Strongest predictors of voting for Trump predictive
This graphic created by visualizes a machine learning model that was trained to predict who planned to vote for Trump vs. Clinton based on 138 factors we collected about each person. Our organization is non-partisan, with the mission of studying and improving societal decision making, and we offer many free decision making tools.

On the left (in red) are those factors in our predictive model for which agreeing was predictive of supporting Trump, and on the right (in blue) those where agreeing was predictive of Clinton support. On each side, the strongest factors are at the top, the weakest at the bottom. When a factor is strong, that means that the prediction model found it to be useful for prediction even taking into account the other variables. When a factor is weak, that means that once the other variables were taken into account, the predictive model did not need to use that factor much to make accurate predictions. The model achieved 91% accuracy at predicting which of the two candidates each person would vote for when it was applied to data that was not used to train the model. The data used to build the model came from our survey of respondents recruited online (on Oct 2 and 3, 2016) consisting of 800 people who said they had already decided to vote for Trump or Clinton. A random sample of 640 data points out of the 800 were used to train the model, and the remaining data to test it. The data does not come from a nationally representative sample. However, we found that this sample matched larger and more representative samples along all 12 dimensions that we tested it (i.e. both in our data and in these larger and more nationally representative polls, Trump supporters relative to Clinton supporters were more male, white, aged, republican, socially conservative, fiscally conservative, anti-immigration, uneducated, married, rural dwelling, authoritarian, and religious). The prediction algorithm used was logistic regression with L2 regularization (C=.0025). Predictive strength refers to the absolute value of the coefficients in the prediction model (where each feature has had its mean subtracted, and been divided by its standard deviation, to put all variables in comparable units). Our data set can be downloaded here. If you use this data please let us know at and cite our work.
Strongest predictors of voting for Clinton
I'm a registered member of the Republican Party. 0.130 Many have claimed that this presidential election is fundamentally different than those prior, or that trump is a new type of candidate. While true in a number of ways, the single strongest predictors that we found of supporting Trump versus Clinton is whether the person is a registered Republican or Democrat. In other words, as unique as Trump is, and despite the unusual turn of affairs where a number of Republican politicians turned their back on him, party affiliation is still extremely relevant in this election. If we examine Trump and Clinton supporters that registered with a political party, we found that 40% of the Trump supporters and 51% of Clinton supporters (in one of our study samples) agreed that they would almost certainly vote for the presidential candidate of their registered party regardless of who that person is and who the opposing party candidates happen to be. So party loyalty is strong, but there is still room for deviation, and the remaining factors we investigate will show that there are many useful factors to take into account for predicting candidate support once registered party has already been controlled for.

The chart (below) shows the percentage of Clinton supporters (blue) and Trump supporters (red) in our sample that responded "No" and "Yes" to each question shown.
-0.149 I'm a registered member of the Democratic Party.
There is too much political correctness in this country. 0.086 The Washington post says, "If there is one uniting principle the defines Donald Trump's campaign for president -- besides, perhaps, winning and being classy -- it is that ." political correctness is bad. It's probably not a coincidence that he talks about this issue so much, as we found it to be a very strong predictor of support for him. In a follow up study, we found that Clinton supporters are highly divided on whether there is too much political correctness, but those who don't think there is too much like Political Correctness because they think it reduces how often people are offended, sets a minimum standard for politeness, and reduces societal tolerance for prejudice. On the other hand, Trump supporters in our study nearly all agreed that there is too much political correctness, and tend to believe that our society has become far too sensitive to being offended. They see political correctness as frustrating because it limits their freedom of speech, and dangerous because it prevents topics they feel are important, related to radical Islamic terrorism, illegal immigration, and race, from being discussed.

