisan hearts and minds: Political parties and the social identities of voters. New Haven, CT: Yale University Press.

40.

Harrington

D. E.

(1989). Economic news on television: The determinants of coverage. Public Opinion Quarterly, 53, 17-40.

41.

Hellwig

T. T.

Ringsmuth

E. M.

Freeman

J. R.

(2008). The American public and the room to maneuver: Responsibility attributions and policy efficacy in an era of globalization. International Studies Quarterly, 52, 855-880.

42.

Hetherington

M. J.

(1996). The media’s role in forming voters’ national economic evaluations in 1992. American Journal of Political Science, 40, 372-395.

43.

Hibbs

D. A.

(1977). Political parties and macroeconomic policy. American Political Science Review, 71, 1467-1487.

44.

Hibbs

D. A.

(1979). The mass public and macroeconomic policy: The dynamics of public opinion toward unemployment and inflation. American Journal of Political Science, 23, 705-731.

45.

Hsiao

(2003). Analysis of panel data. Cambridge, UK: Cambridge University Press.

46.

Huber

G. A.

Hill

S. J.

Lenz

G. S.

(2012). Sources of bias in retrospective decision making: Experimental evidence on voters’ limitations in controlling incumbents. American Political Science Review, 106, 720-741.

47.

Jacobson

G. C.

(2010). Perception, memory, and partisan polarization on the Iraq war. Political Science Quarterly, 125, 31-56.

48.

Jerit

Barabas

(2012). Partisan perceptual bias and the information environment. Journal of Politics, 74, 672-684.

49.

Johnston

(2006). Party identification: Unmoved mover or sum of preferences? Annual Review of Political Science, 9, 329-351.

50.

Johnston

Sarker

Jones

Bolster

Propper

Burgess

(2005). Egocentric economic voting and changes in party choice: Great Britain, 1992-2001. Journal of Elections, Public Opinion and Parties, 15, 129-144.

51.

Key

(1966). The responsible electorate. Cambridge, MA: Harvard University Press.

52.

Kiewiet

D. R.

(1983). Macroeconomics and micropolitics: The electoral effects of economic issues. Chicago, IL: University of Chicago Press.

53.

Kim

S.-Y.

Taber

C. S.

Lodge

(2010). A computational model of the citizen as motivated reasoner: Modeling the dynamics of the 2000 presidential election. Political Behavior, 32, 1-28.

54.

Kinder

D. R.

Kiewiet

D. R.

(1979). Economic discontent and political behavior: The role of personal grievances and collective economic judgements in congressional voting. American Journal of Political Science, 23, 495-527.

55.

Kinder

D. R.

Kiewiet

D. R.

(1981). Sociotropic politics: The American case. British Journal of Political Science, 11, 129-161.

56.

King

Tomz

Wittenberg

(2000). Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science, 44, 347-361.

57.

Kiviet

J. F.

(1995). On bias, inconsistency, and efficiency of various estimators in dynamic panel data models. Journal of Econometrics, 68, 53-78.

58.

Klorman

(1978). Trend in personal finances and the vote. Public Opinion Quarterly, 42, 31-48.

59.

Kramer

G. H.

(1971). Short-term fluctuations in U.S. voting behavior, 1896-1964. American Political Science Review, 64, 131-143.

60.

Kramer

G. H.

(1983). The ecological fallacy revisited: Aggregate- versus individual-level findings on economics and elections, and sociotropic voting. American Political Science Review, 77, 92-111.

61.

Kunda

(1990). The case for motivated reasoning. Psychological Bulletin, 108, 480-498.

62.

Ladner

Wlezien

(2007). Partisan preferences, electoral prospects, and economic expectations. Comparative Political Studies, 40, 571-596.

63.

Lau

R. R.

Sears

D. O.

Jessor

(1990). Fact or artifact revisited: Survey instrument effects and pocketbook politics. Political Behavior, 12, 217-242.

64.

Lavine

H. G.

Johnston

C. D.

Steenbergen

M. R.

(2012). The ambivalent partisan: How critical loyalty promotes democracy. New York, NY: Oxford University Press.

65.

Lebo

M. J.

Cassino

(2007). The aggregated consequences of motivated reasoning and the dynamics of partisan presidential approval. Political Psychology, 28, 719-746.

66.

Lewis-Beck

M. S.

(1985). Pocketbook voting in U.S. national election studies: Fact or artifact? American Journal of Political Science, 29, 348-356.

67.

Lewis-Beck

M. S.

(1988). Economics and elections: The major Western democracies. Ann Arbor: University of Michigan Press.

68.

Lewis-Beck

M. S.

Jacoby

W. G.

Norpoth

Weisberg

H. F.

(2008). The American voter revisited. Ann Arbor: University of Michigan Press.

69.

Lewis-Beck

M. S.

Stegmaier

(2000). Economic determinants of electoral outcomes. Annual Review of Political Science, 3, 183-219.

70.

Lewis-Beck

M. S.

Stegmaier

(2007). Economic models of voting. In Dalton

R. J.

Klingemann

H.-D.

(Eds.), The Oxford handbook of political behavior (pp. 518-537). Oxford, UK: Oxford University Press.

