Nina van Ee s1934643
Leiden University
Bachelor Project: Political Behaviour: Can we Trust Democracy to the Voters? Thesis Supervisor: Dr. Joshua Robison
December 20th, 2019 Word Count: 8100
Introduction
Donald Trump; the President of the United States of America, the first elected president with no political or military experience (Crockett, 2017). Many scholars try to grapple with the reason why Trump won the 2016 Presidential Elections; a candidate, who used explicit racial rhetoric throughout his campaign. The unexpected victory of Donald Trump under his slogan ‘Make America Great Again’ has been the subject in many newspaper articles and upcoming scholarly articles (Beaumont, 2016; Krieg, 2016; Zurcher, 2016; Smith, 2016; Schaffner, 2018).
The victory of Donald Trump has posed puzzling questions to many scholars as it relates to the broader phenomena of how ‘partisanship’ has returned in a more ideological form (Bafumi & Shapiro, 2009, p. 1). The partisan gap between Republicans and Democrats is growing on several issues, such as job ratings, whether the U.S. should be active in world affairs and how much the government should help the needy, and this partisan polarization is becoming a powerful force in American politics (Doherty, 2017). There are many explanations given to explain the massive support for Trump. One explanation for Trump’s popularity is that the perceived status threat felt by high-status groups (Mutz, 2018). This is related to the narrative given by Sides et al. (2018); where attitudes about immigration and economic entitlements were strong predictors for voting for Trump. On the other hand, Whitehead et al (2018) show that Christian nationalism is actually the main reason for the victory of Trump. Luttig et al. (2017), also adds that scholars have focused on Trump as a product of feelings of “economic anxiety, the growing populism of the Republican Party, sexism and authoritarianism” (p. 1). Finally, Schaffner et al. (2018), found that racial attitudes and sexism are the main reasons for the massive support for Donald Trump, as opposed to economic (dis) satisfaction, where Dwyer et al. (2009) also provides evidence that issues on race are becoming prominent in voting behavior. However, these last two articles mentioned only look at a specific time period, the Obama era or the new Trump era respectively. What is missing from the analysis is determining whether the Trump case is unique, or whether issues on race have been growing more and more salient in voting behavior over time. In order to determine whether the effect of racial attitudes was significant specifically in the case of Trump, we must consider at the effects of these two variables over time.
In order to understand why Trump won the 2016 elections, I will attempt to answer the following question; “What was the effect of racial attitudes on voting behavior during the Presidential Elections between 1992 and 2016?”. I conclude that racial resentment has a statistically significant effect on voting behavior. When comparing the data across time, it is evident that the effect of racial attitudes on voting behavior was significantly stronger in 2016, indicating that racism had an important role in Trump’s victory, which can be explained by the racial priming theory and social sorting.
Theoretical Framework
The surprising outcome of the 2016 American Presidential Elections has sparked a lot of debate in political science. An elected president with no political experience motivates us to try and understand the voting behavior during the 2016 Presidential Elections. Many explanations for the victory of Donald Trump have been given, such as the perceived status-threat theory, (anti)-immigration attitudes, economic (dis)satisfaction and Christian nationalism. Racial attitudes also proved to be high predictors for voting for Trump (Schaffner et al., 2018). What this paper aims to investigate is whether racial attitudes are likely to be more important in 2016 than 1992, and whether this reflects the 2016 elections or a general trend over time.
Race
Literature shows that racial attitudes can explain voting behavior (Schaffner et al., 2018; Luttig et al., 2017; Sears et al., 1997). Issues on race still remains central to understanding American politics, it is interesting to look how this is related in the case of Donald Trump. There is a big debate in political science about the role of race in Trump’s victory. Some articles find that race indeed influenced the victory of Donald Trump (Schaffner et al., 2018; Luttig et al, 2017), however was this specific to the year 2016, or an overall trend that we are seeing in American politics. In order to understand this debate, we first need to conceptualize racial attitudes.
There are many ways of defining racism; the first is the classic theory that ones’ race is superior to an other due to racial differences, and the second is ‘racial prejudice’ and discrimination (Sears et al., p. 19). Following the example of Sears et al. (1997). we will be focusing on the latter definition. Modern racism, expressed more subtly as opposed to Old Fashioned Racism, involves negative attitudes towards people of colour (Swim et al., 1995, p. 199). It “represents the resistance to change in the racial status quo”, which is that African Americans violate
traditional American values (Kinder & Sears, 1981, p. 416) such as working hard (Rabinowitz et al., 2009, p. 806). Old Fashioned Racism, a belief held by a majority of White Americans in the 20th century, had the following three elements: “(1) desire for social distances between the races, (2) beliefs in the biological inferiority of blacks, and (3) support public policies insuring racial segregation and formalized discrimination” (Tesler, 2013, p. 114).
Where Old Fashioned Racism (OFC) is focused on the belief of black biological and social inferiority, modern racism describes the feeling that Black Americans are violating the traditional American values (Tesler, 2013, p. 114). This is present in issues concerning welfare, an issue that was addressed in Trump’s campaign message; a strong support for social welfare benefits and entitlement programs (Sides et al., 2018, p. 138; Kinder & Sears, 1981, p. 416). The hostility towards African Americans may stem from the feeling that they symbolize a threat to their own private life; which can be generated from the means-ends formula from rational decision-making (Kinder & Sears, 1981, p. 415). Racial Resentment is one form of modern racism, although its exact definition is debatable. Tesler (2013) defines racial resentment as “attitudes that emphasize lack of black commitment to traditional American values” (p. 110). Wilson and Davis (2011) on the other hand understand racial resentment as a feeling of “animosity or antipathy toward a person or group of people who are perceived to be unfair or unjust recipients of some outcome” (p. 119). Kam & Burge (2017) base this concept upon three pillars; “(1) Anti-black affect; (2) a belief that African Americans have failed to conform to the Protestant work ethics; and (3) a denial of continuing discrimination against African Americans” (p. 1). The core difference between Modern Racism and Racial Resentment is that the latter portrays the attitudes individuals have towards a certain race, like not trying hard enough or getting too many favours (Enders & Scott, 2019, p 276), which influences things like voting behaviour. Wilson and Davis (2011) add to this debate by arguing that the level of racial resentment an individual has is based upon the perceived level of effort and determination they think that African Americans have (p. 119). This implies the thought that African Americans are undeserving of ‘special treatment’ and extra help (Wilson and Davis, 2011, p. 119). In sum; racism is not understood anymore as “segregation, discrimination, and biological superiority”, but is instead seen has the belief that Black’s don’t put in enough effort and the extent to which they are undeserving (Dweyer et al., 2009, p. 224-225).