The charts below show the percentage of Clinton supporters (blue) and Trump supporters (red) in our sample that gave each response on a seven-point agreement scale ranging from -3 (totally disagree) to +3 (totally agree). The dotted lines show the average score for Clinton supporters (blue) and Trump supporters (red). This scale is used because it provides more nuance about how strongly people feel than a typical "agree"/"disagree" scale.
Why is belief that the government should require everyone to have health insurance such a strong predictor of supporting Clinton? There are at least three plausible reasons. First, Trump said he wanted to repeal the Affordable Care Act and replace it with a plan based more on free market principles, so anti-healthcare advocates may find Trump more appealing. Second, Obama is very strongly associated with universal healthcare, so supporting it may be a proxy for supporting Obama, and Clinton is sometimes viewed as a successor to continue Obama's work. Third, Clinton supporters tend to believe that the government should look out for all of us, whereas Trump supporters tend to want a smaller government, believe that free market solutions work better than government programs, and think that we all should look out for ourselves (rather than depend on Government). So what did we find when we asked respondents to explain their agreement or disagreement regarding required health insurance? Clinton supporters who agreed we should require it tended to support their belief by appealing to the idea that health care is a human right which everyone deserves, as well as to the idea that requiring it may make insurance cheaper for everyone. Trump supporters who opposed it argued that people should have the freedom to choose since not everyone needs or wants insurance, and see requiring it as the government doubling down on a healthcare system that they view as broken or excessively expensive. -0.103 It's better if the government requires everyone to have health insurance.
It is very valuable for a politician to have previous experience in *business*. 0.086 Thinking that business experience is valuable in politics is very predictive of supporting Trump, whereas thinking that political experience is valuable is predictive of Clinton. Why? First, economic conservatives (who naturally tend to support the Republican candidate) likely trust companies more than economic liberals, and so may have a more favorable view of business people in general. They also tend to prefer free market solutions, and so may trust a business person to implement these more than a career politician. Add on top of this the fact that Trump supporters tend to have less trust of politicians in general, and we can see why we would find a strong link between agreement to these statements and which candidate you support.

But it seems highly likely that part of the explanation for the strength of this factor is rationalization. If you support Clinton, you're going to want to find reasons to justify her being the better candidate, and if you support Trump you'll want to find reasons to back up your view that he is better. Business experience is a highly noticeable advantage of Trump, and political experience a highly noticeable advantage of Clinton. A reasonable person will agree that both political and business experience can be useful in politics. But if your preferred candidate has a lot of one and none of the other, you may become biased in favor of the type of experience your candidate happens to have. We expect that the high predictive strength of this factor is related to Confirmation Bias, where we attempt to gather evidence to prove that our beliefs are right, rather than to objectively test whether we really are right. Sadly, we all suffer from this bias at least some of the time.

The charts below show the percentage of Clinton supporters (blue) and Trump supporters (red) in our sample that gave each response on a seven-point agreement scale ranging from -3 (totally disagree) to +3 (totally agree). The dotted lines show the average score for Clinton supporters (blue) and Trump supporters (red).
-0.095 It is very valuable for a politician to have previous experience in *politics*.
It's important for a leader to say what's truly popping into their mind rather than talking points. 0.083 What does it mean for a person to be honest? Is honesty about getting facts and numbers correct, or is it about saying what is truly running through your mind (and your true feelings) rather than filtering what you're thinking? This debate over what honesty means is a fascinating difference between Clinton and Trump supporters. Both candidates say a disturbingly large number of false and misleading things (sometimes by mistake, other times intentionally). But fact checkers find that Trump miscites facts and statistics far more often than Clinton. For instance, before the election PolitiFact rated 71% of the statements of his that they had fact checked as mostly false or worse, versus 27% for Clinton. When Clinton supporters see his factual errors they think he's a liar (whereas Trump supporters have high distrust of the media and may say the fact checkers are biased). But even if Clinton is much more factually accurate, there is a way in which Trump is more honest than Clinton: he is more likely to say what happens to be popping into his mind and to express what he feels in the given moment, rather than carefully controlling and filtering what he says. It's factual accuracy versus authenticity. Trump supporters tend to care more about this form of honesty than Clinton supporters. With Clinton it's hard to know what she is truly thinking due to her controlled nature. But that's okay to Clinton supporters, who tend to focus more on factual accuracy. So which is actually more honest, getting the facts about the world right, or saying what's really popping into your mind and what you're feeling? It depends on what honesty means to you. It's a classic liberal position that there should be restrictions on the purchase of guns, tying into the idea that the government should protect us.