71.

Lodge

Taber

C. S.

(2000). Three steps toward a theory of motivated political reasoning. In Lupia

McCubbins

M. D.

Popkin

S. L.

(Eds.), Elements of reason, cognition, choice, and the bounds of rationality (pp. 183-213). Cambridge, UK: Cambridge University Press.

72.

Lord

C. G.

Ross

Lepper

M. R.

(1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of Personality and Social Psychology, 37, 2098-2109.

73.

MacKuen

M. B.

Erikson

R. S.

Stimson

J. A.

(1992). Peasants or bankers? The American electorate and the U.S. economy. American Political Science Review, 86, 597-611.

74.

Markus

(1992). The impact of personal and national economic conditions on presidential voting, 1956-1988. American Journal of Political Science, 36, 829-834.

75.

Nadeau

Lewis-Beck

M. S.

(2001). National economic voting in U.S. presidential elections. Journal of Politics, 63, 159-181.

76.

Nickell

(1981). Biases in dynamic models with fixed effects. Econometrica, 49, 1417-1426.

77.

Nir

(2011). Motivated reasoning and public opinion perception. Public Opinion Quarterly, 75, 504-532.

78.

Nordhaus

(1975). The political business cycle. Review of Economic Studies, 42, 169-190.

79.

Parker-Stephen

(2013). Tides of disagreement: How reality facilitates (or inhibits) partisan public opinion. Journal of Politics, 75, 1077-1088.

80.

Payne

J. W.

Bettman

J. R.

Johnson

E. J.

(1993). The adaptive decision maker. New York, NY: Cambridge University Press.

81.

Rahn

W. M.

(1993). The role of partisan stereotypes in information processing about political candidates. American Journal of Political Science, 37, 472-796.

82.

Redlawsk

D. P.

Civettini

A. J.

Emmerson

K. M.

(2010). The affective tipping point: Do motivated reasoners ever “get it”? Political Psychology, 31, 563-593.

83.

Rousseau

D. M.

Snehal

T. A.

(1999). What’s a good reason to change? Motivated reasoning and social accounts in promoting organizational change. Journal of Applied Psychology, 84, 514-528.

84.

Sanders

Gavin

(2004). Television news, economic perceptions, and political preferences in Britain, 1997-2001. Journal of Politics, 66, 1245-1266.

85.

Schlozman

K. L.

Verba

(1979). Injury to insult: Unemployment, class, and political response. Cambridge, MA: Harvard University Press.

86.

Sears

D. O.

Lau

R. R.

(1983). Inducing apparently self-interested political preferences. American Journal of Political Science, 27, 223-252.

87.

Sigelman

C. K.

Bullock

(1991). Reconsidering pocketbook voting: An experimental approach. Political Behavior, 13, 129-149.

88.

Sniderman

P. M.

Brody

R. A.

(1977). Coping: The ethic of self-reliance. American Journal of Political Science, 21, 501-521.

89.

Soroka

S. N.

(2006). Good news and bad news: Asymmetric responses to economic information. Journal of Politics, 68, 372-385.

90.

Stigler

G. J.

(1973). Micropolitics and macroeconomics: General economic conditions and national elections. American Economic Review, 63, 160-167.

91.

Taber

C. S.

Cann

Kucsova

(2009). The motivated processing of political arguments. Political Behavior, 31, 137-155.

92.

Taber

C. S.

Lodge

(2006). Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science, 50, 755-769.

93.

Taylor

S. E.

(1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, 110, 67-85.

94.

Tufte

E. R.

(1975). Determinants of the outcomes of midterm congressional elections. The American Political Science Review, 69, 812-826.

95.

Tufte

E. R.

(1978). Political control of the economy. Princeton, NJ: Princeton University Press.

96.

Tverdova

Y. V.

(2012). The formation of economic perceptions in post-communist countries of east central Europe. Political Behavior, 34, 137-158.

97.

van der Brug

van der Eijk

Franklin

. (2007). The economy and the vote: Economic conditions and elections in fifteen countries. New York, NY: Cambridge University Press.

98.

Weinschenk

A. C.

(2010). Revisiting the political theory of party identification. Political Behavior, 32, 473-494.

99.

Westen

Blagov

P. S.

Harenski

Kilts

Hamann

(2006). Neural bases of motivated reasoning: An fMRI study of emotional constraints on partisan political judgment in the 2004 U.S. presidential election. Journal of Cognitive Neuroscience, 18, 1947-1958.

100.

Wlezien

Franklin

Twiggs

(1997). Economic perceptions and vote choice: Disentangling the endogeneity. Political Behavior, 19, 7-17.

101.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data. Cambridge, MA: MIT Press.

102.

Wooldridge

J. M.

(2006). Introductory econometrics: A modern approach. Mason, OH: South-Western Cengage Learning.

103.

Wright

J. R.

(2012). Unemployment and the democratic electoral advantage. American Political Science Review, 106, 685-702.

104.

Zaller

(1992). The nature and origins of mass opinion. Cambridge, UK: Cambridge University Press.

105.

Zellner

(1962). An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association, 57, 348-368.