Relative Unemployment,Political Information,and the Job Approval Ratings of State Governors and Legislatures

Abstract

Research finds that voters benchmark the state’s unemployment level to the nation’s when holding state policy makers accountable. Yet benchmarking requires some voter knowledge if the standard is to be applied correctly as an accountability rule. This article leverages the fact that voters have more knowledge about their state governors than legislatures assess how much knowledge is necessary for holding these policy makers accountable. Using pooled Cooperative Congressional Election Study data from 2006 to 2016, results find that knowledge has stronger mediating effects for the state legislature than governor. Furthermore, despite the low knowledge levels among voters about the state legislature, collectively there appears to be enough knowledge to hold that policymaking body accountable. The conclusion offers directions for future research.

Keywords

job approval governors executive politics state legislatures economic policy public policy issue preferences public opinion

What are the effects of federalism on voters’ ability to hold policy makers accountable? By obscuring clarity of responsibility and by increasing the costs of acquiring information, voters may have a harder time holding policy makers accountable in federal than in more centralized systems (Anderson 2006; Arceneaux 2005, 2006; Powell 2000; Powell and Whitten 1993). As federalism is a defining characteristics of the U.S. political system, there has been much attention to the implications of federalism on politics and policy making.

From a functional accountability standpoint, voters should hold state policy makers accountable for issues and concerns within the jurisdiction of the policy makers (Carsey and Wright 1998). In a federal system this means that state policy makers should be held accountable for state but not national unemployment. Consistent with the functional accountability perspective, state unemployment has dominated research on accountability in the states (Ferguson 2014; Newman and Johnson 2012). But debate exists over whether voters hold state policy makers accountable for state unemployment. (The research is discussed below.)

One possible reason for this debate is the high cost for voters in acquiring and using relevant information. Such information costs may make it hard for voters to hold state policy makers accountable for, say, state unemployment. Benchmarking has been offered as a theory of voters decision making in burdensome informational contexts, like in federal systems (Aytaç 2018; Cohen and King 2004; Kayser and Peress 2012; Wolfers 2002). Voters benchmark when they compare some performance across governmental levels. For unemployment, voters would compare the state and national unemployment levels, termed by Cohen and King (2004) relative unemployment. The voters’ relative unemployment decision rule would be to approve the incumbent when state unemployment is lower than national unemployment, but disapprove when state unemployment is higher than national unemployment. Several studies have found support for relative unemployment effects, including for gubernatorial approval (Cohen and King 2004), gubernatorial elections (Wolfers 2002), and big city mayoral elections (Hopkins and Pettingill 2017), but not for New York mayoral approval (Arnold and Carnes 2012).

Still there are some informational needs and costs for voters to apply relative unemployment reasoning correctly. They need to know something about the relative unemployment rates in their state and the nation and something about the policy responsibilities and policy achievements of their state policy makers. Voters vary in the amount of information and knowledge they possess, which leads to the knowledge mediation hypothesis, that voters with more knowledge may be better able to apply the relative unemployment standard correctly than those with less knowledge. This article tests this hypothesis using data from pooled Cooperative Congressional Election Studies (CCES) from 2006 to 2016, which has questions on gubernatorial and state legislative approval and a set of political knowledge questions, which are merged with data on state and national unemployment.

Past studies on the relative unemployment hypothesis mostly have used aggregate data, but individual-level data are necessary for testing the knowledge mediation hypothesis. Furthermore, this article compares application of relative unemployment reasoning for both the governor and state legislature. This comparison is useful, because information about the governor is fairly widespread, but is scant for the state legislature, as detailed below. The low level of information about the state legislature allows us to address how much information in the electorate is necessary to hold policy makers accountable. Rarely do studies compare voter assessments of both the governor and state legislature. Kelleher and Wolak’s (2007) is a rare study that compares public confidence across state institutions. They find differences and similarities in the basis for public confidence in the state governor and legislature, but they find that state unemployment has similar effects for confidence across the two bodies.

This research is important for several reasons. First, federalism in an important characteristic of American democracy but it may undermine voter accountability of state policy makers. Second, state governments have substantial policy making responsibility and authority. Through policy making, state officials can affect the lives of their citizens. Voters also think those governments are consequential and powerful. For these reasons, it is important to understand whether and under what conditions voters hold state policy makers accountable.

The next section discusses the importance of unemployment for voters and policy makers in the states, followed by a review of the research on approval of the governor and state legislature. Then, benchmarking theory and the relative unemployment hypothesis are discussed. Following sections present the data and empirical results. Findings indicate that voters apply relative unemployment reasoning for both institutions, and that knowledge mediation effects are much stronger for the state legislature than the governor. Knowledge does not mediate relative unemployment effects for the governor presumably because of voters’ widespread knowledge of the state executive. The conclusion puts the results into perspective and suggests directions for future research.

Voters, State Policy Makers, and Jobs

State unemployment is an important issue for policy makers and voters. The importance of the issue is a necessary first step in making a case that voters would hold state policy makers accountable for the state’s unemployment. State unemployment and jobs are important because governors have prioritized the issue (Bernick 2016; Brace 1993; Grady 1989; Taylor 2012). They travel extensively nationally and globally to bring jobs to their states. In their State of the State Addresses (SOSA), they stress job creation more than most other issues (Coffey 2005; DiLeo 2001; Herzik 1991; Weinberg 2010). Taylor’s (2012) analysis of SOSA finds that economic policy is always on governors’ legislative agendas at high levels (p. 274).

In addition, voters think state governments can affect the state’s economy and their personal financial circumstances (Jacobs 2017; Kincaid and Cole 2010). A 1982 CBS News/New York Times poll found 69% saying the governor was responsible for the state’s economy, and a 2017 Morning Consult/Bloomberg poll found 45% saying that state or local business laws and regulations affected their personal finances, compared to 40% for the national government.¹

Voters also rate state unemployment (and other state economic issues) among the top issues for the state to address. A Kaiser Health Tracking poll of March 2016 found that jobs/unemployment/wages was the most cited concern at 18%. And 7% cited the economy more generally, while another 10% split across infrastructure, taxes, and “Government spending/Budget/Deficit.” Combined, about 35% of voters cited something relevant to the economy as the most important problem for their state.² The National Center for State Courts poll in March/April 2006 found similar results, with 12% citing jobs, 4% the economy, 3% infrastructure, 3% the cost of living, 9% taxes, and 4% state budgets; 35% cited an aspect of the economy. Only education received a large number of mentions, with 18%.³ Thus, we should expect voters to at least try to hold policy makers accountable for state unemployment.

Research on State Elections and Approval

In part because of the above reasons, state unemployment has dominated the research on accountability in the states (Ferguson 2014; Newman and Johnson 2012). But factors besides state unemployment have also been found to affect voters’ assessment of their state policy makers. King and Cohen (2005) offer a useful framework for organizing the research on approval and elections in the states. Their framework has two dimensions: (1) state versus national and (2) economic versus political factors (p. 230). Presidential approval is an obvious national political factor and national unemployment a national economic one. State political factors include the party of the governor and state legislature, with state unemployment the most investigated state-level economic factor. When employing individual-level data, it also may be useful to include voter political predispositions and attitudes, such as party identification. Recent studies also have expanded the factors that can affect approval and elections to issues besides economics and unemployment, such as immigration (Newman and Johnson 2012) and Medicaid (Fording and Patton 2019).