Considering this idea dispassionately for a moment, we can all agree that it's generally bad to restrict a person's freedom to do what they choose, so long as their actions aren't harming anyone. And we can also likely agree that the vast majority of gun owners will never hurt a person with a gun. On the other hand, we can all agree that it's horrendous when guns are used to murder people. The socially conservative position here is that the law abiding gun purchasers shouldn't be punished due to the fact that some small fraction of gun owners use them to kill people. The socially liberal view is that the freedom to buy guns is not worth a potentially higher murder and suicide rate that easy access to guns may help cause.

Here both factual issues and questions of values are at stake. The factual question is: how much would murders and suicides fall if we initiated different forms of gun legislation? The values question is: to what extent is it worth accepting a greater number of murders and suicides to enable more freedoms around owning a gun? And of course, people in rural areas (i.e. areas where Trump supporters tend to live) are more likely to own guns, and therefore are more likely to be personally impacted by gun regulation.
-0.092 We should have a lot of restrictions on the purchase of guns.
It's very important that we protect U.S. jobs, and not allow them to disappear to foreign countries. 0.075 At face value, the importance of this factor seem strange, as the official unemployment rate has been falling since 2009 and is currently (Nov, 2016) at only about 5%, which is certainly not a high level. So why is there so much fear about job loss? Part of the explanation may have to do with disagreements about how to measure the rate of unemployment, as Trump has claimed the official numbers don't take into a huge number of people, but some of the numbers he has claimed have also been far too high. The official numbers omit some groups for good reason, since it makes sense to exclude the retired, full time students, full time homemakers who take care of the home by choice, and people who have no interest in working. It is reasonable to disagree however about how to count people that have not looked for a job in months (but want one) and those who work part-time but would prefer to be full-time. If we add those people back into the unemployment figures they end up about twice as high. And there are still other groups we could consider adding back in. Another issue is that since the economic recovery began after the financial collapse, the recovery has not been equal in different parts of the country. So perhaps unemployment is worse in areas where Trump supporters tend to live. In our study we found Clinton supporters reported that they believe a median of 6% of people they know are unemployed, compared to 12.5% for Trump supporters. However, an analysis by the Washington Post suggests that Trump supporters are not more likely to be affected by economic hardship, though may be extra worried abouttheir children having trouble finding jobs. Some Clinton supporters have accused people of supporting Trump largely due to racism. Defining what it means for someone to be racist though (so that you can actually measure it) is extremely tricky, and establishing that racism is the cause of a person voting for Trump (and not just correlated with it) is even harder still. It is very likely that less than 50% of both Trump and Clinton supporters explicitly endorse racist views, whereas measures of implicit racial bias tend to find that a majority of both democrats and republicans have negative implicit judgements based on race. Articles claiming that support for Trump is caused mainly by racism often suffer from three methodological flaws. (1) They assume that answers to multiple choice on issues related to race are reliably indicators of racism, whereas in fact usually these questions have multiple interpretations, and there are ways to agree to them other than because a person holds a racist attitude. (2) They overlook or gloss over the fact that a substantial portion of Trump's support came just from him being the republican candidate (i.e. that support is much more about party affiliation than race). (3) They jump from Trump support being correlated with views on race to concluding that it is the actual cause of Trump support, which is an inference that is extremely hard to prove from survey data. -0.088 People of color in the U.S. are not treated as well, on average, as White people.
I lean conservative on *social* issues. 0.072 Naturally, being liberal on social issues is a strong predictor of supporting Clinton, and being conservative on social issues is a strong predictor of supporting Trump. Traditionally, social conservatives are thought to oppose abortion, affirmative action, gun control, environmental protection, universal health care, and gay marriage while social liberals tend to support them. -0.087 I lean liberal on *social* issues.
Immigrants threaten American customs and values. 0.072 Trump supporters seem to worry about immigration not just because of fear of job loss, but also because of fear that too much immigration threatens American customs and values. For instance, one might think that immigrants will have values or preferences that differ in at least some ways than your typical American values, as all cultures have at least somewhat differing views on what's most important, and on how society should be configured to support those values.

