Explaining White Polarization in the 2016 Vote for President:

The Sobering Role of Racism and Sexism

Brian F. Schaffner
University of Massachusetts Amherst
[email protected]
Matthew MacWilliams
MacWilliams Sanders Communication
Tatishe Nteta
University of Massachusetts Amherst

Paper prepared for presentation at the Conference on The U.S. Elections of 2016: Domestic and
International Aspects. January 8-9, 2017, IDC Herzliya Campus.

The 2016 presidential campaign featured candidates who explicitly put issues of race and
gender at the forefront of the discourse. Notably, 2016 also witnessed the largest gap between the
presidential vote preferences of college- and non-college educated whites since at least 1980.
While Trump enjoyed just a four-point margin over Clinton among whites with a college degree
(10 points smaller than Romneys margin over Obama among that group in 2012), his advantage
among non-college educated whites was nearly 40 points. This gap between college- and noncollege educated whites was possibly the single most uniquely important divide documented in
2016.
While many election post-mortems were quick to make note of the education gap among
whites in terms of presidential vote choice in 2016, explanations for the origins of this gap were
a subject of greater debate. Two prominent explanations have been offered. The first is that white
working class Americans have been left behind during the economic recovery that took place
during the Obama presidency. Trumps populist economic message, focusing on protectionism
and other policies to help working people, resonated with this group. A second explanation is
that Trumps willingness to make explicitly racist and sexist appeals during the campaign,
coupled with the presence of an African American president and the first major party female
nominee, made racism and sexism a dividing line in the vote in this election. This led less
educated whites, who tend to exhibit higher levels of sexism and racism, to support Trump, while
more educated whites were more supportive of Clinton.
In this paper, we use data from a national survey conducted during the final week of
October to adjudicate between these explanations. Using unique measures of attitudes on racism
and sexism, coupled with a question to tap into dissatisfaction with economic conditions, we are
able to determine to what extent each of these explanations helped to explain vote choices in

2016 and, ultimately, whether either of these explanations can explain the education gap in vote
choices among whites. We find that while economic dissatisfaction was part of the story, racism
and sexism were much more important and can explain about two-thirds of the education gap
among whites in the 2016 presidential vote.

Explaining the Education Divide Among Whites

Figure 1 plots the Republican share of the two-party vote for president among whites
with and without a college degree in each presidential election since 1980. These data are from
the national exit polls, as compiled by the New York Times. Note that white voting behavior in
presidential elections was rather similar from 1980 through 1996, regardless of education.
Indeed, in none of these elections was there more than a five percentage point difference in how
college and non-college educated whites voted. In 2000, a small but notable gap did begin to
emerge, with non-college whites providing more support for the Republican presidential
nominee. This gap remained relatively small, ranging from 5 to 7 points in the elections held
from 2000 to 2012.
In 2016, however, the gap in vote preferences between college and non-college whites
widened considerably to 18 points, nearly three-times larger than it had been in any election
during the series. Importantly, this 18-point gap resulted from an apparent polarization among
whites; college-educated whites became more supportive of Clinton than they had been for
Obama in 2012, while whites without a college degree moved even more dramatically toward
Trump. In fact, Trump won over 70% of the two-party vote among whites without a college
degree, which easily exceeded the performance of any Republican going back to at least 1980.
His success with this group was particularly important in the three states that ultimately decided

Figure 1: Percent of two-party vote for the Republican presidential candidate among
whites with and without a college degree, 1980-2016

Source: National exit polls. Accessed at:

http://www.nytimes.com/interactive/2016/11/08/us/politics/election-exit-polls.html

the election Wisconsin, Michigan, and Pennsylvania. In those three states, whites without a
college degree made up between 40 and 47% of the electorate and in each state they favored
Trump by about 30-points over Clinton.
The emergence of the education gap in vote choice among whites after the turn of the
century sparked attention to understanding why working class whites were abandoning the
Democratic Party. In the most prominent example of work directed at explaining this gap,
Thomas Frank (2005) argued that cultural wedge issues were causing white working class voters
to abandon the Democratic Party. Generally speaking, the cultural wedge issues that Frank
pointed to included abortion, gun control, gay marriage, and the death penalty. Frank argued that