The Role of Racism
Race has always been an important issue in Americans’ partisan preferences, which has stemmed from elite-level differences in “support for 1960s civil rights initiatives”, between the Democratic and Republican Party (Tesler, 2013, p. 110). Race has continued to divide the two parties, where racial resentment “shifted White Southerners support from the Democratic party to the Republican Party” (Tesler, 2013, p. 110). This marked the beginning of party polarization on matters of race. Valentino et al. (2017) state that:
Secular partisan realignment that begun in the 1960s, sorted the most racially conservative whites into the Republican party. Whites’ perceptions of their group’s racial distinctiveness and disadvantage have risen due to the demographic shifts and economic stagnation, which has boosted white identification. (p. 2)
Many scholars argue that race has been crucial in the transformation of American politics (Abramowitz, 1994, p. 1-2). The evolution of racial issues in the United States started in the 1960s when Democratic and Republican party leaders became polarized on issues of race, which lead to the polarization of party supporters amongst party lines (Abramowitz, 1994, p. 3). Carmines and Stimson (1989) conclude that “racial issues have transformed U.S. politics by shaping the development of party loyalties among voters who have entered the electorate since 1964” (as cited in Abramowitz, 1994, p. 1). It is therefore evident that there is some sort of relationship between race and political parties. I believe that because attitudes on race have aligned with partisan lines, race will be a strong determinant for voting behavior.
The alignment of racial attitudes amongst partisan lines has created an environment where parties are more polarized on racial standpoints. Since the dominant group (White Americans) feel threatened by the out-group, symbolic racism can arise. It is evident from existing literature that we can make the following hypothesis:
H1: As racial resentment attitudes increase, so does the probability that an individual will vote for a Republican candidate.
Existing work (Enders & Scott, 2019; Schaffner et al., 2018; Knuckey, 2011; Dweyer et al., 2009) have focused specifically on the role of racism in the 2016 Presidential elections. However, it is still puzzling whether the role of racism in American politics has gradually
intensified over time, or if racism was an especially important issue in 2016. Past work has focused on the relationship between racial attitudes and vote choice during a certain American Presidential campaign, however up till now there hasn’t been any concrete time-series analysis in order to determine whether there is a significant difference between the role of racism in 2016 and previous years. I aim to address this limitation in this analysis. Two competing theories can help us grapple with the issue of race in voting behavior. Racial Priming Theory potentially defends the position that race played a significant role in the outcome of the 2016 Presidential Elections. On the other hand, Social Sorting Theory holds that the issue of race has always played a role in American politics but has overtime become more significant, and that in the case of Donald Trump, race was not something special.
Racial Priming Theory
In order to understand the relationship between racial attitudes and voting behavior in the 2016 Presidential Elections, we must introduce the Racial Priming Theory. Racial Priming Theory holds that “subtle cues in news coverage, political advertising, or candidate speeches activate racial attitudes, boosting their impact on candidate evaluation or policy opinion” (Valentino et al, 2017 & Bracic et al., 2018). When issues about gender become salient, which was the case in Clinton and Trump’s campaign, a process of association between beliefs about gender and vote choice are activated (Bracic et al., 2018, p. 282). This must also be true for issues about race (Mendelberg, 2008). Statements such as calling all Mexicans “rapists”, “bringing crime” and “bringing drugs” and by calling a ban on all Muslims coming to the U.S. (Lopez, 2019) activated racial cues throughout the campaign. Similarly, Trump was able to dominate media coverage during his presidential campaign, where he was able to activate longstanding sentiments across Republican voters, such as racial predispositions (Sides et al., 2018, p. 136). Trump was able to appeal to Whites with conservative racial views which activated racial attitudes and made these important for voting behavior. When issues about race are made salient, it activates the racial schema, causing these attitudes to influence political behavior (Bracic et al., 2018). Donald Trump has used explicit racial rhetoric during many occasions throughout his presidential campaign which has activated a process making racial attitudes important in voting behavior. Racism will likely be a high predictor of Trump voters because as Schaffner et al. (2018) observed, a “voter who scored high on racism denial was about three times more likely to be an Obama/Trump voter than one that was acknowledging of racism” (p. 28). Schaffner et al. (2018) also notes that because whites now view themselves as an ‘embattled racial group’, it has made in-group identity stronger and more anger towards
out-groups (p. 13). This might be a possible reason for why race has been such an important factor in the voting behavior in 2016, especially after an African American President that was in office for eight years.
A core debate in Racial Priming literature, is the difference between implicit and explicit priming, and whether Trump’s rhetoric was effective or not. Let us first determine the difference between implicit and explicit racial cues. Implicit racial cues use a subtle communication style to convey a negative message about a racial group without using racial nouns or adjectives (Mendelberg, p. 110). For example, a photo of a black hand grabbing money, suggesting the African Americans are a threat to our well-being. Countless mentioning of the high amount of poverty experienced amongst African Americans, creates a perception that a large sum of welfare benefits goes to African Americans (Valentino & Neuner, 2016, p. 8). Explicit racial cues on the other hand are not subtle; these are cues that use racial nouns and adjectives in order to portray negative stereotypes of African Americans (Mendelberg, p. 110). Explicit racial cues are however not effective for White Americans because these racial cues go against the norm of racial equality, whereas implicit priming on the other hand is effective because they make “racial attitudes more accessible in memory” (Hutchings & Jardina, 2009, p. 398). A question that then comes to mind is how effective these implicit racial cues are in American politics. According to the study done by Huber and Lapinski, there is no evidence that any kind of racial cues have a significant effect on racial opinions (???). However, as Mendelberg states, there is so much literature that proves the opposite; that racial cues actually do work and that the process of racialization happens implicitly (p. 110).
According to Mutz (2017), “the 2016 election raised the salience of people’s pre-existing views on racist views, so that they mattered more in the Presidential vote choice in 2016” (p. 2). When members of the in-group group feel threatened it results in a bigger importance of conformity to group norms, and increased out-group negativity (Mutz, 2017, p. 2). According to Sides et al. (2017, p. 42) and Major et al. (2016, p. 932), the demographic shift in American politics has made issues related to racial, ethnic and social identities very important in the 2016 elections, which Trump was able to profit on. One of the explanations that Schaffner et al. (2018) provides for why Trump won is that Trump was willing to make racial appeals during his campaign to the less educated whites, which were a majority in swing states, who tend to exhibit higher levels of and racism (p. 10). What is interesting to note is that there is a shift of racial rhetoric in the US, namely that the white Americans are subject to racism, which has led
to a stronger in-group identity. As we saw from the outcome of the elections, the racial rhetoric has been fairly successful among less educated whites (Schaffner et al., 2018, p. 13).