There is no consensus, however, on the effects of state unemployment on gubernatorial or state legislative elections or approval ratings. For governors, some studies find higher state unemployment weakens incumbent electoral performance (Atkeson and Partin 1995; Carsey and Wright 1998; Partin 1995; Svoboda 1995; Wolfers 2002), but others detect either no or mixed effects (Besley and Case 1995; Brown 2010; Chubb 1988; Ebeid and Rodden 2006; Hummel and Rothschild 2014; Kenney and Rice 1983; Krause and Melusky 2014; Leyden and Borrelli 1995; Peltzman 1987; Rodden and Wibbels 2011; Stein 1990; Wright 2012). Similarly several studies find higher state unemployment harms gubernatorial approval (Cohen 2018; Cohen and King 2004; Crew et al. 2002; Hansen 1999a, 1999b; Howell and Vanderleeuw 1990; Kelleher and Wolak 2007; King and Cohen 2005; Niemi, Stanley, and Vogel 1995; Orth 2001), but neither Crew and Weiher (1996) nor MacDonald and Sigelman (1999) find such an effect on approval.

There too are mixed findings with regard to the state legislature. Several studies find no effect of the state economy, including unemployment, on state legislative elections (Chubb 1988; Lowry, Alt, and Ferree 1998; S. Rogers 2017; Stein 1990). But Richardson, Konisky, and Milyo (2012); Richardson and Milyo (2016); Kelleher and Wolak (2007); and Langehennig, Zamadics, and Wolak (2019) report higher state unemployment dampens approval of and confidence in the state legislature.

Why the divergence in findings on whether voters hold state policy makers accountable for state unemployment? Studies use different databases, time frames, dependent variables, estimation strategies, and model specifications. A second reason is that many studies have not specified fully the theoretical linkages from state unemployment to the approval ratings or the vote other than the simple proposition that when unemployment rises, voters will withdraw their support. Often missing in these accounts are the high informational hurdles for voters in federal systems, like the U.S. states.

For instance, the state and national economies are to some degree integrated. This has implications for voters’ ability to partition blame(reward) for the state’s unemployment level. Economic conditions and policies from one state may spillover across state lines, and national economic policies may affect a state’s unemployment rate. Thus, it may thus be difficult for voters to segment how much of a state’s unemployment level is due to state policies or these other factors. Compounding this situation, there may not be enough news about state government for voters to learn about state economic policies or conditions. News coverage of state government activities is sparse and state capitol news bureau staffs have been shrinking (Althaus, Cizmar, and Gimpel 2009; Delli Carpini, Keeter, and Kennamer 1994; Hayes and Lawless 2015; Layton and Dorroh 2002; Lyons, Jaeger, and Wolak 2013; Shaker 2009; Vining et al. 2010). For this and other reasons, the audience for local and state news has shrunk (Hopkins 2018). Voter knowledge about state politics appears quite limited in general (Delli Carpini and Keeter 1996; Lyons, Jaeger, and Wolak 2013).

But other research suggests that when it comes to state unemployment, voters possess some degree of accurate information. While Niemi, Bremer, and Heel (1999) do not assess the accuracy of voter assessments of state and national unemployment, they find that “the condition of the state economy has a big effect on how it is perceived” (p. 185). J. Rogers (2016) finds that metropolitan statistical area (MSA) unemployment rates affect voter perceptions of MSA economic conditions. And Ansolabehere, Meredith, and Snowberg (2012) find that state unemployment rates affect voter perceptions of the state’s economy. A November 2010 PIPA/Knowledge Networks poll asks respondents whether the state unemployment rate is higher, lower, or about the same as the nation’s.⁴ There is a significant correlation between the difference in the national and state unemployment levels for October 2010 and voter perceptions (r = .52, p < .001). Voters may not be so lacking in knowledge about state and national unemployment.

With regard to political knowledge, the CCES asks respondents to identify the party of the governor and both legislative chambers. Knowledge of the governor’s party is widespread, at 73.6%, but only 45.7% and 42.8% could identify the party of the state House or state Senate correctly. The limited knowledge of the state legislature may impede voters from holding that institution responsible for state unemployment.

Benchmarking Theory and Accountability in the States

When information is costly for voters to obtain, as it appears to be in the states, voters may have a hard time holding policy makers accountable. Benchmarking theory suggests a method for voters to employ when information is costly—voters compare performance on some attribute between two governments, for instance between their state and the nation. The theoretical appeal of benchmarking is that it provides voters with a relatively easy to use accountability decision rule compared to other standards, such as trying to partition the relative effects of state and national policies on state unemployment. As Kayser and Peress (2012) explain, “No economic figure is innately high or low; what passes for booming growth in one period or place might be considered sluggish in another. To assess economic performance, voters necessarily must compare an outcome to others . . .” (p. 662).

The relative unemployment hypothesis, one application of benchmarking, suggests that voters will reward state policy makers by approving when state unemployment is lower than that of the nation, but will punish by disapproving when state unemployment is higher than for the nation. Studies find support for the relative unemployment hypothesis for gubernatorial approval (Cohen and King 2004), gubernatorial elections (Wolfers 2002), and mayoral elections (Hopkins and Pettingill 2017), but not New York mayoral approval (Arnold and Carnes 2012).

Still relative unemployment and benchmarking require some information, and there may be times when voter information levels may be too low to apply such reasoning correctly. For instance, voters should have a relatively accurate sense of the comparative unemployment levels of the state and the nation and also know something about state policy makers. As noted above, less than half of voters could identify the party controlling either the state House or Senate. In a two-party system, we would expect 50% to get this right by guessing.

This leads to the knowledge mediation hypothesis, that voters with more knowledge should be better able to apply relative unemployment reasoning correctly than voters with less knowledge. Ideally, we want a study that includes questions tapping voters’ both economic and political knowledge. Unfortunately, the CCES used does not have economic knowledge questions, and the PIPA/Knowledge Networks poll does have questions on political knowledge or gubernatorial or state legislative job approval.

The CCES, however, has questions on knowledge of the party affiliation of the governor, the state House, and state Senate. Knowledge of the governor’s party is widespread and accurate, but smaller fractions of voters could identify correctly the party of either legislative chamber. CCES does not have other relevant items, such as knowledge about the policymaking responsibility of the state executive and legislature or knowledge of state policies to address unemployment. It is unlikely that many respondents who do not know the party of state policy makers would know about policymaking responsibilities or state unemployment policies.

Data and Variables

To test the above hypotheses, I pool CCES polls from 2006 through 2016, merged with data on state and national unemployment. The dependent variables for this study are job approval ratings for the governor and state legislature, measured on a 5-point scale from strongly approve to strongly disapprove. Don’t know/not sure responses are retained and coded in the middle category.⁵ Table 1 lists the definitions of variables used in the analysis.

Table 1.

Variables Used in the Analysis and Their Definitions.