such issues repelled white working class voters from the Democratic Party, which would be a
more natural match for that group if their attention was focused on economic concerns.
In a response to Franks book, Bartels noted that the partisan significance of educational
attainment has largely evaporated (2006, p. 209). Bartels also convincingly demonstrated that
economic issues were still a primary determinant of voting behavior in the 2004 presidential
election, casting doubt on the claims made by Frank regarding the importance of cultural wedge
issues. While other scholars found more support for Franks thesis about a growing divide
among the white working class (e.g. Brady et al. 2008), Bartelss argument that (1) the education
divide among whites was not particularly pronounced in 2004 and (2) cultural issues were not
particularly influential in driving voters away from the Democratic Party, was convincing.
But as Figure 1 shows, if 2004 did not bring us a particularly large split in the voting
preferences of whites based on education, 2016 clearly did. Based on Bartelss analysis of ANES
data going back to the 1952 presidential election, the 18-point gap in the vote choices of college
whites compared to non-college whites would be the largest such gap since 1964. However, in
1964, that gap was reversed, with non-college whites voting Democratic at a much higher rate
than college whites. As Seth Masket wrote shortly after the election, Franks book explains the
2016 election far better than it did the election cycle in which it was published.
But why did education emerge as such a dividing line among whites in 2016? On one
hand, it may be that the second part of Bartelss analysis remains correct even in 2016. That is,
economic issues may still be the most important determinant of vote decisions, but economics
may simply be dividing whites along education lines more powerfully than they have in past
election cycles. Indeed, Trump ran an especially populist campaign for a Republican nominee,
focusing on protectionist positions on trade issues while generally refusing to call for cuts to

popular government programs like Medicare and Social Security. With some analyses indicating
that working class whites saw the least amount of benefit from the economic recovery (e.g.
Porter 2016), it may very well be the case that this group voted decisively for the populist
nominee of the out-party in 2016.
On the other hand, the explicit nature of the campaign rhetoric on race and sex in 2016
may be the culprit for the education gap among whites. With regard to racism, Trumps rhetoric
frequently violated norms that were supposed to inhibit politicians from making explicitly racist
appeals. Specifically, one of the core tenants of the implicit/explicit model of racial priming is
the expectation that racial appeals will be rejected by the mass public and will, therefore, be
ineffective (Mendelberg 2001; Valentino et al. 2002). Yet, Valentino et al. (2016) find that the
norms of racial political rhetoric have been shifting in recent years. Through a series of survey
experiments they find that Whites now view themselves as an embattled racial group, and this
has led to both strong ingroup identity and a greater tolerance for expressions of hostility toward
outgroups (2016, p. 28).
Thus, by 2016, it may have been possible for a candidate like Trump to make explicitly
racist appeals to whites without undermining the effectiveness of those appeals in winning over
voters with more racist attitudes. Importantly, education has been found to be related to views on
race; whites with less education generally are less tolerant of other racial/ethnic groups and tend
to exhibit more conservative racial attitudes than those with more education (Bobo and Licari
1989; Sniderman and Piazza 1993; Schuman et al. 1997). Thus, if Trumps racial rhetoric was
effective, it was most likely to win him votes among less educated whites.
Of course, Trumps rhetoric went far beyond targeting racial and ethnic groups; he also
invoked language that was explicitly hostile towards women. These remarks were often focused

directly at opponents, such as Carly Fiorina and Hillary Clinton, or news reporters, such as
Megyn Kelly. Adding to the litany of sexist remarks he had made during and before the
campaign was the release of the Access Hollywood tape, which made major news about a month
before Election Day, and caused many Republicans to withdraw their support of him.
Such rhetoric was likely all the more salient given the presence of the first female major
party nominee for president in the race. Scholarship on the role of sexism and gender stereotypes
on vote decisions involving women candidates is mixed. While many studies find that women
candidates do not suffer a penalty from voters (e.g. Claassen and Ryan 2016; Dolan 2014; Hayes
2011; Pearson and McGhee 2013), other work has pointed to important challenges faced by
women when they run (e.g. Huddy and Terklidsen 1993; Bauer 2016; Streb et al. 2008). Bos et
al. point to the importance of role incongruity theory (RCT) for understanding when a female
candidates gender may become salient to voters during a campaign. Specifically, RCT is based
on the notion that people tend to think that women should behave, but that political leaders ought
to be assertive and independent. It may be the case that when a campaign highlights the way in
which a female candidate is behaving incongruously, attitudes on sexism may become a stronger
predictor of vote choice. As Bos et al. note, Prejudice against female candidates is likely to
occur when context favors male stereotypical strengths, highlighting womens poor fit with the
leader role. Prejudice should be reduced when the context favors female stereotypical strengths,
such as cooperation and flexibility (2017, p. 18). Thus, for example, when Trump referred to
Clinton as a nasty woman during a debate, the reaction from voters may have been conditioned
by their underlying views about how women should behave. For those with more sexist views,

Trumps remark may have drawn attention to the fact that Clinton was not acting in the
stereotypical way that they expect from a woman.1
Thus, while it is certainly possible that economic dissatisfaction was largely responsible
for the education gap among whites in 2016, there is even more reason to expect that racism and
sexism may be predominantly behind this gap. If racism and sexism are associated with support
for Trump in 2016, and if non-college whites are more likely to hold racist and sexist views, then
the uniquely explicit role of racism and sexism in the 2016 campaign may account for the
education gap among whites. The education-based polarization in vote choice evidenced in 2016
may have resulted from college educated whites, who are more likely to reject racist and sexist
rhetoric, turning away from Trump while non-college educated whites, who may be more likely
to embrace such rhetoric, voting for him.