Barack Obama, the first American President of colour, who served for two terms (2008-2016) has made issues on race more salient in American politics (Luttig & Callaghan, 2016, p. 628). Explicit racial attitudes had a significant effect on the voting behavior in the 2008 Presidential Elections, however this is an unavoidable consequence due to Obama’s race (Luttig & Callaghan, 2016, p. 644-645). Obama activated a dual process of racial resentment and ‘Old Fashioned Racism’ (OFR), which has made this white in-group identity stronger and has tapped into a feeling that blacks have too much influence in politics (Tesler, 2012, p. 121). In terms of communication throughout Obama’s campaign, racial attitudes were not necessarily primed, instead the fact that Obama is Black was enough to spark racial attitudes throughout the 2008 campaign.
Donald Trump has used explicit racial rhetoric countless times throughout his presidential campaign (Valentino & Neuner, 2016, p. 4). Even though study shows that explicit racial cues are not effective, new studies have actually proved the opposite; this is because Whites now view themselves as an “embattled racial group” which has made this group more tolerant to explicit racial rhetoric (Schaffner et al., 2018, p. 13). Examples include government displays of the Confederate flag not being rejected by some whites and that explicit racial cues can activate in-group identification, which has made American voters more tolerant towards racial rhetoric (Valentino & Neuner, 2016, p. 8). In line with the Racial Priming Theory, racial rhetoric used by Trump made racial attitudes salient in the elections, making Trump so successful. Similarly, the longstanding predispositions on race caused by centuries of socialization that has been marked by negative attitudes towards African Americans, can be activated through the use of political symbols; which is also known as symbolic politics theory (Sears et al. 1997, p. 18). Racial predispositions can also be activated through racial priming (Bracic et al., 2018; Mendelberg, 2008). Luttig et al. (2017), also add to this discussion by finding that “when race becomes salient in public discourse, support for Donald Trump will serve as a fulcrum for divergent policy judgements” (p. 7). This implies that racial priming is especially occurring in the case of Donald Trump.
Social Sorting Theory
A competing argument to attempt to understand the influence of race in the election of Donald Trump is the Social Sorting Theory. Alliances between parties and other social identities have been growing stronger in a process called social sorting, which has motivated partisans to cling more strongly to their partisan groups, intensifying the partisan social and affective polarization (Mason & Wronski, 2018, p. 259). Social sorting means that personal attachments to each of the objectively aligned racial, religious, ideological and partisan groups merge to create more loyal partisans (Mason & Wronski, 2018, p. 264). Sorting reduces identity-based cross pressures (Mason & Wronski, 2018) and facilitates motivated cognition (Mason, 2018). Since group identification is driven by a desire to positively distinguish one’s in-group, ingroup bias arises which pushes ingroup members to judge the members of their own group superior to the members of the outgroup. The relationship between social identities and partisan identities creates a politicization of Americans’ otherwise nonpolitical identities. In accordance to symbolic politics theory, racial predispositions which have occurred due to a process of socialization, aligns these attitudes closer with their party identity (Sears et al., 1997, p. 18). Racial identities have moved into a greater alignment with Democratic and Republican identities. Mason and Wronksi (2018) state that:
As partisan and ideological identities move into alignment, the same increase in partisan identity should be visible. The party-group alliance cues allow individuals to perceive their own social identity has allied or unallied with the political parties. The general identity of the Republican party, White and Christian conservative, make it easier to identify who the in-group and out-group are. (p. 261-262)
Trump being a Republican President, also pushes us to look at the importance of the Republican party in the effect process between racism and voting behavior. To understand the importance of the Republican party we must first understand the concept of polarization.
Study shows that political polarization in the United States has increased in the past few decades (Bafumi & Shapiro, 2009; Fiorina & Abrams, 2008; Westfall et al., 2015). Partisan identity voting in the US has always been high, but according to some political scientists, partisanship has turned more ideological and issue-based along liberal-conservative lines than before; which is evident through the strength between partisanship and ideology and partisanship voting in general (Bafumi & Shapiro, 2009, p. 1). We define partisan identity as
“a long term, affective attachment to a political party- one that developed early on in people’s lives” (as cited in Weinschenk, 2013, p. 608). Ensley (2007) argues that as ideological differences between Democratic and Republican candidates grow further apart, political polarization increases as well as ideological voting (as cited in Lachat, 2008, p. 689). There are multiple ways to define polarization; as alignment or divergence. Alignment refers to the “degree to which party identity increasingly matches ideology (sorting)” (Lelkes, 2016, p. 394). Divergence refers to “the degree to which the distribution of ideology has moved apart” (Lelkes, 2016, p. 394). I will be focusing on the former definition throughout the rest of the paper. There is a positive correlation between the relationship of partisanship and ideology to an individual’s economic, social, racial and religious attitudes and opinions (Bafumi & Shapiro, 2009, p. 1).
Changes in the political landscape that have occurred in the past decades in the US have led to a rise in party polarization. These two trends have changed the role of racism in voting behavior. The first trend is the partisan realignment in the 1930s and 1960s which made way for new issues, conflicts and cleavages which has resulted in an electorate that is led by liberal/conservative ideological concerns (Bafumi & Shapiro, 2009, p. 3). The new ideological positioning of individuals has been driven by new issues (racial, social and religious) and by leadership that has created a bigger divide in partisanship (Bafumi & Shapiro, 2009, p. 3). A leader such as Donald Trump that openly speaks about racial, social and religious issues, could create an even bigger divide between the Democratic and Republican Party, and could potentially make these attitudes essential characteristics of the Republican Party. By the 21st century, racially conservative White Americans were sorted into the Republican party as a result of the partisan realignment that occurred in the 1930s and 1960s (Valentino & Neuner, 2016, p. 4). The second change in the American political landscape has been the demographic shift that has led white Americans to believe that their racial group is under threat, making it acceptable to make explicit racial claims against the out-group (Valentina & Neuner, 2016, p. 5). Potentially, this shift in tolerance towards racial rhetoric has allowed Trump to successfully use explicit racial cues.