Variable name	Variable values
Gubernatorial approval	4 = strongly approve, 3 = approve, 2 = don’t know or not sure, 1 = disapprove, 0 = strongly disapprove
State legislative approval	4 = strongly approve, 3 = approve, 2 = don’t know or not sure, 1 = disapprove, 0 = strongly disapprove
Knowledge	Number correct, 0–3, Know which party controls the Governorship, State Senate, State House
Relative unemployment	State unemployment – National unemployment (in %)
National unemployment	National unemployment (in %)
Legislature–respondent same ideology	5 = very liberal (conservative) consistent with party controlling the state legislature, 4 = liberal (conservative) consistent, 3 = moderate, 2 = liberal (conservative) consistent, 1 = very liberal (conservative) inconsistent
Legislature–respondent same party	7 = strong same-party identification as party controlling the state legislature, 6 = not strong same party, 5 = lean same party, 4 = independent, 3 = lean opposite party, 2 = not strong opposite party, 1 = strong opposite-party identification as party controlling the state legislature
Governor–respondent same ideology	5 = very liberal (conservative) consistent with party of governor, 4 = liberal (conservative) consistent, 3 = moderate, 2 = liberal (conservative) consistent, 1 = very liberal (conservative) inconsistent party
Governor–respondent same party	7 = strong same-party identification as party of governor, 6 = not strong same party, 5 = lean same party, 4 = independent, 3 = lean opposite party, 2 = not strong opposite party, 1 = strong opposite-party identification as party of governor
Presidential approval	4 = strongly approve, 3 = approve, 2 = don’t know or not sure, 1 = disapprove, 0 = strongly disapprove
President–legislature same party	1 = president–both chambers of the legislature same party, 0 = 1 or both chambers different party from the president
Governor–legislature same party	1 = governor–both chambers of the legislature same party, 0 = 1 or both chambers different party from the governor
President–governor same party	1 = president and governor same party, 0 = president and governor different party

The primary independent variable, relative unemployment, is the difference in the average annual unemployment levels of the state and nation during the year of the surveys (state unemployment % – national unemployment %). The relative unemployment hypothesis predicts a negative sign—when state unemployment is higher than the nation’s, voters will disapprove of the governor and state legislature, but when state unemployment is lower, voters approve. Using actual unemployment, rather than voter perceptions of the economy, avoids endogeneity concerns between job approval and economic perceptions (Evans and Pickup 2010; Gerber and Huber 2009).

Control Variables

The analysis also includes several control variables. The first is national unemployment. Why would voters blame or reward the governor for national unemployment? Especially when the national unemployment is high, voters may feel a need to vent their discontent, and governors are a handy target because they are well-known and highly visible public figures (Cohen and King 2004). National unemployment has been used in several studies of gubernatorial approval with mixed effects, sometimes finding that national unemployment depresses gubernatorial approval (Cohen and King 2004; Jacobson 2006; King and Cohen 2005), but other studies detecting no effect of national unemployment on gubernatorial approval (Crew and Weiher 1996; Hansen 1999b; Howell and Vanderleeuw 1990; Orth 2001).

A second set of controls concerns political predispositions. Political predispositions, especially partisanship and ideological identification, have been found to be strong predictors of voter evaluations and support for political leaders. Survey research finds voter partisanship and ideological identification affect approval of the governor (Cohen 1983; Hamman 2004; Howell and Vanderleeuw 1990; Orth 2001) and state legislature (Langehennig, Zamadics, and Wolak 2019; Richardson, Konisky, and Milyo 2012; Richardson and Milyo 2016).

First are variables for whether the respondent shares the party and ideological leanings of the governor and legislature. These variables begin with the 7-point party identification (strong Democrat to strong Republican) and 5-point ideological identification (strong liberal to strong conservative) questions. These variables are aligned so that respondent and governor/legislature party (ideology) point in the same directions. Shared partisanship is scored “7” when the respondent is a strong identifier of the governor’s party and “1” when a strong identifier of the opposition party. Shared ideology assumes Democrats lean in a liberal direction and Republicans in a conservative direction. Shared ideology is coded “5” when the respondent is a strong ideological identifier consistent with the governor (legislature) and “1” when the respondent is a strong ideological identifier opposite to that of the governor (legislature), for instance, when the governor is a Democrat and the respondent is a strong conservative. When there is split party control of the state legislature, the shared party and ideology are scored at the variables’ midpoints (4 for partisanship and 3 for ideology). For both institutions, positive signs are expected for both shared partisanship and ideology with approval.

There is also a control for presidential approval. Research finds that presidential approval affects evaluations of governors but the direction of impact depends upon the whether the governor and president are of the same party (Cohen and King 2004; Crew et al. 2002; Crew and Weiher 1996; King and Cohen 2005; Orth 2001).⁶ Voters may view office holds as members of the president’s team or the opposition team. By using dis(approval) of the president to evaluate the governor or state legislature, voters may be sending a signal to the president of their (dis)pleasure with the president. Hence, presidential approval will have differing effects depending upon whether the governor and state legislature are of the same party as the president: presidential approval will positively affect governors and legislatures of the president’s party, but negatively affect opposition governors and legislatures. To test this hypothesis, I employ an interaction between a dummy variable for whether the president and governor (state legislature) are of the same party (=1) and the respondent’s approval of the president. Split party control of state legislature is coded as opposition control.

Finally, research shows that gubernatorial approval affects state legislative approval in parallel fashion to the effects of presidential approval (Richardson, Konisky, and Milyo 2012; Richardson and Milyo 2016). Thus, voters may view the state legislature as part of the governor’s team. Gubernatorial approval should then have a positive effect when the state legislature and governor are of the same party, but a negative effect when they are of different parties. This idea is measured and tested similarly to presidential approval effects with an interaction between a dummy variable for whether the governor and legislature are of the same party (=1) and gubernatorial approval. Split party control of state legislature is coded as opposition control.

Results

Ordered logit is used as approval is measured on a 0 to 4 scale. The analysis clusters on state-year (a dummy variable for each combination of state and year, for example, Delaware-2010).⁷ The online appendix presents results of several alternative estimations, including using regression, clustering on either state or year but not both simultaneously, and with a multilevel model. No matter the estimation type, substantive results remain the same.

Table 2 presents results of estimations that only include the unemployment variables, relative unemployment, state unemployment in levels, and national unemployment in levels. Each variable, when entered singly, shows the proper negative sign on approval for both the governor and state legislature. Although ordered logit coefficients do not have an intuitive interpretation, the relative unemployment coefficient is larger than either state or national unemployment in levels. Two additional models enter (1) both relative employment and national unemployment or (2) state unemployment in levels with national unemployment at the same time. One concern with entering both national and relative unemployment is that they may be highly collinear, which would affect the interpretation of their coefficients. But this is not the case, as relative and national unemployment are not correlated (r = .03). In contrast, state and national unemployment (both in levels) are strongly correlated at r = .82.

Table 2.

Impact of Unemployment Variables on Gubernatorial and State Legislative Approval, CCES 2006–2016, Ordered Logit (Clustered on State-Year).