Data
To test whether economic dissatisfaction or racist/sexist attitudes explain the education
gap among whites, we analyze a nationally representative survey of American adults
administered online by YouGov, from October 25th October 31st, 2016. YouGov uses a
matched sampling approach, which begins with a randomly selected target sample taken from the
2010 American Community Survey. YouGov then matched respondents from their volunteer
panel on a variety of characteristics including gender, age, race, education, party identification,

1 Of course, in our discussion of Trumps explicit rhetoric on race and sex, it is important to make note of the fact
that Clinton did not shy away from these issues during the campaign either. As Tessler notes, Hillary Clinton
moved to the left of Obama in both her rhetoric and policies on race-related issues in order to retain support from a
coalition increasingly comprised of minorities and racially progressive whites (2016). She also explicitly
campaigned on her intention to become the first woman president. Thus, to the extent that Trumps rhetoric primed
racism and sexism during the campaign, Clintons campaign did little to avoid those topics.

ideology, and political interest. The survey included interviews with 2,000 American adults, with
an oversample of African Americans and Latinos to ensure at least 400 respondents in each of
those groups. Propensity score weights accounted for the minority oversamples and were also
designed to ensure that the sample was representative of the adult population on age, gender,
race/ethnicity, education, ideology, and region.
The survey questionnaire began by asking respondents whether they intended to vote in
the November election and then asked which candidate they intended to vote for. A follow-up
question was asked of those who said they were not sure who they would vote for to determine
whether they leaned toward voting for a particular candidate. If we limit our analysis to
individuals who said they would definitely vote or had already voted, the survey showed Clinton
with a 3-percentage point lead over Trump (46% to 43%). This margin is close to the 2.1 points
by which Clinton actually won the national popular vote.
Our primary dependent variable is the two-party vote for president. This variable includes
people who said that they were leaning toward one of the two major candidates. Individuals who
chose a candidate other than Clinton or Trump are excluded from our analysis. We also restrict
our analysis only to likely voters, defined as those who said they would definitely vote or who
had already voted. However, extending our scope even to those who were not likely voters does
not alter the conclusions reported below.
Our two primary independent variables are measures of attitudes regarding sexism and
racism. For sexism attitudes, we create a scale from four items taken from the hostile sexism
battery (Glick and Fiske 1996). The hostile sexism battery is part of the Ambivalent Sexism
Inventory and is designed to measure prejudiced attitudes toward women. The four items we use
from this scale are:

1. Women are too easily offended.

2. Many women are actually seeking special favors, such as hiring policies that favor them
over men, under the guise of asking for "equality."
3. Women seek to gain power by getting control over men.
4. When women lose to men in a fair competition, they typically complain about being
discriminated against.
Respondents were asked to indicate their agreement or disagreement with these items on a fivepoint scale. We then scaled these four items using an IRT graded response model, which resulted
in a single standardized variable for hostile sexism, with a mean of 0 and a standard deviation of
1.
To measure racism, we use three items that capture the extent to which an individual
acknowledges and empathizes with racism. These items are related first and foremost to the
concept of color-blind racial attitudes. As Neville et al. (2000, p. 60) explain, color-blind racial
attitudes refers to the belief that race should not and does not matter. People who hold such
attitudes essentially do not acknowledge the existence of racism in the United States. Thus, the
two items we use from the CoBRAS scale developed by Neville et al. are:
1. White people in the U.S. have certain advantages because of the color of their skin.
2. Racial problems in the U.S. are rare, isolated situations.
As DeSante and Smith (2016) note, the CoBRAS items are useful at tapping the cognitive
awareness or acknowledgement of racism in America, but additional items are needed to
measure the extent to which people feel empathetic about the costs of racism. Thus, based on the
advice offered by DeSante and Smith, we add an additional item from the Psycho-social Costs of

10

Racism to Whites (PCRW) battery (Spanierman and Heppner 2004, Spanierman, et al. 2006,
Poteat and Spanierman 2008):
3. I am angry that racism exists.
For each of these three items, respondents indicated on a six-point scale the strength with which
they agreed or disagreed with each statement. These three questions were then scaled using an
IRT graded response model to create a single racism measure on a standardized scale.
Figure 2 shows the distribution of responses on each of these two items, first for all likely
voters (top row) and then just for likely voters who are white (bottom row). Note that about 15
Figure 2: Distribution of likely voters on sexism and racism scales
All likely voters