The relationship between partisanship and ideology is important in understanding why racial attitudes played a strong role in Trump’s victory. Salient racial identities push individuals to vote in a particular way because of the way that they move into a greater alignment with ideologies and partisan identity. “Voters construct their partisan identities using the aggregate
alliances between party and other social identities such as religion, race and ideology” (Mason & Wronski, 2018, p. 258). Partisan identity, which can be treated as a social identity, hand in hand with a salient racial identity can motivate individuals to pursue political action on the basis of racial issues and drive political judgement (Mason & Wronski, 2018, p. 259-260). This relationship between partisan and social identities politicizes Americans’ identities, which would otherwise be non-political. These racial identities have moved into a great alignment with ideologies and partisan identities, and this together with an elite partisan polarization has created an environment for Trump to take his victory in 2016. Mason and Wronksi (2018) state that:
As partisan and ideological identities move into alignment, the same increase in partisan identity should be visible. The party-group alliance cues allow individuals to perceive their own social identity has allied or unallied with the political parties. The general identity of the Republican party, White and Christian conservative, make it easier to identify who the in-group and out-group are. (p. 261-262)
When racial identities align with partisan identities it can lead to Partisan Motivated Reasoning, which is when “people defend their partisan identity by seeking information that reinforces their party’s positions and counter-arguing the which challenges their party” (Mullinix, 2015, p. 384). Racism still plays an important role in American politics and so a political candidate that recognizes the feelings these individuals on issues of race might strengthen an individuals’ partisanship. When looking at the American political landscape as a whole we can see that elite partisan polarization alters the effect partisanship has on policy preferences (Mullinx, 2015, p. 386). This is mainly because elite polarization “increases the salience of partisan identities” and it is clearer where party’s stand on particular issues (Mullinx, 2015, p. 386). Because elite polarization highlights partisan response to information, partisan identities are made more salient across party lines.
The Republican Party is more responsive to objective social sorting, due to their homogenous social makeup; a concept that states that personal attachments to each of the objectively aligned racial, religious, ideological and partisan groups merge to create more loyal partisans (Mason & Wronski, 2018, p. 264). According to Mason & Wronski (2018) “party group alliances should only lead to stronger partisan attachment when individuals can successfully perceive the cumulative alignment between their ingroups and in-party subjective social sorting” (p. 264). This means that stronger partisan attachment results only when individuals have a
personal attachment to racial, religious, ideological and partisan groups which is aligned with their in-group. This leads us to assume that there has been a linear trend since 1992 in the effect of race on voting behavior.
Donald Trump was not disadvantaged due to his explicit racial rhetoric. Trump’s racial rhetoric has activated the racial predispositions of certain American voters, which has made issues on race salient in the 2016 Elections. This leads us to our second hypothesis:
(H2) The effect of racial resentment on voting behavior is higher for Donald Trump than for any other elected president since 1992.
Operationalization
In order to research the stated research question (‘What was the effect of racial attitudes on voting behavior in the 2016 Presidential Elections’), we must take a large-N approach when it comes to the methodology. This is necessary because it allows us to make generalizable conclusions and is suitable for national data surveys. The data is taken from the Cumulative Data File from the The American National Election Survey (ANES) for the years 1948 to 2016. This dataset includes data from 59944 respondents where questions regarding voting behavior are asked. In order to answer the research question, I will be comparing the effect of racial and attitudes on the outcome of the Presidential elections between 1992 and 2016 and observe whether there is an increase in the strength of voting probability of these two variables due to an overall trend in American Politics, or if the effect was more significant in the case of Donald Trump. The years 1992-2016 were chosen because before 1992 there isn’t sufficient data on racial resentment. Furthermore, I would like to determine whether economic dissatisfaction had a bigger effect on vote choice compared to racial attitudes. I want to do this because of the strong relationship between economic perceptions and voting (Godbout & Bélanger, 2007, p. 542). The main independent variable in this analysis will be racial resentment, and the dependent variable will be vote choice. Partisanship is the moderator variable and the variables I will be controlling for are age, gender, income, education, race and ideology. To answer the research question I will be performing a logistic regression.
Independent Variables
In this study we will be focusing on three main independent variables; racial attitudes, sexist attitudes and economic (dis)satisfaction. Racial attitudes will be measured by a combination of four indicators included in the data set. Here opinions are asked on the following statements; (1) conditions that make it difficult for Blacks to succeed, (2) special favors for Blacks in order to succeed, (3) Blacks must try harder to succeed, and finally, (4) that Blacks have gotten less than they deserve over the past few years. Respondents must indicate how much they agree or disagree with these four statements (1=strongly agree and 5=strongly disagree).
In order to create one combined variable to explain racial attitudes, I first made sure the scale meant the same for all four indicators; I therefore recoded the second and third indicator making 1= 5, 2=4 and 3=3. This now means that each 1-point increase on the scale of 1-5, racist attitudes become more present. To create my racial attitudes variable I created a mean of the four variables, creating a interval-ratio variable ranging from a scale of 1-5, where 1 indicates low racist attitudes/racial resentment and 5 indicates very high racist attitudes/racial resentment. According to the Cronbach’s Alpha (0.778), the combined racial resentment scale is reliable to measure a respondent’s racial resentment. Schaffner et al. (2018), measures racist attitudes with the following three statements; “(1) white people in the US have certain advantages because of the colour of their skin, (2) racial problems in the US are rare, isolated situations and (3) I am angry that racism exists (p. 17-18). This is measured on a six-point scale. However, Dwyer et al. (2009) use the same questions I used to create the racist scale (p. 229).
One of the most important incentives to vote is economic interest (Wlezien et al., 1997; Markus, 1988; Godbout & Bélanger, 2007). This is why we chose to include this variable in our analysis. Economic (dis)satisfaction will be measured by the variable asking individual how much better or worse the national economy has gotten in the past year. This is measured on a scale of 1-5 and is treated as a continuous variable. This allows us to determine whether an individual is satisfied with how the economy is developing during the past year. Ideally a question asking how an individual thinks about the economy at this moment in time would be better as it can give us a better indication of how satisfied/unsatisfied that person is. Another question that would be helpful is asking individuals how well they think the economy has improved during a Presidential term, however this would not necessarily let us know whether they are satisfied or not. Schaffner et al. (2018) uses a similar variable to measure economic
dissatisfaction; namely how satisfied the individual is with their current economic situation (p. 19).