When we asked Clinton supporters who disagree with the statement  "immigrants threaten American customs and values" why they disagree, they tended to argue that America is a country where everyone was at one time an immigrant, and that diversity makes the U.S. better by bringing diverse perspectives and enriching culture.

On the other hand, Trump supporters who instead agree (rather than disagree) with the statement tended to argue that some fraction of immigrants may be dangerous (e.g. terrorists or those with radical anti-American views), that some immigrants choose not to assimilate to American values, that traditional American customs and values are diluted when a greater share of the population have different customs or values,  that some immigrants are here illegally, and that some people move here to mooch off of the system.
In one study we conducted, 64% of Clinton supporters agreed that the economy is performing well economically, while only 21% of Trump supporters agreed. How could there be so much disagreement about what seems like a pure matter of fact?

Clinton supporters in our study who agreed that the U.S. is performing well economically tended to point out that there has been a large recovery since 2008, that the official unemployment rate has fallen, that the stock market has risen quite a lot since the crash, and that relative to many other countries in the world we're doing great economically.

Trump supporters who said we are not doing well economically tended to argue that the national debt is too high, that there is still too much unemployment, that economic growth is too low, and that prices have risen too much (relative to wages).

While some of these points are debatable, both sides do make some good points that provide genuine evidence for their perspective. The Clinton and Trump supporters in our sample were simply focused on different aspects of what it means for America to perform well (or badly) economically.