White likely voters only

11

percent of respondents take the least sexist position on the four items. However, the remaining
85 percent of respondents are distributed fairly evenly across the distribution of values. This
distribution looks quite similar when we restrict the analysis only to whites (bottom-left plot). A
somewhat similar pattern exists with the racism scale, with over 10 percent of all likely voters
taking the least racist positions on the three items, with the remaining adults distributed across
the spectrum of racism acknowledgement. Notably, the distribution for white likely voters looks
somewhat different, with fewer whites (about 8%) providing the least racist responses to the
items.
The correlation between the two scales is .49 among likely voters and .60 among whites.
Thus, individuals who score higher on the racism battery are also more likely to score higher on
the sexism battery, but the scales are conceptually and statistically distinct.
In addition to these scales for racism and sexism, we also test for the role of economic
satisfaction in affecting vote choice. The survey included an item asking, All things considered,
how satisfied are you with your overall economic situation? Respondents could choose from
five options ranging from extremely satisfied to not satisfied at all. Figure 3 shows the
distribution of responses from likely voters to this question. Nearly one-quarter of likely voters
reported that they were not satisfied at all with their economic situation, while very few reported
that they were extremely satisfied. Notably, responses to this question are not strongly related to
the racism or sexism scales both scales correlated with the economic dissatisfaction item at just
.10.

12

Figure 3: Distribution of responses on question asking about satisfaction with personal

economic situation

For any of these three variables to explain the education gap in vote choice among
whites, we would expect to find that non-college whites would score higher on these items than
those with a college degree. That is, we expect that non-college whites would be more racist,
more sexist, and more dissatisfied with their economic situation. Table 1 presents the average
value for non-college whites and college whites on each of these three measures. The pattern is
consistent across the three items whites without a college degree expressed more economic
dissatisfaction and scored higher on the racism and sexism scales. Specifically, on the question
about economic dissatisfaction, whites without a college degree were more than a half-point less
satisfied with their economic conditions on the five-point scale. And on the racism and sexism
scales, non-college whites scored about one-quarter of a standard deviation more racist/sexist

13

than whites with a college degree. These differences are all statistically significant and the
magnitude of these differences is non-trivial.
Table 1: Average values of measures of economic dissatisfaction, racism, and sexism among
whites by education
Measure
Whites with a college
Whites without a
Difference
degree
college degree
Economic
3.06
3.64
0.58
dissatisfaction
(0.07)
(0.06)
(p<.001)
Racism scale
0.06
0.30
0.24
(0.04)
(0.04)
(p<.001)
Sexism scale
-0.11
0.18
0.29
(0.05)
(0.04)
(p<.001)
Note: Entries are means. Standard errors in parentheses.

In the analysis of presidential vote choice that follows, we re-scale each of these three
variables so that they range from 0 to 1. In addition to these three independent variables, we also
include several control variables including partisanship (on the 7-point scale), ideology (5-point
scale), gender, age, education, income, and race. We re-scaled the partisanship and ideology
measures so that they also range from 0 to 1, and the remaining control variables are simply
incorporated as dummy variables for each relevant category. Finally, all of our analyses
incorporate sampling weights to ensure that our results are generalizable to the population of
likely voters.

Sexism, Racism, Economic Dissatisfaction and Voting for Trump

Table 2 presents the results from two vote choice models estimated using probit. The first
model includes all likely voters in our survey while the second model limits the analysis only to
white likely voters. Notably, the coefficients for the variables are relatively similar across both
models. While it is certainly true that whites score higher on the racism scale than blacks and

14

Latinos, racism does not operate much differently among whites in predicting support for Trump
than it does for those minority groups. It is also worth noting that since we have re-scaled each of
these variables from 0 to 1, the coefficients are somewhat comparable. Of particular note is the
fact that the coefficients for the racism and sexism scales are quite large, exceeded only by
partisanship in terms of their strength of association with support for Trump.
Table 2: Probit estimates of factors affecting two-party vote for Trump
All likely voters
Hostile sexism scale

White likely voters

2.969
3.293
(0.408)**
(0.606)**
Racism scale
2.913
3.023
(0.585)**
(0.839)**
Economic dissatisfaction
1.527
1.670
(0.291)**
(0.403)**
Ideology
2.038
2.499
(0.435)**
(0.699)**
7 point Party ID
3.858
4.158
(0.330)**
(0.471)**
Female
0.538
0.645
(0.187)**
(0.249)**
Age 30-54
1.561
2.131
(0.427)**
(0.562)**
Age 55+
1.590
1.991
(0.432)**
(0.544)**
College degree
-0.003
-0.159
(0.195)
(0.253)
Income <$40k
-0.314
-0.508
(0.264)
(0.359)
Income $40k - $100k
-0.215
-0.435
(0.255)
(0.348)
Income >$100k
-0.765
-0.924
(0.307)*
(0.429)*
White
-0.377
(0.348)
Black
-0.620
(0.396)
Hispanic
-0.713
(0.389)
Constant
-7.539
-8.840
(0.732)**
(1.038)**
N
1,304
721
Note: Entries are probit coefficients with standard errors in parentheses. * p<.05, ** p<.01.