Dependent Variable
The dependent variable in this analysis is Vote Choice. We want to find out whether attitudes on racism has an effect on an individuals’ vote choice. Vote choice is measured by the ‘Vote Choice’ indicator, where 1=Democrat and 2= Republican. These are recoded to make 0=Democrat and 1=Republican. There are some limitations to using this indicator; namely that not all individuals are honest when stating who they voted for in the elections. Another way to measure voting behavior could be to look at the favorability of each presidential candidate, however this is not a good enough indicator to measure for who respondents voted for. Therefore the Vote Choice indicator is for the purpose of this analysis the most accurate variable.
Controlled Variables
Partisanship will be used as a moderator variable, where I predict that partisanship has created an environment for racial and sexist attitudes to become salient and become important in the voting behavior. Partisanship will be measured on a scale of 1-7 where 1= Strong Democrat and 7 is a Strong Republican. We must also control for a number of control variables. These are ‘Ideology’, ‘Education’, ‘Gender’, ‘Race’ and ‘Age’ and ‘Income’.
Results
I performed a Bivariate Logistical Regression for respondents between 1992 and 2016. The data is taken from the ANES Cumulative File. When simply looking at the racial resentment means for Democrats and Republicans across time, we already see that there is a difference in mean values. When looking at the mean racial resentment in total (between 1992 and 2016), for Democratic voters this is 2.88 and for Republican Voters this is 3.89. This means that between 1992, individuals that voted for a Democratic candidate scored 2.88 on average on the 1-5 racial resentment scale. Individuals that voted for a Republican candidate scored 3.89 on average on the 1-5 racial resentment scale. Figures 1-5 (see below) depict a difference in mean between racial resentment for Democratic Voters and racial resentment for Republican Voters between 1992 and 2016. The mean is higher in all years for Republican voters. The difference
of racial resentment levels for Democratic and Republican voters can also be seen with the use of a histogram.
Table 1
Mean of Racial Resentment for Democratic and Republican Voters (1992-2016)
Figure 1
Simple Boxplot of Racial Attitudes by Vote Choice (1992)
Note: 0= Democrat and 1= Republican
1 Data on racial resentment indicators is missing from the ANES cumulative data set.
Election Democratic Voters Republican Voters Difference
2016 2.4367 (1.03494) 3.9232 (0.81154) 1.4865 2012 2.9684 (0.98057) 4.0258 (0.74651) 1.0574 2008 3.0529 (0.91614) 3.9008 (0.78038) 0.8479 2004 2.8590 (1.03240) 3.7850 (0.81898) 0.926 2000 3.0766 (0.97752) 3.7166 (0.85883) 0.64 19961 - - - 1992 2.9661 (1.04759) 3.6564 (0.81452) 0.6903
Figure 2
Simple Boxplot of Racial Attitudes by Vote Choice (2000)
Note: 0= Democrat and 1= Republican
Figure 3
Simple Boxplot of Racial Attitudes by Vote Choice (2004)
Note: 0= Democrat and 1= Republican Figure 4
Simple Boxplot of Racial Attitudes by Vote Choice (2008)
Figure 5
Simple Boxplot of Racial Attitudes by Vote Choice (2012)
Note: 0= Democrat and 1= Republican
Figure 6
Simple Boxplot of Racial Attitudes by Vote Choice (2016)
Note: 0= Democrat and 1= Republican
When taking a first look at the relationship between racial resentment and vote choice, we can see that there is a strong positive, statistically significant, relationship present. Spearman’s Rho is 0.625 and is statistically significant (p<0.001). As racial resentment increases, individuals are more likely to vote for a Republican. Kendall’s Tau-b gives us a more accurate understanding of the correlation between racial attitudes and vote choice. Kendall’s Tau-b is 0.526, which means there is a strong positive correlation between racial resentment and vote choice (p<0.001).
Now that we know that there is a positive relationship between racial resentment and vote choice, we can perform a more detailed analysis in order to gain a better understanding of what this relationship exactly entails. A Binary Logistic Regression is the appropriate statistical analysis because our dependent variable is a dichotomous variable and our independent variable is interval-ratio. Let us first consider the first hypothesis; in order to prove whether or not high levels of racial resentment affect voting behavior, we must focus on the odds coefficients of the logistic regression. Additionally, a third model will be created in order to analyze the second hypothesis, which will compare the years 1992 to 2016 by creating an interaction variable and observing whether the odds coefficient in 2016 was significantly larger than in the other Presidential elections. The control variables must also be taken into account in this analysis because we can then see whether other factors such as education, age, income, gender and ideology have any effect on the relationship between racial attitudes and voting behavior.
Before performing the logistic regression, we must first test out if the analysis will violate any of the assumptions of linearity, independence of errors and multicollinearity. Since my variables are either binary or categorical (that have now been turned interval-ratio) the assumption of linearity of the logit has been met. Since the tolerance level is above 0.1 and VIF is smaller than 10, there are no collinearity problems.
Racial Resentment and Vote Choice
In the years 1992, 2000, 2004, 2008 and 2016, information about the racial attitudes of an individual increase the odds more than the level of economic dissatisfaction in guessing for which candidate an individual voted for in the Presidential Elections. This is summarized in table 3.
Table 3
Logistic Regression analysis of the probability of voting Democrat or Republican (Ex(B) scores)
Model 1 Model 2
Level of Racial Attitudes 2.396*** 2.393***
(0.041) (0.042) Economic Dissatisfaction 1.326*** 1.356*** (0.031) (0.032) Partisanship 2.933*** (0.022) Ideology 0.740*** (0.027) Age 1.014*** (0.002) Education 1.005 (0.048) Gender (Male) 79043266.1 (16184.956) Income (0-16 percentile) 0.701 (0.187) Income (17-33 percentile) 0.909 (0.181) Income (34-67 percentile) 0.836 (0.164) Income (68 to 95 percentile) 0.898 (0.514) -2LL 6104.855 5216.879
Cox and Snell R Square 0.559 0.574
Nagelkerke R Square 0.749 0.770
N 10184 10184
Note: binary logistic regression coefficients with standard errors in brackets. *** p < 0,001, **p < 0,01, *p < 0,05
In Model 1 we can observe a couple of things about our predictors; racial resentment and economic dissatisfaction. The Ex(B) tells us the change in odds resulting from a one unit change in the predictor (Field, 2009, p. 270). As the predictor increases, the odds of the outcome occurring increases. As the level of racial resentment increases, the odds of an individual voting for a Republican is 2.297 times higher than someone with very low racial attitudes. Attitudes on how well a respondent thinks the economy is doing, is also a statistically significant predictor for vote choice. As individuals become less satisfied with the economy, the odds of that individual voting for a Republican is 1.280 times higher than someone that is more satisfied with the economy. Both limits of the confidence intervals for racial resentment and economic dissatisfaction are above 1, which gives us confidence that the direction of the relationship we have observed, racial resentment increases the odds of voting for a Republican candidate, is true for our population. (CI= 2.127 & 2.481). The Wald Statistic tells us whether the predictor variable is making a significant contribution to the prediction of the outcome (Field, 2009, p. 287). For both predictors the Wald Statistic is significantly different from zero (Race: Wald= 449.737; p<0.001, Economy: Wald=66.196; p<0.001).