Variables	Gubernatorial approval
Relative unemployment	−13.23*** (2.33)			−12.93*** (1.90)
State unemployment (levels)		−8.96*** (1.27)			−12.93*** (1.90)
National unemployment (levels)			−7.29*** (1.74)	−7.03*** (1.50)	5.90* (2.32)
/cut1	−1.12*** (0.03)	−1.73*** (0.09)	−1.61*** (0.12)	−1.60*** (0.11)	−1.60*** (0.11)
/cut2	−0.23*** (0.03)	−0.85*** (0.09)	−0.73*** (0.12)	−0.71*** (0.11)	−0.71*** (0.11)
/cut3	0.19*** (0.03)	−0.43*** (0.09)	−0.31** (0.12)	−0.29** (0.11)	−0.29** (0.11)
/cut4	1.82*** (0.04)	1.21*** (0.09)	1.32*** (0.12)	1.35*** (0.11)	1.35*** (0.11)
Observations	342,110	342,110	342,110	342,110	342,110
df	1	1	1	2	2
Log likelihood	−524,834	−524,102	−525,403	−523,896	−523,896
Variables	State legislative approval
Relative unemployment	−17.16*** (3.51)			−16.61*** (2.88)
State unemployment (levels)		−11.67*** (1.84)			−16.61*** (2.88)
National unemployment (levels)			−9.53*** (2.36)	−9.02*** (1.89)	7.58* (3.11)
/cut1	−1.33*** (0.04)	−2.15*** (0.11)	−2.00*** (0.14)	−1.97*** (0.11)	−1.97*** (0.11)
/cut2	−0.16*** (0.03)	−0.98*** (0.11)	−0.83*** (0.15)	−0.79*** (0.12)	−0.79*** (0.12)
/cut3	0.62*** (0.03)	−0.19 (0.12)	−0.05 (0.15)	−0.00 (0.12)	−0.00 (0.12)
/cut4	2.94*** (0.05)	2.13*** (0.13)	2.26*** (0.16)	2.32*** (0.13)	2.32*** (0.13)
Observations	291,119	291,119	291,119	291,119	291,119
df	1	1	1	2	2
Log likelihood	−435,430	−434,520	−436,340	−434,245	−434,245

Note. Robust standard errors in parentheses. CCES = Cooperative Congressional Election Studies.

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

When national unemployment is entered along with relative unemployment, both retain the correct negative sign while the magnitude of the coefficients does not alter much. In the estimation with both national and state unemployment in levels, state unemployment is signed negative but national unemployment now has a positive (and wrong) sign for both the governor and state legislature. It makes little sense that voters would reward governors and state legislatures when national unemployment is high.

Do the above findings of the effects of relative unemployment persist with controls for political predispositions and political contexts? Table 3 presents the results. The relative unemployment hypothesis receives support for both institutions: when state unemployment is higher (lower) than the nation’s, both governors and state legislatures receive lower (higher) marks from voters.

Table 3.

Impact of Relative Unemployment on Gubernatorial and State Legislative Approval, CCES 2006–2016, Ordered Logit (Clustered on State-Year).

Variables	(1)Governor	(2)State legislature	(3)Governor	(4)State legislature
Relative unemployment	−14.45*** (1.95)	−12.24*** (2.61)	−12.68*** (2.02)	−4,76* (2.09)
National unemployment	−8.66*** (1.72)	−4.93*** (1.42)	−8.67*** (1.70)	−5.49*** (1.33)
Knowledge			−0.03* (0.02)	−0.14*** (0.02)
Knowledge × Relative Unemployment			−1.45 (1.41)	−6.17*** (1.53)
Legislature–respondent same ideology		0.22*** (0.02)		0.22*** (0.02)
Legislature–respondent same party		0.06*** (0.01)		0.06*** (0.01)
Presidential approval	−0.17*** (0.03)	0.03 (0.02)	−0.16*** (0.03)	0.03* (0.02)
President–legislature same party		−0.65*** (0.08)		−0.65*** (0.08)
Presidential Approval × President–Legislature Same Party		0.21*** (0.03)		0.20*** (0.03)
Gubernatorial approval		0.97*** (0.04)		0.97*** (0.04)
Governor–legislature same party		−0.02 (0.01)		−0.01 (0.01)
Governor Approval × Governor–Legislature Same Party		0.02*** (0.01)		0.02*** (0.01)
Governor–Respondent Same Ideology	0.30*** (0.02)		0.31*** (0.02)
Governor–Respondent Same Party	0.17*** (0.01)		0.17*** (0.01)
President–Governor Same Party	−0.98*** (0.10)		−0.99*** (0.10)
Presidential Approval × President–Governor Same Party	0.56*** (0.04)		0.56*** (0.04)
/cut1	−0.56 (0.16)	0.45 (0.13)	−0.60 (0.16)	0.23 (0.13)
/cut2	0.50 (0.15)	2.14 (0.14)	0.46 (0.15)	1.94 (0.14)
/cut3	0.96 (0.15)	3.21 (0.15)	0.92 (0.15)	3.02 (0.15)
/cut4	2.88 (0.15)	6.15 (0.17)	2.84 (0.15)	5.93 (0.17)
Observations	327,969	277,265	327,969	277,265
df_m	7	10	9	12
Log likelihood	−454,821	−341,755	−454,755	−340,668
AIC	909,664	683,537	909,536	681,368
BIC	909,782	683,685	909,675	681,537

Note. Robust standard errors in parentheses. CCES = Cooperative Congressional Election Studies; AIC = Akaike information criterion; BIC = Bayesian information criterion.

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

There are two complications in interpreting the results on Table 2. First, with such large Ns, tests of statistical significance become meaningless, as almost any variable will reach conventional significance levels. Second, ordered logit coefficients are not intuitively interpretable. To address both of the concerns, the ordered logit coefficients are converted into probabilities. Figure 1 presents plots of the effect of relative unemployment for gubernatorial and state legislative approval, with separate subpanels for each approval category.

Figure 1.

Impact of relative unemployment on gubernatorial and state legislative approval. Panel 1a: Gubernatorial approval (based on estimations from Model 1, Table 2). Panel 1b: State legislative approval (based on estimations from Model 2, Table 2).

Figure 1 reveals several patterns of the effect of relative unemployment on gubernatorial and state legislative approval. First, voters are responsive to relative unemployment when evaluating both institutions, but relative unemployment appears to have stronger impacts on gubernatorial than state legislative approval. Especially for the strong disapproval, approval, and strong approval categories, the probability slopes are steeper for gubernatorial approval. As an example, a shift from the best relative unemployment rate (0.05 or 5%) to the worst (–0.05 or –5%) shows a probability shift for approval of the governor of 0.23 (from 0.38 to 0.15). For state legislative approval the probability effect is 0.16 (from 0.30 to 0.14).

Second, voters seem to blame state policy makers more strongly for poorer relative unemployment than reward them for better relative unemployment. For governors, the shift from best (–0.05) to worst (0.05) relative unemployment has a 0.13 and 0.15 effect for approval and strong approval, respectively, both much smaller than the 0.23 for strong disapproval. For state legislatures, there is little difference in the probability effects when comparing strong disapproval (0.16) and approval (0.14), but the difference between strong disapproval and approval (0.06) is pronounced at 0.10.