15

Figure 3 uses the first model in Table 1 to plot the predicted probability of voting for
Trump across the range of the economic satisfaction scale, racism scale, and sexism scale, while
holding all other variables in the model at their mean values. This figure begins to answer the
question of whether support for Trump was more about economic dissatisfaction or attitudes on
race and gender. The first panel in the figure shows that economic dissatisfaction was clearly
associated with support for Trump. Moving from the highest to the lowest level of satisfaction
with ones personal economic situation increased the predicted probability of voting for Trump
by .13.
However, the effect of economic dissatisfaction is dwarfed by the relationship between
sexism and racism and voting for Trump. For example, an individual who was average on all
other variables in the model but registered the most sexist attitudes on the hostile sexism scale
Figure 3: Predicted probability of voting for Trump based on values of economic
dissatisfaction, racism, and sexism

Note: Predicted probabilities based on first model in Table 2 while holding all other variables
in model at their mean values. Vertical lines represent 95% confidence intervals.

16

had a .65 probability of voting for Trump. That same individual would have just a .35 predicted
probability of voting for Trump if she registered the least sexist attitudes. Thus, moving from one
end of the sexism scale to the other produced a 30-point increase in support for Trump among
the average likely voter. The effect for the racism scale was nearly identical moving from the
highest levels of acknowledgement and empathy for racism in American to the lowest levels was
associated with about a 30-point increase in support for Trump.
In contextualizing the strong effects for racism and sexism in the 2016 vote choice
model, it is important to keep several things in mind. First, the 30-point change in the probability
of voting for Trump as one moves from low to high levels of racism or sexism occur while
holding all other variables in the model at their mean values. This includes variables that are
highly predictive of vote choice, such as partisanship and ideology. Second, these effects hold
even when we attempted to control for other related concepts, such as authoritarianism and
populism (see the Appendix for this analysis).

Was racism and sexism uniquely important in 2016?

An important question regarding the powerful relationships between racism, sexism and
vote choice described above is whether those effects are unique to the 2016 presidential election,
or if this is simply the continuation of a trend in recent elections. Answering this question is not
entirely straight forward, since we are unaware of any previous election surveys that have
included measures of hostile sexism or the acknowledgement of racism scale that we analyze
here. With regard to the role of race, Michael Tessler (2016) has provided preliminary evidence
that the role of racism was stronger in 2016 than in the 2012 presidential election, but using

17

alternative measures of racism. In this paper, we provide additional suggestive evidence that
2016 was unique in this regard as well.
First, the national survey that provides the main source of data for this paper included
questions asking respondents to rate how favorable their views were regarding not only Trump
and Clinton, but also the major party nominees from the previous two election cycles (Barack
Obama, John McCain, and Mitt Romney). If views toward racism and sexism played a similar
role in those campaigns as it did in 2016, then we might expect to see similar patterns in how
sexism and racism are associated with favorability ratings for each of these candidates.
Alternatively, if we see that racism and sexism is associated with favorability ratings for the
2016 nominees but not for previous candidates, then this would be evidence that the 2016 vote
was uniquely affected by racism and sexism.
Figure 4 shows the relationship between a respondents value on the racism and sexism
scales and his predicted favorability rating for each politician. These predictions were generated
from identical models to those analyzed above, except in this case we use the favorability rating
as the dependent variable rather than vote choice and we use OLS as our estimator rather than
probit. The patterns in Figure 4 are quite striking, particularly for Republicans. Specifically, we
find no statistically significant relationship between either the racism or sexism scales and
favorability ratings of either John McCain or Mitt Romney. However, the pattern is quite strong
for favorability ratings of Donald Trump. In fact, people who score among the highest values of
racism or sexism rate Trump about twice as favorably as those with the lowest values on those
scales, holding all other variables in the model at their mean values. From this analysis, it
certainly appears as though support for the previous two Republican nominees was not affected
by racism and sexism in the same way that support for Trump was.

18

Figure 4: Predicted favorability ratings based on values of racism and sexism

Note: Predicted ratings while holding all other variables in model at their mean values.
Vertical lines represent 95% confidence intervals.

Interestingly, the patterns for Clinton and Obamas favorability ratings are quite similar
to each other. While Obama had higher favorability ratings than Clinton across the board, both
were related to sexism and racism in similar ways. Yet, the 2016 election featured the only