In Model 2 we added all of the controlled variables; age, gender, education, ideology, income and partisanship. Here the Ex(B) coefficients decline a little except for economic dissatisfaction; here the odds increase. When looking at the -2LL we notice that in Model 2 this value is significantly smaller (-2LL= 5216.879) than in Model 1 (-2LL= 6104.855). This means that the model is more accurate when adding the controlled variables. The ‘b’ value represents the change in the logit of the outcome variable associated with a one-unit change in the predictor variable (Field, 2009, p. 286). This means that when racial attitudes increase by one unit, the logit of vote choice increases by 0.832 (p<0.001). When feelings of economic dissatisfaction increase by one unit, the logit of vote choice decrease by 0.247 (p<0.001). In Model 2, the Wald Statistic is significantly different from zero for racial resentment and economic dissatisfaction (Race: Wald= 83.952; p<0.001, Economy: Wald=25.320).
It was to be expected that partisanship would have a significant effect on vote choice, since when looking at the data, about 85% of Republicans said they voted for Trump and 87% of Democrats said they voted for Clinton. In order to determine what the effect is of our moderator variable, partisanship, on the effect of racial resentment on vote choice, we must create an interaction variable between our moderator and predictor variable. In our logistical regression
model the Ex(B) is not statistically significant (Ex(B)= 0.818, p=0.445). This means that according to our model, the level of partisanship does not increase the effect that racial resentment has on vote choice.
Racial Resentment and Donald Trump
To see if the victory of Donald Trump was something unique in terms of the effect of racial resentment , we must compare the presidential elections between 1992 and 2016 (see figure 7). When creating 6 separate models for each 4-year term of Presidential Elections we can see a pattern occurring in the odds ratio. There is a gradual increase in the odds ratio for racial attitudes. This means that across time, when having information about a respondent’s racial attitudes, the odds we can correctly predict that they voted for a Democrat/Republican increases. When one’s racial attitudes worsen, the odds that they voted for a Republican increase. In terms of economic dissatisfaction the pattern is harder to find; there is no gradual increase or decrease. What we can see is that in 2012 the odds ratio for economic satisfaction is almost five times higher than in 2008. An explanation for this could be the financial crisis of 2008 where almost every U.S. citizen was affected “either through investment and housing price declines, or adverse changes in the labour market” (Malhotra & Margalit, 2010, p. 853). Because of the bad economic environment caused by the financial crisis, it is understandable that individuals were unsatisfied with the economic situation.
Figure 7
Odds Ratio of Voting for Democrat/Republican between 1992 and 2016 (Ex(B)).
0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 1992 2000 2004 2008 2012 2016
Odds Ratio of Voting for Democrat/Republican between
1992 and 2016
Another way we can see whether the effect of racial resentment on the voting was unique for Donald Trump is by creating an interaction variable of year x racial attitudes. This allows us to determine whether racial resentment is significantly more related to voting in 2016 than in 1992. The summary of the logistical model including this interaction is shown below.
Table 4
Logistic Regression analysis of the probability of voting Democrat or Republican (Ex(B) scores)
Model 2 Model 3
Level of Racial Attitudes 2.393*** 2.217***
(0.042) (0.052) Economic Dissatisfaction 1.356*** 1.571*** (0.032) (0.037) Partisanship 2.933*** 2.911*** (0.022) (0.022) Ideology 0.740*** 0.744*** (0.027) (0.028) Age 1.014*** 1.012*** (0.002) (0.002) Education 1.005 0.973 (0.048) (0.049) Gender (Male) 79043266.1 111564053 (16184.956) (15957.938) Income (0-16 percentile) 0.701 0.669* (0.187) (0.189) Income (17-33 percentile) 0.909 0.852 (0.181) (0.182) Income (34-67 percentile) 0.836 0.825 (0.164) (0.165) Income (68 to 95 percentile) 0.898 0.846 (0.514) (0.165) Racex2016 1.237*** (0.039)
Racex2012 1.064 (0.035) Racex2008 0.869*** (0.039) Racex2004 1.191*** (0.047) Racex2000 1.212*** (0.047) Racex1992 -2LL 5216.879 5112.017
Cox and Snell R Square 0.574 0.578
Nagelkerke R Square 0.770 0.776
N 10184 10184
Note: binary logistic regression coefficients with standard errors in brackets. *** p < 0,001, **p < 0,01, *p < 0,05
The coefficients of Ex(B) for the interaction between year and racial attitudes are positive and statistically significant for the years 2016, 2008, 2004 and 2000. The Racial Resentment x2016 coefficient is positive and statistically significant (Ex(B)= 1.237, p<0.001), indicating that its effect is greater than in the previous years. The odds coefficient is not statistically significant for 2012, but it is for 2008. The odds coefficient is substantially greater for the 2016 Presidential election than for the 2008 Presidential Elections, indicating that race played a greater role in Trump’s victory.
Discussion & Conclusion
The aim of this paper was to determine whether the role of racism on voting behavior has had a significantly higher effect in the 2016 American Presidential Elections, or if this effect has been gradually increasing over time. By performing a logistic regression analysis measuring the odds that an individual voted for a Republican when we know where they were place on the racial resentment scale. This odds coefficient was done for the years 1992-2016.
From the results we can conclude that that high racial resentment levels has an effect on vote choice. When looking at the coefficients between 1992 and 2016, we see a gradual increase in the effect of racial attitudes on vote choice, but not abnormally high for 2016 when Trump was elected. This could mean that overall racial issues are becoming more salient in American Politics, and Trump was able to make use of this environment and also make racial issues important in his Presidential Campaign. However, the years 2000 and 2016 are statistically different in terms of the effect of racial resentment on vote choice.