This asymmetric effect has been noted in other studies of gubernatorial approval. Hansen (1999b) notes in her comparative study of eight states that “high unemployment hurts governors’ ‘poor’ ratings worse than lower unemployment lifts ‘good’ ratings” (p. 179), but she does not offer an explanation for this asymmetry. Owen (2011) also finds asymmetric unemployment effects in his study of gubernatorial approval (p. 89).

Several reasons may account for the asymmetric effect of unemployment. One, there may be more news about economic performance when the state’s economy is doing poorly compared to the nation. Plus, when the state economy lags behind the national economy, politicians seeking to oust incumbents may try to exploit the situation as a campaign issue. The behavior of challenger candidates, thus, may stimulate increased news coverage on relatively bad economic performance. Two, negative conditions may outweigh positive one in voter evaluations, the negativity effect. Psychologists provided an underpinning for such asymmetric effects, in which people respond more strongly to negative than positive information and stimuli (Rozin and Royzman 2001) and loss aversion seems a stronger reaction than the prospect of gains (Tversky and Kahneman 1991). Soroka’s (2014) review of research on negativity and politics reports that negative evaluations often dwarf positive ones. The data used here, however, are limited in sorting through these alternatives.

Impact of Controls

All of the controls statistically perform as hypothesized. Again, to facilitate interpretation, the ordered logit coefficients are converted in probabilities. Rather than discuss the impact of each control on each approval outcome, to conserve space, I only discuss effect on the approval outcome category.

First, even with relative unemployment included, higher national unemployment depresses approval of the governor and the state legislature. For example, a change from the lowest national unemployment rate (0.025) to the highest (0.14) lowers the probability of approval for the governor by 0.10 and for the state legislature by 0.065, or approval changes of 10% (governor) and 6.5% (state legislature). These are substantively significant effects of national unemployment, similar in magnitude to that found for relative unemployment.

Shared party identification and ideology also increase approval of the governor and state legislature. Shifting from being a strong partisan of the governor’s (state legislative) party to a strong opposition identifier lowers the probability of approval by 0.105 for the governor but only about 0.03 for the state legislature. A shift from being a strong ideologue of the governor’s (state legislative) party to a strong opposition ideologue lowers the probability of approval by 0.13 for both the governor and state legislature. Except for shared partisanship’s effects on state legislative approval, all of these effects are quite large and substantively meaningful.

The political context also affects approval of the two bodies. Party control should condition the effects of presidential approval, showing positive impacts when the governor (state legislature) is of the president’s party, but negative effects when the opposition party controls these institutions. Results provide support for this conditional effect. To interpret the interaction effects, presidential approval on the table denotes when the opposition is in control, but the interaction term (Presidential Approval × President [governor or legislature] Same Party) displays the effect on same-party control. For the governor, the logit coefficients are −0.17 for the opposition and 0.56 for same party, and for the state legislature the coefficients are −0.56 and 0.04. Stated in probability terms, shifting from strong presidential disapproval to strong approval increases approval of governors of the same party by 0.17 and lowers approval for opposition governors by 0.08, or 17% and 8% respectively. For the state legislature, comparable effects for same party control are 0.10, but opposition party displays a meager, and incorrectly signed, 0.02 probability effect. Limited political knowledge of the state legislature may account for this reversed sign.

Finally, gubernatorial approval, conditioned by party control, also affects approval of the state legislature. Again, in reading the table, the coefficient for gubernatorial approval indicates the effect when the opposition controls at least one legislative chamber, while the interaction term (Governor Approval × Governor–State Legislature Same Party) indicates gubernatorial approval when the governor’s party controls both legislative chambers. Both coefficients are large (0.40 for same party and 0.70 for the opposition party), but the positive sign for gubernatorial approval coefficient is contrary to the conditional effects’ hypothesis. Turning to probability effects, shifting from strong gubernatorial disapproval to strong approval increases legislative approval by a massive 0.56 under same-party control. But contrary to the conditional effects hypothesis, shifting from strong gubernatorial disapproval to strong approval uplifts legislative approval by 0.44. Both of these effects are massive, and again the incorrect sign for opposition party control may be due to the limited political knowledge about the state legislature. Many voters may be guessing as to the party controlling the state legislature, thinking in many cases that the legislature and governor are of the same party.

Voter Knowledge and Accountability

Above analyses found stronger relative unemployment effects for governors than state legislatures. Does political knowledge mediate voters’ ability to hold these policy makers accountable for the state’s relative unemployment? At the individual-level, the knowledge mediation hypothesis predicts that voters with more knowledge will be better able to apply the relative unemployment standard than those with less knowledge. The governor and state legislature offer useful comparisons cases for testing this hypothesis because of the aggregate differences in voter knowledge about them, the widespread knowledge about the governor but comparatively sparse knowledge about the state legislature.

An interaction term between knowledge and relative unemployment is used to test this hypothesis. The knowledge index counts the number of correct responses to questions asking voters to identify the party controlling the governor, state Senate, and state House. Responses naming the wrong party or not naming a party were coded as incorrect. The index varies from 0 to 3—32.2% got all three correct, 18.3% two, 28.4% one, and 21.2% none.

Table 2 presents the results. Interactions are difficult to interpret from statistical results, and given the possibility of collinearity among the variables that comprise the interaction terms, significance tests are not very meaningful. As above, the ordered logit coefficients are converted to probabilities and plotted on two graphs, Figure 2 for gubernatorial approval and Figure 3 for state legislatures. Both figures have subpanels for each approval outcome.

Figure 2.

Impact of knowledge-relative unemployment interaction on gubernatorial approval.

Figure 3.

Impact of knowledge-relative unemployment interaction on state legislative approval.

For governors, Figure 2 indicates no apparent interaction between knowledge and relative unemployment. The slopes for each knowledge level are indistinguishable in most cases for most approval outcomes. Only for strong disapproval and approval do we see a slight spreading in the probability of these outcomes when state unemployment soars above that of the nation’s, with more knowledgeable respondents appearing marginally more responsive to a relatively worse economic state than national economy than less knowledgeable voters. For instance, there is no discernible difference in strong disapproval by knowledge when relative unemployment is at its best compare to the nation (–0.05). But when relative unemployment is at its worst (0.05), the most knowledgeable have are about 8% more likely to strongly disapprove (0.41) than the least knowledgeable (0.33). This however is the strongest interactive effect.

Knowledge has stronger interactive effects with relative unemployment for state legislative approval. This effect is most apparent for strong disapproval and approval, and when blaming the state legislature for bad times compared to rewarding for good times. When relative unemployment is at its best (–0.05), there is only a 0.04 difference in the probability of strong disapproval for the least (0.16) versus the most knowledgeable (0.12). But when relative unemployment is at its worst (0.05), the difference expands to 0.19, to 0.40 for the most knowledgeable and 0.21 for the least knowledgeable. The most knowledgeable are more responsive to changing relative unemployment conditions than the least knowledgeable, increasing their disapproval rate by 0.28 from the best to worst relative unemployment, while the least knowledgeable only move 0.05.