19

pairing of candidate whose favorability ratings were both affected by peoples levels of racism
and sexism.
As an additional test of whether 2016 was unique, we draw on a survey of likely voters
conducted in New Hampshire in October 2016. In that survey, we asked respondents not only
who they would support in the 2016 election, but we also asked them to recall who they voted
for in 2012. While the questionnaire for this survey did not include the racism battery or the
question related to economic dissatisfaction, it did include the same four hostile sexism items
from our national survey. Using this data, we estimated two vote choice models one for
whether the likely voter said they were going to vote for Trump or Clinton in 2016, and a second
for whether the likely voter said they voted for Romney or Obama in 2012. As with our other
models, we include controls for partisanship, ideology, gender, age, and education. We exclude a
control for race in this model since nearly all of the likely voters in New Hampshire were white.
Table 3 shows the results from the two vote choice models. We are particularly interested
in comparing the coefficients for the hostile sexism scale. While hostile sexism does have a
positive coefficient for the 2012 model, the coefficient is not statistically significant and the size
of the effect is less than one-fourth as large as what is in the 2016 vote choice model. A
difference of coefficients test indicates that we can be highly confident (p < .01) that the
relationship between sexism and vote choice in 2016 was larger than it was for the 2012 vote.
Figure 5 shows how the predicted probability of voting for Trump and Romney was affected by
increasing sexism while holding the other variables in the model at their mean values. Based on
this analysis, it does appear that sexism played a much more important role in affecting the 2016
vote than it did for 2012.

20

Table 3: Probit estimates of factors affecting two-party vote for Republican presidential
candidate in New Hampshire in 2012 and 2016
2012 Presidential Vote
Hostile sexism scale

2016 Presidential Vote

0.732
3.132
(0.486)
(0.637)**
Ideology
3.145
3.370
(0.689)**
(0.602)**
7 point Party ID
4.056
3.670
(0.599)**
(0.525)**
Female
-0.116
0.175
(0.198)
(0.204)
Age 30-54
0.241
-0.222
(0.426)
(0.452)
Age 55+
0.035
-0.188
(0.434)
(0.445)
College degree
-0.050
-0.277
(0.195)
(0.210)
Income <$40k
-0.116
0.558
(0.311)
(0.322)
Income $40k - $100k
-0.467
-0.059
(0.250)
(0.337)
Income >$100k
-0.244
0.228
(0.279)
(0.370)
Constant
-3.766
-5.061
(0.578)**
(0.611)**
N
554
551
Note: Entries are probit coefficients with standard errors in parentheses. * p<.05, ** p<.01.

21

Figure 5: Predicted probability of voting for Republican nominee in 2012 and 2016 based on
values of sexism

Note: Predicted probabilities based on the models in Table 3 while holding all other variables in
model at their mean values. Vertical lines represent 95% confidence intervals.
Ideally, we would have survey data from previous election cycles that would allow us to
make cross-election comparisons in terms of the importance of our racism and sexism scales on
the presidential vote. Nevertheless, the analysis of the survey data from New Hampshire and the
national data on favorability ratings for current and past nominees provides a relatively strong
indication that racism and sexism were more important in 2016 than they had been in previous
elections (see also Tessler 2016).

Can Sexism and Racism Explain the Education Gap?

So far, we have demonstrated that sexism and racism were strongly associated with
presidential vote choice in 2016. We have also provided some evidence to support the notion that

22

these associations were uniquely potent in 2016 compared to recent presidential elections. But
can racism and sexism help to explain the large gap in voting behavior between college and noncollege whites in 2016?
The 2016 exit polls found that 52% of the two-party vote among whites with at least a
college degree went to Trump, while Trump won 71% of the two-party vote among whites
without a college degree. This amounts to a 19-point gap in the vote choices of whites based on
education. In the pre-election survey that we analyze in this paper, we found a 22-point gap in
the vote choices of college and non-college educated whites. This gap is reflected by the
coefficient on gender in the first column of Table 4. This table presents a series of simple OLS
models for white likely voters in our sample. The aim is to examine how controlling for each of
our key variables might help to explain the education gap among whites. Thus, the greater a
reduction in the size of the coefficient for the college variable in a particular model, the more
those variables help to account for the gap.2
Table 4: The college gap among white likely voters in the two-party vote for
Controlling for
College degree

Base
Gap
-0.221
(0.042)**

Economic dissatisfaction

Econ.
Satisfaction
-0.176
(0.042)**
0.370
(0.078)**

Hostile sexism scale

Sexism
-0.099
(0.036)**

R2
N

Racism &
Sexism
-0.071
(0.031)*

1.481
(0.058)**
-0.053
(0.036)
0.42
800

0.641
(0.070)**
1.029
(0.087)**
-0.147
(0.034)
0.50
800

1.097
(0.054)**

Racism scale
Constant

Racism
-0.108
(0.032)**

0.636
(0.027)
0.04
800

0.394
(0.055)
0.09
796

0.115
(0.038)
0.37
800

Note: Entries are OLS coefficients with standard errors in parentheses. * p<.05, **p<.01.