My first hypothesis was that racial resentment increases the probability of an individual having voted for a Republican. The logistic regression shows that the odds coefficient is positive and is statistically significant, meaning that when racial attitudes are present, the odds that someone voted for a Republican candidate is higher than having voted for a Democratic candidate. Schaffner et al. (2018), Dweyer et al. (2009) and Knuckey (2011) also concluded that attitudes on race are powerful forces for driving vote choice. Additionally, racial attitudes were especially important in the 2016 Presidential Elections, especially amongst whites without a college degree (Schaffner et al., 2018, p. 30).
My second hypothesis was that the effect of racial attitudes on vote choice would be significantly stronger for Trump than any other presidential candidate since 1992 due to Racial Priming. The effect of racial resentment on vote choice is significantly higher in 2016 than the year 2000. Schaffner et al. (2018), find that racial appeals appeared to have won Trump even more support, as Valentino et al. (2018) found, politicians are no longer hindered when making racial appeals in campaigns (as cited in Schaffner et al., 2018, p. 31). After a Black President that served for two terms, race was definitely a salient issue. We would have to look at this effect over a larger period of time in order to really determine whether 2016 was a unique year for racial issues.
My findings, showing that there is a positive significant effect between racial resentment and vote choice, coincide with the articles of Schaffner et al. (2018), Dweyer et al. (2009) and Knuckey (2011). What I add to this debate is that racial resentment played an especially important role in the victory of Donald Trump. The effect of racial resentment on vote choice increases over time, indicating a linear trend, which coincides with the conclusions of Enders & Scott (2019, p 298). It is evident that racial resentment plays a large role in voting behavior
in the U.S., which Donald Trump was able to use to his advantage in the 2016 Presidential Elections.
Throughout the analysis I experienced some methodological issues. The first was missing data. Even though the ANES cumulative file includes a vast amount of data on political behavior in the United States, attitudes on racial issues was not complete. Data was missing from the year 1996, which impacts the accuracy of the generalizations taken from the results. Additionally, we could not go further back the 1992, because there was insufficient data on racial resentment items. The year 1992 also did not have enough data to create an interaction variable between time and racial attitudes; meaning we could not determine whether there was a significant difference in racial resentment between the year 1992 and 2016, and thus had to start at the year 2000; making our conclusions less reliable.
Another limitation is the time frame; since I am starting at the year 1992, I am not fully able to determine the effect of our moderator ‘partisanship’ on the relationship between racist attitudes on vote choice. This is because since the 1960s, political parties became polarized on issues of race, which means that in order to see the full effect we would have to go back to before this occurred; which is not possible with the data we have. Furthermore, it is difficult to find the exact role of partisanship in the relationship between racism and voting behavior. When a Republican voter voted for Trump, and we can see that their level of racial resentment is high, we cannot assume that their views on racism are the only reason they voted for Trump. One might have agreed with many of his other policies, or rather one could have low racial attitudes and have voted for Trump because they are a strong Republican and identity with the other core values of the Republican party. A way we could better determine the role of partisanship is by looking at Obama-Trump voters and Romney-Clinton voters. If an Obama-Trump voter has high racial attitudes, we might assume that this individual voted for Trump because of high racial resentment. Unfortunately this was not possible to investigate with my current data set, but is definitely a relationship worth considering for further research.
Choosing the year 1992 as my starting point for my analysis also provides us with some implications. We are only making conclusions based on the last 28 years however the change in American politics including the tolerance towards racial rhetoric is a trend that has been occurring since the 1930s.
A final issue that I experienced with my data set is the use of my controlled variables. I did not choose to control for race because the type of racial resentment that is of importance in my analysis, mostly comes from ‘white’ Americans. Perhaps it could have improved the results if I controlled for this variable by adding the race of the individual into my model.
The relationship between political parties and social identities have been growing stronger, which has resulted in individuals clinging more to their partisan groups (Bafumi & Shapiro, 2009, p. 1). Racial attitudes stem from Americans predispositions that have resulted from a long process of socialization. Racial views have never really left, instead they evolved into ‘symbolic racism’. According to my results, a high racial resentment level has a significant effect on voting behavior in the 2016 Presidential Elections. However, what is missing from the analysis is whether racial attitudes have pushed Obama voters to vote for Trump. Finding the answer for this question could give us a better understanding of the extent that racism had a role in the Presidential Elections in 2016.
References
Abramowitz, A.I. (1994). Issue Evolution Reconsidered: Racial Attitudes and Partisanship in the U.S. Electorate. American Journal of Political Science, 38(1), 1-24
American National Election Studies. Time Series Cumulative Date File [Data file] Retrieved from: https://electionstudies.org/data-center/anes-time-series-cumulative-data-file/
Bafumi, J., & Shapirom, R.Y. (2009). A New Partisan Voter. The Journal of Politics , 71(1), 1-24
Beaumont, A. (2016, December 17). US 2016 Election Final Results: How Trump Won. The Conversation. Retrieved from http://www.theconversation.com
Bracic, A., Israel-Trummel & M., Shortle, A.F. (2019). Is Sexism for White People? Gender Stereotypes, Race, and the 2016 Presidential Election. Political Behaviour, 41, 281-307.
Blankenship, B.T., & Stewart, A.J. (2019). Threat, Trust, and Trump: Identity and Voting in the 2016 Presidential Election. Politics, Groups and Identities, 7(3), 724-736
Crockett, Z. (2017, January 25). Donald Trump is the Only US President Ever With No Political or Military Experience. Vox. Retrieved from https://www.vox.com/
Doherty, C. (2017, October 5). Key Takeaways on Americans’ Growing Partisan Divide Over Political Values. Pew Research Center. Retrieved from https://www.pewresearch.org/
Dwyer, C.E., Stevens, D., Sullivan, J.L., & Allen, B. (2009). Racism, Sexism and Candidate Evaluations in the 2008 U.S. Presidential Election. Analyses of Social Issues and Public Policy, 9(1), 223-240
Enders, A.M. & Scott, J.S. (2019). The Increasing Racialization of American Electoral Politics, 1988-2016). American Politics Research, 47(2), 275-303
Field, A. (2009). Discovering Statistics Using SPSS: Third Edition. London: Sage Publications. Fiorina, M.P., & Abrama, S.J. (2008). Political Polarization in the American Public. Annual Review of Political Science, 11, 563-588.
Graham, D.A., Green, A., Murphey, C., & Richards, P. (2019, June). An Oral History of Trump’s Bigotry. The Atlantic. Retrieved from: http://www.theatlantic.com
Godbout, J., & Bélanger, E. (2007). Economic Voting and Political Sophistication in the United States. Political Research Quarterly, 60(3), 541-554
Hutchings, V.L., & Jardina, A.E. (2009). Experiments on Racial Priming in Political Campaigns. Annual Review of Political Science, 12, 397-402.