-0.075 America is currently performing well economically.
I identify as being White. 0.069 -0.073 I lean liberal on *economic* issues.
America has too many immigrants. 0.064 -0.071 Diversity improves the United States.
I don't at all trust the mainstream media. 0.059 -0.069 It's important to continue fighting for women's equality in the U.S.
I lean conservative on *economic* issues. 0.059 -0.065 Taxes should be substantially *increased* on the wealthiest Americans.
It's important to have a leader who is not afraid to do the unorthodox. 0.057 -0.059 There is too much wealth *inequality* in America.
The government should have less power than it does now. 0.055 -0.058 Humans have caused permanent warming to the planet.
What we need most in the U.S. is for things to change. 0.055 -0.057 The laws of the U.S. currently are setup to unfairly benefit the wealthy.
America used to be a much better place than it is now. 0.053 -0.055 I identify as Black or African American.
Torture is an effective technique that our country should use to get information from confirmed terrorists. 0.052 -0.052 It's very bad for a leader to be narcissistic.
Racial profiling is worthwhile because it makes us safer. 0.051 -0.051 A good leader should not brag.
The U.S. government is not currently looking out for my interests. 0.049 -0.047 It's very bad for a leader to be impulsive.
You can't trust a leader that filters what they are thinking before they speak. 0.048 -0.045 We should raise the minimum wage.
We should reduce the number of immigrants that are *legally* allowed to enter the United States. 0.047 -0.042 It is important for a leader to be a *warm* (not cold) person.
America spends too much money trying to help other countries. 0.045 -0.040 It's very bad for a leader to be greedy.
America needs to show the world that it's the strongest country on earth. 0.043 -0.038 It's important for a leader to think carefully before they speak.
I am male. 0.039 -0.036 America needs to show the world that it helps other countries that are in need.
Taxes in this country are too high. 0.039 -0.036 Highest level of your education.
People who are wealthy usually worked hard and contributed to society in order to become wealthy. 0.036 -0.035 Women make just as good leaders as men do.
I hope that the current governmental system we have collapses. 0.036 -0.035 It's important that a leader is a *kind* person.
Criminals who commit truly horrible crimes should receive the death penalty. 0.032 -0.033 If a leader often gives inaccurate numbers and figures when they speak to the public then they can't be trusted in other important matters.
One of the United States' top priorities should be to stop terrorism. 0.029 -0.033 America should invest more in alternative forms of energy that are less polluting.
Level of committed romantic relationship. 0.028 -0.033 There are many valuable services that only the government can provide for us.
Nearly anything is better than the current status quo in the United States. 0.027 -0.031 I live in the Western region of the U.S.
You really can't trust politicians. 0.027 -0.031 It is important for a leader to be highly intellectual.
It's extremely important that we *not* elect leaders that are corrupt. 0.025 -0.030 America should never threaten other countries with the use of nuclear weapons.
The businessman and the manufacturer are more important to our country than artists and writers. 0.025 -0.028 America should *decrease* its military spending.
I live in a *rural* (rather than an *urban* area). 0.025 -0.027 Police officers in the U.S. are more violent than they should be.
Science is not as trustworthy for figuring out the truth as many people think. 0.024 -0.027 Gay people should have the right to marry.
It's extremely important that our president support our troops and military officers. 0.024 -0.025 American should only use its military as a very last resort.
Obedience and respect for authority are the most important virtues children should learn. 0.024 -0.022 It's very bad if a leader gets the facts or numbers wrong when giving a speech.
Penalties for criminals should be harsher than they are right now. 0.023 -0.022 It's very important for a leader to be a *sincere* person.
Every American adult should be responsible for their own safety, not relying on others to protect them. 0.023 -0.020 Maintaining the existing alliances of the United States is a good way to achieve our policy goals.
I am happy with the way my own life is currently going. 0.023 -0.020 The government should give students loans so that they can afford college.
All politicians are liars. 0.023 -0.019 The U.S. should have an active role in world affairs.
Free markets usually are a better solution than government programs. 0.022 -0.019 Students in middle school and high school *don't* spend enough hours at school every year.
America has become an increasing *immoral* place. 0.022 -0.017 It's important that a leader be a *likable* person.
Woman and men are best suited towards different kinds of work. 0.022 -0.017 It's very bad for a leader to be power hungry.
It's very important for the president to deeply love America. 0.021 -0.016 I identify as Hispanic or Latino.
There are currently far too few jobs available in America. 0.021 -0.014 Corporations should not be trusted.
It's very important for our leaders to be *decisive*. 0.021 -0.014 We should remove tariffs and promote free trade.
I want the leader of our country to be *entertaining*. 0.019 -0.013 Number of children.
It's important for a leader to inspire the people. 0.019 -0.011 I am a person that tends to follow the rules.
Insults to our honor should always be punished. 0.019 -0.011 Total annual household income.
It's very important that a leader be *charismatic*. 0.018 -0.010 It's America's responsibility to topple corrupt regimes in other countries.
Abortion is morally wrong. 0.017 -0.006 I typically feel like none of the politicians running for office represent my views.
If you are a good person and work hard, ultimately your life will go well and you'll get good things. 0.016 -0.005 It is very important for a leader to be in control of their emotions.
It's important that the president of the U.S. brings in new ideas. 0.014 -0.005 A person who is terminally ill and suffering should be allowed to *choose* to prematurely end their own life.
Marijuana should be fully legalized in the entire U.S. 0.013 -0.001 All factors considered, America is the best country on earth to be a citizen of.
Number of years old (i.e. Age). 0.013
America is a wonderful place. 0.013
I live in the Midwestern region of the U.S. 0.012
I would feel really bad for an animal caught in a trap. 0.011
One of the most important things for a leader to do is show *strength*. 0.011
I live in the Southern region of the U.S. 0.011
It's important to do what authorities tell us to do. 0.010
Religion is important to me. 0.009
There are important differences between different races. 0.008
It's a very bad sign if a leader is caught telling a lie to the public. 0.008
I have children. 0.008
When two things are different from each other, it's usually the case that one is better than another. 0.007
It's very important that the leaders we elect have a prior strong track record of success. 0.007
I live in a neighborhood where most people are similar to me in race and ethnicity. 0.007
I am a deeply religious person. 0.006
Freedom is what makes America great. 0.006
I am an intellectual. 0.004
I live in the Northeastern region of the U.S. 0.004
It is a good thing to really enjoy having sex. 0.004
For the most part, people get what they deserve in life. 0.003
The advancement of technology will improve the U.S. more than anything else. 0.003
I have a blue-collar (rather than white-collar) profession. 0.003
I generally trust other people. 0.003
People like me don't have any say about what the government does. 0.002
I consider myself to be of high socio-economic status. 0.002
Courage is one of the most important traits for a leader to have. 0.001
I am retired from working. 0.000
It's important that a leader not change their mind about important issues. 0.000