A similar pattern is observed for approval. When relative unemployment is at its best (–0.05), there is only a 0.04 difference in the probability of approval for the least (0.35) and most (0.39) knowledgeable. But when relative unemployment is at its worst (–0.05), the gap in the probability of approval grows to 0.15, to 0.31 for the least knowledgeable and 0.16 for the most knowledgeable. The probability of approval falls by 0.04 for the least knowledgeable when comparing the best and worst relative unemployment contexts, but 0.23 for the most knowledgeable. Knowledge effects on state legislative approval are substantively consequential.

Collective Accountability Effects of Knowledge

Analysis thus far has focused on the individual-level attributes associated with applying the relative unemployment standard to rate state policy makers. Knowledge was hypothesized to affect voters’ ability to apply the relative unemployment standard. The mediating effect of knowledge was found to affect ratings of the state legislature, but not the governor. Compared to the state legislature, knowledge about the governor generally is widespread. But the differences in application of the relative unemployment standard across knowledge levels for the state legislature raise the question of whether voters collectively can hold the state legislature accountable. How much knowledge in the electorate collectively is necessary to hold the state legislature accountable for the state’s economy?

To address this question, I simulated the impact of changes in relative unemployment for different levels of knowledge in the electorate. Results of these simulations are plotted on Figure 4. On the figure, the mean approval rating is plotted for different relative unemployment levels for different distributions on the knowledge index. The aggregate level of knowledge for the state legislature is 1.7; that is, on the 0 to 3 knowledge index, voters on average get 1.7 items correct. This serves as a baseline for assessing collective accountability as knowledge levels increase or decrease from that observed in these data.

Figure 4.

Simulating the effects of political knowledge on state legislative approval by relative unemployment levels.

At the baseline knowledge level, there appears to be collective responsiveness to changing relative unemployment levels. Results indicate that when the state economy is performing much better than the nation’s (relative unemployment = −0.05), the state legislature receives a higher average approval rating than when state unemployment is much higher (0.05) than the nation’s—mean approval drops from 2.06 to 1.39. When knowledge levels rise, responsiveness to relative unemployment becomes somewhat sharper too. Thus, when there is complete knowledge across the electorate (knowledge = 3), the mean approval rating drops from 2.15 to 1.13 when comparing the best and the worst state relative unemployment contexts.

As knowledge decreases from the baseline, aggregate voter responsiveness to relative unemployment also fades. When aggregate knowledge falls to 1, that is, on average voters get one of the three items in the knowledge index correct, legislative approval from the best to worst relative unemployment figure shifts from 2.02 to 1.54. Even at this low level of aggregate knowledge, there is still some aggregate level accountability. Only when knowledge levels fall much below 1.0 do we observe little or no aggregate voter responsiveness to relative unemployment. At the extreme of zero knowledge, when no voter gets any of the knowledge questions correct, there is little difference in the mean approval ratings of the state legislature as relative unemployment shifts from best to worst—with an approval rating of 1.95 for the best relative performance and 1.76 for the worst, a difference hardly worth noting. Consistent with other research on the effects of aggregation (Erikson, MacKuen, and Stimson 2002; Page and Shapiro 1992), these results suggest that collective accountability can occur if a small number of voters apply the relative unemployment standard correctly, while a large number of voters appear to randomly apply the standard; random responses cancel out when aggregated.

Conclusion

Research on whether voters hold state policy makers accountable for the state’s unemployment rate has produced mixed results. Benchmarking theory argues that voters compare the state’s unemployment level with the national level, the relative unemployment rate. The appeal of benchmarking theory is consistent with the limited information that voters possess about politics and policy making, and benchmarking imposes less onerous informational demands on voters than other accountability rules. Still benchmarking unemployment requires some level of voter information. This article addresses this question by comparing voter assessments of their state governor with their state legislature. Such is a useful comparison because there is widespread knowledge about the governor but scant knowledge concerning the state legislature.

Using CCES polls from 2006 to 2016, analysis finds support for the mediating impact of voter knowledge on holding the governor and state legislature accountable for the state’s relative unemployment level. The mediating effect of knowledge, however, is much stronger for the legislature than the governor, arguably because of broad levels of knowledge of the executive. For the legislature, voters lacking in knowledge are less likely to apply the relative unemployment standard than those possessing more knowledge. Yet even with the low level of knowledge about the state legislature, collectively there may be enough to hold the legislature accountable. Knowledge levels would have to fall some increment below current observed levels to immunize the state legislature from the state’s relative unemployment rate. Thus, while federal arrangements raise the hurdle for voters to hold the state legislature accountable, at present and observed knowledge levels this hurdle is not insurmountable.

This study suggests several directions for future research. The knowledge questions used focus exclusively on political control. Knowledge about the economic performance at the state and national levels may also facilitate voter application of the relative unemployment standard. Future studies should incorporate economic performance knowledge questions, alongside political knowledge, to test more fully for the effects of knowledge on the voters’ ability to hold policy makers accountable and to compare the relative importance of the two types of knowledge.

Research should also investigate relative unemployment in voting decisions for the governor and state legislature, as well as other state/local policymaking bodies. Voting and job approval differ. Most basically, voting requires a comparison between candidates, while job approval ratings do not. Voters and nonvoters may also differ in certain ways (Leighley and Nagler 2013) that may be relevant to the relative unemployment comparison, and accountability in general. For instance, voters know more about politics than nonvoters. If knowledge levels are higher among voters than nonvoters, there may be a tighter linkage between relative unemployment in elections compared to approval ratings.

Supplemental Material

Online_Appendix_relative_unemployment_r_and_r_sppq_8_1_19 – Supplemental material for Relative Unemployment, Political Information, and the Job Approval Ratings of State Governors and Legislatures

Supplemental material, Online_Appendix_relative_unemployment_r_and_r_sppq_8_1_19 for Relative Unemployment, Political Information, and the Job Approval Ratings of State Governors and Legislatures by Jeffrey E. Cohen in State Politics & Policy Quarterly

Footnotes

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.

ORCID iD

Jeffrey E. Cohen

Supplemental material

Supplemental material for this article is available online.

Notes

Author Biography

Jeffrey E. Cohen is a professor of Political Science at Fordham University. His major research interests are on American executives, especially presidents and governors. His most recent book is “The President on Capitol Hill: A Theory of Institutional Influence,” (Columbia University Press, 2019).

References

Althaus

Scott L.

Cizmar

Anne M.

Gimpel

James G

. 2009. “Media Supply, Audience Demand, and the Geography of News Consumption in the United States.” Political Communication 26 (3): 249–77.

Anderson

Cameron D.

2006. “Economic Voting and Multilevel Governance: A comparative Individual-level Analysis.” American Journal of Political Science 50 (2): 449–63.

Ansolabehere

Stephen

Meredith

Marc

Snowberg

Erik

. 2012. “Sociotropic Voting and the Media.” In Improving Public Opinion Surveys: Interdisciplinary Innovation and the American National Election Studies, ed. Aldrich

J. H.

McGraw

K. M

. Princeton: Princeton University Press, 175–89.

Arceneaux

Kevin

. 2005. “Does Federalism Weaken Democratic Representation in the United States?” Publius: The Journal of Federalism 35 (2): 297–311.