2
We note here that controlling for a respondents income does not affect the size of the
education gap at all.

23

The second column in Table 4 adds our variable capturing respondents levels of
economic dissatisfaction. As we saw in our previous vote choice models, this variable is
statistically significant and clearly important. However, controlling for economic dissatisfaction
only reduces the size of the education gap from 22 points to 18 points. Thus, economic
dissatisfaction does not account for most of this gap. In the third and fourth columns, we add our
sexism and racism scales, respectively. Adding each of those variables individually to the model
results in a much larger reduction in the education gap. In fact, controlling for racism or sexism
reduces the size of the education gap by more than half.
The final model in Table 4 includes both the scales for racism and sexism to see what
combined effect both items have on reducing the education gap among whites. When we control
for both an individuals attitudes on racism and sexism, the college gap drops to 7-points; this is
less than one-third of the size of the original education gap among whites. It is perhaps worth
remembering here that the previous four presidential elections witnessed a college vote choice
gap among whites of between 5 and 7 percentage points. Thus, controlling for racism and sexism
effectively restores the education gap among whites to what it had been in every election since
2000.

Conclusion
The 2016 campaign witnessed a dramatic polarization in the vote choices of whites based
on education. In this paper, we have demonstrated that very little of this gap can be explained by
the economic difficulties faced by less educated whites. Rather, most of the divide appears to be
the result of racism and sexism in the electorate, especially among whites without college

24

degrees. Sexism and racism were powerful forces in structuring the 2016 presidential vote, even
after controlling for partisanship and ideology. Of course, it would be misguided to seek an
understanding of Trumps success in the 2016 presidential election through any single lens. Yet,
in a campaign that was marked by exceptionally explicit rhetoric on race and gender, it is
perhaps unsurprising to find that voters attitudes on race and sex were so important in
determining their vote choices.
Whether the 2016 election will simply be an aberration or the beginning of a trend
remains to be seen. However, there is reason to think that Trumps strategy of using explicitly
racist and sexist appeals to win over white voters may be followed again by candidates in future
elections. After all, Valentino et al. (2016) show that there is no longer a price to be paid by
politicians who make such explicit appeals. Explicit racist and sexist appeals appeared to cost
Trump some votes from more educated whites, but it may have won him even more support
among whites with less education. If Republicans see little prospect of winning over racial or
ethnic minorities in the near future, they have two choices moderate their appeals in order to
restore their advantage among more educated white voters (even if it costs them some votes
among less educated whites) or repeat the Trump strategy to maximize their support among less
educated whites (even at the expense of winning large margins among college educated whites).
As the norms governing political rhetoric appear to have largely been shattered in 2016, the latter
strategy is at least as plausible as the former, and that may have significant consequences for the
stability of American democracy.

25

References
Bartels, Larry M. "Whats the Matter with Whats the Matter with Kansas?." Quarterly Journal
of Political Science 1, no. 2 (2006): 201-226.
Bos, Angela L., Monica C. Schneider, and Brittany L. Utz. 2017. Navigating the Political
Labyrinth: Gender Stereotypes and Prejudice in U.S. Elections. In APA Handbook of the
Psychology of Women, Cheryl Travis and Jackie White, Editors.
Brady, David, Benjamin Sosnaud, and Steven M. Frenk. "The shifting and diverging white
working class in US presidential elections, 19722004." Social Science Research 38, no.
1 (2009): 118-133.
Claassen, Ryan L., and John Barry Ryan. "Social Desirability, Hidden Biases, and Support for
Hillary Clinton." PS: Political Science & Politics 49, no. 4 (2016): 730-735.
DeSante, Christopher D. and Candis W. Smith. 2016. The Two Dimensions of Whites Racial
Attitudes, or: The New New Racism. Working paper.
Dolan, Kathleen. "Gender Stereotypes, Candidate Evaluations, and Voting for Women
Candidates What Really Matters?." Political Research Quarterly 67, no. 1 (2014): 96107.
Frank, Thomas. What's the matter with Kansas?: how conservatives won the heart of America.
Macmillan, 2007.
Glick, Peter, and Susan T. Fiske. "The ambivalent sexism inventory: Differentiating hostile and
benevolent sexism." Journal of personality and social psychology 70, no. 3 (1996): 491.
Hayes, Danny. "When gender and party collide: Stereotyping in candidate trait
attribution." Politics & Gender 7, no. 02 (2011): 133-165.
Huddy, Leonie, and Nayda Terkildsen. "Gender stereotypes and the perception of male and
female candidates." American Journal of Political Science (1993): 119-147.
Masket, Seth. Whats the Matter With Kansas? aptly describes the 2016 election but was
written in 2004. Vox. (2016). Url: http://www.vox.com/mischiefs-offaction/2016/12/1/13807382/thomas-frank-kansas-2016-election
Mendelberg, Tali. The race card: Campaign strategy, implicit messages, and the norm of
equality. Princeton University Press, 2001.
Neville, Helen A., Roderick L. Lilly, Georgia Duran, Richard M. Lee, and LaVonne Browne.
"Construction and initial validation of the Color-Blind Racial Attitudes Scale
(CoBRAS)." Journal of Counseling Psychology 47, no. 1 (2000): 59.