Kam, C.D., & Burge, C.D. (2017). Uncovering Reactions to the Racial Resentment Scale Across the Racial Divide. The Journal of Politics, 80(1), 1-7.
Kinder, D.R., & Sears, D.O. (1981). Prejudice and Politics: Symbolic Racism Versus Racial Threats to the Good Life. Journal of Personality and Social Psychology, 40(3), 414-431
Knuckey, J. (2011). Racial Resentment and Vote Choice in the 2008 Presidential Election. Politics & Policy, 39(4), 559-582.
Krieg, G. (2016, November 10). How Did Trump Win? Here are 24 Theories. CNN. Retrieved from http://www.edition.cnn.com
Lachat, R. (2008). The Impact of Party Polarization on Ideological Voting. Electoral Studies, 27, 687-698.
Lelkes, Y. (2016). Mass Polarization: Manifestations and Measurements. Public Opinion Quarterly, 80, 392-410.
Luttig, M.D., Federico, C.M., & Lavine, H. (2017). Supporters and Opponents of Donald Trump Respond Differently to Racial Cues: An Experimental Analysis. Research & Politics, 4(4), 1-8
Lopez, G. (2019, July 15). Donald Trump’s Long History of Racism, from the 1970s to 2019. Vox. Retrieved from: http://www.vox.com
Major, B., Blodorn, A., & Blascovich, G.M. (2018). The Threat of Increasing Diversity: Why Many White Americans Support Trump in the 2016 Election. Group Processes & Intergroup Relations, 21(6), 931-940.
Malhotra, N., & Margalit, Y. (2010). Short-Term Communication Effects or Longstanding Dispositions? The Public’s Response to the Financial Crisis of 2008. The Journal of Politics, 72(3), 852-867
Markus, G.B. (1988). The Impact of Personal and National Economic Conditions on the Presidential Vote: A Pooled Cross-Sectional Analysis. American Journal of Political Science, 32(1), 137-154
Mason, L. (2015). “I Disrespectfully Agree”: The Differential Effects of Partisan Sorting on Social and Issue Polarization. American Journal of Political Science, 59(1), 128-145. Mason, L., & Wronski, J. (2018). One Tribe to Bind Them All: How Our Social Group
Attachments Strengthen Partisanship. Advances in Political Psychology 39(1), 257-277.
Mason, L. (2018). Losing Common Ground: Social Sorting and Polarization. Forum : Majalah Pengembangan Ilmu-ilmu Social, 16(1), 47-66.
Mullinix, K.J. (2016). Partisanship and Preferences Formation: Competing Motivations, Elite Polarization, and Issue Importance. Political Behavior, 38, 383-411.
Mutz, D.C. (2018). Status Threat, Not Economic Hardship, Explains the 2016 Presidential Vote. Proceedings of National Academy of Sciences of the United States of America, 115(19)
Schaffner, B., Macwilliams, M., & Nteta, T. (2018). Understanding White Polarization in the 2016 Vote for President: The Sobering Role of Racism and Sexism. Political Science Quarterly, 133(1), 9-34.
Sides, J., Tesler, M., & Vavreck, L. (2017). How Trump Lost and Won. Journal of Democracy, 28(2), 34-44.
Sides, J., Tesler, M., & Vavreck, L. (2018). Hunting Where the Ducks Are: Activating Support for Donald Trump in the 2016 Republican Primary. Journal of Elections, Public Opinion and Parties, 28(2), 135-156.
Smith, D. (2016, November 9). How Trump won the election: volatility and a common touch. The Guardian. Retrieved from: http://www.theguardian.com
Sears, D.O., van Laar, C., Carrillo, M., & Kosterman, R. (1997). Is it Really Racism? The Origings of White Americans’ Opposition to Race-Targeted Policies. Public Opinion Quarterly, 61, 16-53
Swim, J.K., Aikin, K.J., Hall, W.S., & Hunter, B.A. (1995). Sexism and Racism: Old Fashioned and Modern Prejudices. Journal of Personality and Social Psychology, 68(2), 199-214 Tesler, M. (2012). The Spillover of Racialization into Health Care: How President Obama Polarized Public Opinion by Racial Attitudes and Race. American Journal of Political Science, 56(3), 690-704.
Tesler, M. (2013). The Return of Old-Fashioned Racism to White Americans’ Partisan Preferences in the Early Obama Era. The Journal of Politics, 75(1), 110-123.
Valentino, N.A., Neuner, F.G., & Vandenbroek, L.M. (2017). The Changing Norms of Racial Political Rhetoric and the End of Racial Priming. The Journal of Politics, 80(3), 1-15. Weinschenk, A.C. (2013). Partisanship and Voting Behavior: An Update. Presidential Studies
Quarterly, 43(3), p. 607-617.
Westfall, J., Van Boven, L., Chambers, J.R., & Judd, C.M. (2015). Perceiving Political Polarization in the United States: Party Identity Strength and Attitude Extremity Exacerbate the Perceived Partisan Divide. Perspectives on Psychological Science, 10(2), 145-158.
Whitehead, A.L., Perry, S.L., & Baker, J.O. (2018). Make America Christian Again: Christian Nationalism and Voting for Donald Trump in the 2016 Presidential Election. Sociology of Religion: A Quarterly Review, 79(2), 147-171.
Wilson, D.C., & Davis, D.W. (2011). Reexamining Racial Resentment: Conceptualization and Content. The Annals of the American Academy of Political and Social Science, 634(1), 117-133.
Wlezien, C., Franklin, M., & Twiggs, D. (1997). Economic Perceptions and Vote Choice: Disentangling the Endogeneity. Political Behaviour, 19(1), 7-17
Zurcher, A. (2016, November 9). US Election 2016 Results: Five reasons Donald Trump won. BBC News. Retrieved from: http://www.bbc.com
Appendix
List of Controlled Variables
Variable Scale Age 1= 17-24 2= 25-34 3= 35-44 4= 45-54 5= 55- 64 6=65-74 7=75-99 and over Gender 1=Female 2=Male Income 1= 0 to 16 percentile 2= 17-33 percentile 3= 34-67 percentile 5= 68- 95 percentile 6= 96-100 percentile
Education 1=Grade school
2= High school
3= Some college, no degree 4= College/advanced degree
Ideology Scale of 1-7 where:
1= extremely liberal 7= extremely conservative