Arceneaux

Kevin

. 2006. “The Federal Face of Voting: Are Elected Officials Held Accountable for the Functions Relevant to Their Office?” Political Psychology 27 (5): 731–54.

Arnold

R. Douglas

Carnes

Nicholas

. 2012. “Holding Mayors Accountable: New York’s Executives from Koch to Bloomberg.” American Journal of Political Science 56 (4): 949–63.

Atkeson

Lonna Rae

Partin

Randall W

. 1995. “Economic and Referendum Voting: A Comparison of Gubernatorial and Senatorial Elections.” American Political Science Review 89 (1): 99–107.

Aytaç

Selim Erdem

. 2018. “Relative Economic Performance and the Incumbent Vote: A Reference Point Theory.” The Journal of Politics 80 (1): 16–29.

Bernick

E. Lee

. 2016. “Studying Governors over Five Decades: What We Know and Where We Need to Go?” State and Local Government Review 48 (2): 132–46.

10.

Besley

Timothy

Case

Anne

. 1995. “Incumbent Behavior: Vote Seeking, Tax Setting and Yardstick Competition.” American Economic Review 85 (1): 25–45.

11.

Brace

Paul

. 1993. State Government and Economic Performance. Baltimore: Johns Hopkins University Press.

12.

Brown

Adam R.

2010. “Are Governors Responsible for the State Economy? Partisanship, Blame, and Divided Federalism.” The Journal of Politics 72 (3): 605–15.

13.

Delli Carpini

Michael X.

Keeter

Scott

. 1996. What Americans Know about Politics and Why It Matters. New Haven: Yale University Press.

14.

Delli Carpini

Michael X.

Keeter

Scott

Kennamer

J. David

. 1994. “Effects of the News Media Environment on Citizen Knowledge of State Politics and Government.” Journalism Quarterly 71 (2): 443–56.

15.

Carsey

Thomas M.

Wright

Gerald C

. 1998. “State and National Factors in Gubernatorial and Senatorial Elections.” American Journal of Political Science 42 (3): 994–1002.

16.

Chubb

John E.

1988. “Institutions, the Economy, and the Dynamics of State Elections.” American Political Science Review 82 (1): 133–54.

17.

Coffey

Daniel

. 2005. “Measuring Gubernatorial Ideology: A Content Analysis of State of the State Speeches.” State Politics & Policy Quarterly 5 (1): 88–103.

18.

Cohen

Jeffrey E.

1983. “Gubernatorial Popularity in Nine States.” American Politics Quarterly 11 (2): 219–35.

19.

Cohen

Jeffrey E.

2018. “From Antisemitism to Philosemitism? Trends in American Attitudes toward Jews from 1964 to 2016.” Religions 9 (4): 107–28.

20.

Cohen

Jeffrey E.

King

James D

. 2004. “Relative Unemployment and Gubernatorial Popularity.” Journal of Politics 66 (4): 1267–82.

21.

Crew

Robert E.

Jr. Branham

David

Weiher

Gregory R.

Bernick

Ethan

. 2002. “Political Events in a Model of Gubernatorial Approval.” State Politics & Policy Quarterly 2 (3): 283–97.

22.

Crew

Robert E.

Jr. Weiher

Gregory R

. 1996. “Gubernatorial Popularity in Three States: A Preliminary Model.” The Social Science Journal 33 (1): 39–54.

23.

DiLeo

Daniel

. 2001. “To Develop or to Redistribute? An Analysis of the Content of Governors’ Agendas.” State and Local Government Review 33 (1): 52–59.

24.

Ebeid

Michael

Rodden

Jonathan

. 2006. “Economic Geography and Economic Voting: Evidence from the US States.” British Journal of Political Science 36 (3): 527–47.

25.

Erikson

Robert S.

MacKuen

Michael B.

Stimson

James A

. 2002. The Macro Polity. New York: Cambridge University Press.

26.

Evans

Geoffrey

Pickup

Mark

. 2010. “Reversing the Causal Arrow: The Political Conditioning of Economic Perceptions in the 2000–2004 US Presidential Election Cycle.” The Journal of Politics 72 (4): 1236–51.

27.

Ferguson

Margaret

. 2014. “State Executives.” In The Oxford Handbook of State and Local Government, ed. Haider-Markel

D. P.

(pp. 319–342). New York: Oxford University Press.

28.

Fording

Richard C

Patton

Dana J

. 2019. “Medicaid Expansion and the Political Fate of the Governors Who Support It.” Policy Studies Journal 47:274–99.

29.

Gerber

Alan S.

Huber

Gregory A

. 2009. “Partisanship and Economic Behavior: Do Partisan Differences in Economic Forecasts Predict Real Economic Behavior?” American Political Science Review 103 (3): 407–26.

30.

Grady

Dennis O.

1989. “Governors and Economic Development Policy: The Perception of their Role and the Reality of their Influence.” Policy Studies Journal 17 (4): 879–94.

31.

Hamman

John A.

2004. “Career Experience and Performing Effectively as Governor.” The American Review of Public Administration 34 (2): 151–63.

32.

Hansen

Susan B.

1999a. “Governors’ Job Performance Ratings and State Unemployment: The Case of California.” State and Local Government Review 31 (1): 7–17.

33.

Hansen

Susan B.

1999b. “‘Life Is Not Fair’: Governors’ Job Performance Ratings and State Economies.” Political Research Quarterly 52 (1): 167–88.

34.

Hayes

Danny

Lawless

Jennifer L

. 2015. “As Local News Goes, so goes Citizen Engagement: Media, Knowledge, and Participation in US House Elections.” The Journal of Politics 77 (2): 447–62.

35.

Herzik

Eric B.

1991. “Policy Agendas and Gubernatorial Leadership.” In Gubernatorial Leadership and State Policy, ed. Herzik

E. B.

Brown

B. W.

(pp. 25–37). New York: Greenwood.

36.

Hopkins

Daniel J.

2018. The Increasingly United States: How and Why American Political Behavior Nationalized. Chicago: University of Chicago Press.

37.

Hopkins

Daniel J.

Pettingill

Lindsay M

. 2017. “Retrospective Voting in Big-city US Mayoral Elections.” Political Science Research and Methods 6:697–714.

38.

Howell

Susan E.

Vanderleeuw

James M

. 1990. “Economic Effects on State Governors.” American Politics Quarterly 18 (2): 158–68.

39.

Hummel

Patrick

Rothschild

David

. 2014. “Fundamental Models for Forecasting Elections at the State Level.” Electoral Studies 35 (1): 123–39.

40.

Jacobs

Nicholas

. 2017. “An Experimental Test of How Americans Think about Federalism.” Publius: The Journal of Federalism 47 (4): 572–98.

41.

Jacobson

Gary C.

2006. “The Polls: Polarized Opinion in the States: Partisan Differences in Approval Ratings of Governors, Senators, and George W. Bush.” Presidential Studies Quarterly 36 (4): 732–57.

42.

Kayser

Mark Andreas

Peress

Michael

. 2012. “Benchmarking across Borders: Electoral Accountability and the Necessity of Comparison.” American Political Science Review 106 (3): 661–84.

43.

Kelleher

Christine A.