26

Pearson, Kathryn, and Eric McGhee. "What It Takes to Win: Questioning" Gender Neutral"
Outcomes in US House Elections." Politics & Gender 9, no. 4 (2013): 439.
Porter, Eduardo. Where Were Trumps Votes? Where the Jobs Werent. New York Times (Dec.
14, 2016): B1.
Poteat, V. Paul, and Lisa B. Spanierman. "Further validation of the Psychosocial Costs of
Racism to Whites Scale among employed adults." The Counseling Psychologist 36, no. 6
(2008): 871-894.
Schuman, Howard, Steeh Charlotte, Bobo Lawrence, and Krysan Maria. "Racial Attitudes in
America: Trends and Interpretations." Cambridge, MA: Harvard (1997).
Sniderman, Paul M., and Thomas Piazza. The Scar of Race Cambridge: Belknap/Harvard Univ.
Press (1993).
Spanierman, Lisa B., and Mary J. Heppner. "Psychosocial Costs of Racism to Whites Scale
(PCRW): Construction and Initial Validation." Journal of Counseling Psychology 51, no.
2 (2004): 249.
Spanierman, Lisa B., V. Paul Poteat, Amanda M. Beer, and Patrick Ian Armstrong.
"Psychosocial costs of racism to whites: Exploring patterns through cluster
analysis." Journal of Counseling Psychology 53, no. 4 (2006): 434.
Streb, Matthew J., Barbara Burrell, Brian Frederick, and Michael A. Genovese. "Social
desirability effects and support for a female American president." Public Opinion
Quarterly 72, no. 1 (2008): 76-89.
Tesler, Michael. Views about race mattered more in electing Trump than in electing Obama.
Washington Post (Monkey Cage). November 22, 2016. URL:
https://www.washingtonpost.com/news/monkey-cage/wp/2016/11/22/peoples-viewsabout-race-mattered-more-in-electing-trump-than-in-electing-obama
Valentino, Nicholas A., Vincent L. Hutchings, and Ismail K. White. "Cues that matter: How
political ads prime racial attitudes during campaigns." American Political Science
Review 96, no. 01 (2002): 75-90.
Valentino, Nicholas A. Fabian G. Neuner, and L. Matthew Vandenbroek. The Changing Norms
of Racial Political Rhetoric and the End of Racial Priming. Working paper. URL:
https://www.researchgate.net/publication/310230276_The_Changing_Norms_of_Racial_
Political_Rhetoric_and_the_End_of_Racial_Priming

27

Appendix
In this appendix, we examine whether our results are robust when we account for
additional variables for our vote choice models. Specifically, our survey included the four-item
child rearing battery to measure authoritarianism and a three-item battery to capture populist
attitudes. We scaled these in a similar way to how we scaled the racism and sexism items using
an IRT graded response model for the populism items and a two-parameter logistic IRT model
for the child rearing items. Table A1 reproduces the probit models presented in Table 2 of the
paper, but adding these two additional scales to the model. The coefficients for the
authoritarianism and populism scales are statistically significant in these models, indicating that
respondents who registered higher levels of authoritarianism and populism were more likely to
support Trump. However, adding these models to the analysis leads to only a small decrease in
the size of the coefficients for the sexism and racism scales. The coefficients for racism and
sexism are more than twice the size of the coefficients for authoritarianism and populism,
indicating that they had a more substantial influence on vote choice. Most importantly, our
findings are robust even when we account for these other items.

28

Table A1: Probit estimates of factors affecting two-party vote for Trump
All likely voters
Hostile sexism scale

White likely voters

2.658
3.038
(0.394)**
(0.552)**
Racism scale
2.774
2.694
(0.525)**
(0.767)**
Dissatisfaction with econ. situation
1.408
1.580
(0.292)**
(0.417)**
Authoritarianism Scale
0.769
1.227
(0.334)*
(0.470)**
Populism scale
1.285
1.696
(0.552)*
(0.675)*
Ideology
2.150
2.900
(0.458)**
(0.817)**
7 point Party ID
3.882
4.280
(0.338)**
(0.511)**
Female
0.509
0.566
(0.182)**
(0.255)*
2bn.agecat
1.500
2.212
(0.458)**
(0.643)**
3.agecat
1.516
2.007
(0.453)**
(0.601)**
College degree
0.012
-0.170
(0.204)
(0.284)
1bn.incomecat
-0.352
-0.599
(0.265)
(0.381)
2.incomecat
-0.245
-0.497
(0.262)
(0.366)
3.incomecat
-0.644
-0.806
(0.315)*
(0.459)
1bn.racecat
-0.457
(0.380)
2.racecat
-0.879
(0.422)*
3.racecat
-0.849
(0.420)*
Constant
-8.293
-10.341
(0.770)**
(1.259)**
N
1,304
721
Note: Entries are probit coefficients with standard errors in parentheses. * p<.05, ** p<.01.

29