Wolak

Jennifer

. 2007. “Explaining Public Confidence in the Branches of State Government.” Political Research Quarterly 60 (4): 707–21.

44.

Kenney

Patrick J.

Rice

Tom W

. 1983. “Popularity and the Vote: The Gubernatorial Case.” American Politics Quarterly 11 (2): 237–41.

45.

Kincaid

John

Cole

Richard L

. 2010. “Citizen Attitudes toward Issues of Federalism in Canada, Mexico, and the United States.” Publius: The Journal of Federalism 41 (1): 53–75.

46.

King

James D.

Cohen

Jeffrey E

. 2005. “What Determines a Governor’s Popularity?” State Politics & Policy Quarterly 5 (3): 225–47.

47.

Krause

George A.

Melusky

Benjamin F

. 2014. “Retrospective Economic Voting and the Intertemporal Dynamics of Electoral Accountability in the American States.” The Journal of Politics 76 (4): 1102–15.

48.

Langehennig

Stefani

Zamadics

Joseph

Wolak

Jennifer

. 2019. “State Policy Outcomes and State Legislative Approval.” Political Research Quarterly 72:929–43.

49.

Layton

Charles

Dorroh

Jennifer

. 2002. “Sad State: Newspaper Budget-cutting has Triggered a Sharp Drop in the Number of Reporters Assigned to Cover the Nation’s Increasingly Important Statehouses.” American Journalism Review 24 (5): 19–35.

50.

Leighley

Jan E.

Nagler

Jonathan

. 2013. Who Votes Now? Demographics, Issues, Inequality, and Turnout in the United States. Princeton: Princeton University Press.

51.

Leyden

Kevin M.

Borrelli

Stephen A

. 1995. “The Effect of State Economic Conditions on Gubernatorial Elections: Does Unified Government Make a Difference?” Political Research Quarterly 48 (2): 275–90.

52.

Lowry

Robert C.

Alt

James E.

Ferree

Karen E

. 1998. “Fiscal Policy Outcomes and Electoral Accountability in American States.” American Political Science Review 92 (4): 759–74.

53.

Lyons

Jeffrey

Jaeger

William P.

Wolak

Jennifer

. 2013. “The Roots of Citizens’ Knowledge of State Politics.” State Politics & Policy Quarterly 13 (2): 183–202.

54.

MacDonald

Jason A.

Sigelman

Lee

. 1999. “Public Assessments of Gubernatorial Performance: A Comparative State Analysis.” American Politics Quarterly 27 (2): 201–15.

55.

Newman

Benjamin J.

Johnson

Joshua

. 2012. “Ethnic Change, Concern over Immigration, and Approval of State Government.” State Politics & Policy Quarterly 12 (4): 415–37.

56.

Niemi

Richard G.

Bremer

John

Heel

Michael

. 1999. “Determinants of State Economic Perceptions.” Political Behavior 21 (2): 175–93.

57.

Niemi

Richard G.

Stanley

Harold W.

Vogel

Ronald J

. 1995. “State Economies and State Taxes: Do Voters Hold Governors Accountable?” American Journal of Political Science 39 (4): 936–57.

58.

Orth

Deborah A.

2001. “Accountability in a Federal System: The Governor, the President, and Economic Expectations.” State Politics & Policy Quarterly 1 (4): 412–32.

59.

Owen

Andrew

. 2011. The Negativity Effect in Retrospective Voting. PhD dissertation, Department of Politics, Princeton University, Princeton.

60.

Page

Benjamin I.

Shapiro

Robert Y

. 1992. The Rational Public: Fifty Years of Trends in Americans’ Policy Preferences. Chicago: University of Chicago Press.

61.

Partin

Randall W.

1995. “Economic Conditions and Gubernatorial Elections: Is the State Executive Held Accountable?” American Politics Quarterly 23 (1): 81–95.

62.

Peltzman

Sam

. 1987. “Economic Conditions and Gubernatorial Elections.” The American Economic Review 77 (2): 293–97.

63.

Powell

G. Bingham

Jr.

2000. Elections as Instruments of Democracy: Majoritarian and Proportional Visions. New Haven: Yale University Press.

64.

Powell

G. Bingham

Jr. Whitten

Guy D

. 1993. “A Cross-national Analysis of Economic Voting: Taking Account of the Political Context.” American Journal of Political Science 37 (2): 391–414.

65.

Richardson

Lilliard E.

Jr. Konisky

David M.

Milyo

Jeffrey

. 2012. “Public Approval of U. S. State Legislatures.” Legislative Studies Quarterly 37 (1): 99–116.

66.

Richardson

Lilliard E.

Milyo

Jeffrey

. 2016. “Giving the People What They Want? Legislative Polarization and Public Approval of State Legislatures.” State and Local Government Review 48 (4): 270–81.

67.

Rodden

Jonathan

Wibbels

Erik

. 2011. “Dual Accountability and the Nationalization of Party Competition: Evidence from Four Federations.” Party Politics 17 (5): 629–53.

68.

Rogers

Jonathan

. 2016. “Tea Party Support and Perceptions of Local Economic Conditions.” Electoral Studies 42 (1): 91–98.

69.

Rogers

Steven

. 2017. “Electoral Accountability for State Legislative Roll Calls and Ideological Representation.” American Political Science Review 111 (3): 555–71.

70.

Rozin

Paul

Royzman

Edward B

. 2001. “Negativity Bias, Negativity Dominance, and Contagion.” Personality and Social Psychology Review 5 (4): 296–320.

71.

Shaker

Lee

. 2009. “Citizens’ Local Political Knowledge and the Role of Media Access.” Journalism & Mass Communication Quarterly 86 (4): 809–26.

72.

Soroka

Stuart N.

2014. Negativity in Democratic Politics: Causes and Consequences. New York: Cambridge University Press.

73.

Stein

Robert M.

1990. “Economic Voting for Governor and US Senator: The Electoral Consequences of Federalism.” The Journal of Politics 52 (1): 29–53.

74.

Svoboda

Craig J.

1995. “Retrospective Voting in Gubernatorial Elections: 1982 and 1986.” Political Research Quarterly 48 (1): 135–50.

75.

Taylor

Charles D.

2012. “Governors as Economic Problem Solvers: A Research Commentary.” Economic Development Quarterly 26 (3): 267–76.

76.

Tversky

Amos

Kahneman

Daniel

. 1991. “Loss Aversion in Riskless Choice: A Reference-dependent Model.” The Quarterly Journal of Economics 106 (4): 1039–61.

77.

Vining

Richard L.

Wilhelm

Teena

Hiers

Sara E.

Marcin

Phil

. 2010. “Patterns of Newspaper Reporting on State Supreme Courts.” Justice System Journal 31 (3): 273–89.

78.

Weinberg

Micah

. 2010. “Measuring Governors’ Political Orientations Using Words as Data.” State Politics & Policy Quarterly 10 (1): 96–109.

79.

Wolfers

Justin

. 2002. Are Voters Rational? Evidence from Gubernatorial Elections. Stanford: Graduate School of Business, Stanford University.

80.

Wright

John R.

2012. “Unemployment and the Democratic Electoral Advantage.” American Political Science Review 106 (4): 685–702.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.24 MB