The US presidential election is imminent. Given the amount of polarization and partisan sorting of political issues in the US, you might wonder: “Who on earth still hasn’t decided who they’ll vote for?!” These mysterious undecided voters, whoever they are, may well determine the election winner and, therefore, part of the future of America.
To better understand who these undecided voters are, we ran a study exploring their beliefs and examined predictors of who remains undecided. We have some surprising results that we’d like to share with you.
Please note: we are a nonpartisan organization, with a policy against taking a stance on political leaders because we hope to reach all sides with information about how to think more clearly and improve decision-making. As such, this newsletter aims to be informative, not to influence the election. As always, we encourage our readers to make their own well-considered choices.
Our Study
Initially, we ran a qualitative pilot study on 24 undecided Americans, asking them a wide variety of open-ended questions about their beliefs, and why they are struggling to decide who to vote for. This was merely to develop hypotheses about why people are struggling with this decision.
Next, we recruited 406 participants through our study recruitment platform, Positly, as well as social media, in multiple waves of recruitment of Americans who are legally allowed to vote, with 91 participants indicating they were undecided who to vote for.
We asked all of our participants a wide variety of questions about their attitudes towards the candidates and to politics in general. Then we built a model to predict which participants are undecided (using a technique known as logistic regression). We also controlled for the effects of age, conservativeness on fiscal issues, and conservativeness on social issues, so that biases in these factors would be less likely to skew results. This allows us to explore which factors predict who is undecided.
So, what did we find? Well, before reading on, see if you can guess which of these factors were predictors of who is undecided:
Seeing the U.S. two-party system as fundamentally flawed
Being a centrist (i.e., near the middle on the political spectrum)
Falling in the middle, between where you perceive the two candidates to be (in terms of social and fiscal conservativeness/progressiveness) so that you are about equidistant from each
Finding it difficult to determine the true positions and values of Donald Trump
Have your guesses? Let's move on to the results, starting with factors that were not useful for predicting who is undecided.
Factors that did not predict being undecided
1️⃣ Viewing the system as ‘bullshit’
Intuitively, it's natural to think that those who are disillusioned with the two-party system might struggle to know who to vote for because they feel that the whole system itself is not doing a good job. This is what we expected to find. But we were wrong!
To test this hypothesis, we asked participants to what extent they agreed with the statement:
"The current two-party system in the U.S. is fundamentally flawed and doesn't serve the public's best interests”
Their responses failed to predict not being decided on who to vote for.
It's worth noting that frustration with the two-party system is widespread across the political spectrum. Simply put: lots of people think the system sucks. In our full set of participants (including both those who are decided and undecided), 29% said they strongly agreed with this sentiment, with an additional 36% at least slightly agreeing. And those who were undecided had similar views on this to the rest of the group.
2️⃣ Being in between the two candidates (ideologically)
We asked each participant where they land on political spectra from progressive to conservative with these two questions:
“Currently, where do your views fall on a scale from completely fiscally progressive, to completely fiscally conservative?”
“Currently, where do your views fall on a scale from completely socially progressive, to completely socially conservative?”
And we offered the following definitions of the relevant terms:
Fiscal progressives usually prefer that the government provides more services, and passes more regulation in order to try to make society better.
Fiscal conservatives usually advocate low taxes, reduced government spending and minimal government debt, and typically are in favor of deregulation, free trade, free markets, privatization, and tax cuts.
Social progressives usually believe that the structure of one's family, one's sexual choices, how one chooses to use one’s body, and the nature of marriage should be left up to individual decision. They tend to say that as long as one does no harm to others, no lifestyle should be treated by the government as inherently better than any other.
Social conservatives usually believe that certain traditional lifesyle choices are preferable to other alternative lifestyle choices, and tend to oppose abortion, same-sex marriage, pornography and the legalization of drugs such as LSD and other psychedelics. They also tend to be more supportive of religion in the public sphere (e.g., within schools, official institutions, etc.)
From this, we were able to plot where people placed themselves, ideologically, in a two-dimensional graph. Here is that graph for the undecided voters:
The darker the dot, the more people are at that location. We can see that there are four quadrants here:
Bottom-left - Progressives: these are people whose fiscal and social conservatism are both negative (meaning they are fiscally and socially progressive). This indicates that they are progressives. 19% of our sample of undecided voters fell into this quadrant.
Top-right - Conservatives: these are people whose fiscal and social conservatism are both positive, indicating that they are conservatives. 18% of our sample of undecided voters were in this quadrant.
Bottom-right - Libertarians: people who are fiscally conservative but socially progressive. This combination of views is most often associated with libertarians, and 34% of our sample of undecided voters had this perspective.
Top-left - Populists: people who are fiscally progressive but socially conservative. This combination is not common in the US, as can be seen by the gap in that corner of the chart above, but it is associated with populist movements around the world - emphasizing ‘traditional’ social values while advocating fiscal policy that aims for a strong and large government - for instance one that is for ‘common people’ against perceived elites. This has also been seen in some nationalist movements. Only 4% of our undecided voters were in this quadrant.
The rest of our undecided voters were on one of the axes (and hence, were on the boundary between quadrants without being within any of them).
Then we asked our participants to say where they thought Donald Trump and Kamala Harris placed on those two spectra. Here’s what the undecideds had to say:
Again, darker dots represent more data points. Here, the red dots show the median answers. It’s interesting to note that there is such a wide spread of answers from participants - indicating quite a lack of consensus about where each candidate falls - and yet the two medians are symmetrical (with Trump at [2,2] and Harris at [-2,-2]).
From this data, we were able to calculate the "distance" that people perceive themselves to be from both Trump and Harris in this two-axis system.
Shouldn't people who see themselves as falling near the middle between the candidates (ideologically) find it harder to decide who to vote for?
We were surprised to find that seeing yourself as in the middle, between the two, was not a predictor of being an undecided voter.
One reason this might be surprising is because it’s common to imagine that the undecided voter is someone who simply sits on the fence, in the middle of the two candidates, struggling to perceive which one is slightly closer to them ideologically. In reality though, there are lots of ways to be an undecided voter - including feeling disengaged from politics entirely, or like the left-right spectrum (including centrism in the middle) doesn’t represent you at all.
3️⃣ How close you are to the closest candidate
For this measure, we calculated how close people perceived themselves to be to the closest of the two candidates (Harris and Trump) on those two-dimensional fiscal and social graphs again. But this also was not predictive.
This is perhaps surprising because you might think that placing yourself as very close to a candidate on these axes would make you more likely to identify with them or feel represented by them, and therefore be more likely to know you want to vote for them.
On the other hand, if someone perceives themselves to be ideologically close to one candidate, other factors — such as doubts about the candidate’s character, honesty, competence, effectiveness or likelihood of achieving their goals — could lead them to hesitate to vote for that candidate. Undecided voters seem to be considering more than just ideology when deciding.
Factors that did predict being undecided
Now, here are the interesting factors we found that did predict being an undecided voter.
1️⃣ Not being able to tell what Trump really believes
One of the questions we asked was:
“How difficult do you feel it is to determine the true positions and values of Donald Trump in this election?”
And our results showed that, the more unsure someone reported being about the true positions and values of Trump, the more likely they were to be an undecided voter. Interestingly, however, having the same feelings about Harris was correlated with being undecided when considered on its own, but when added to the logistic regression with all the other factors, it was not predictive of being an undecided voter (presumably because its effect was already being accounted for by some other factor in the model).
We don't know why this is the case. It might indicate that people are more concerned with or unsure about what Trump really believes than by what Harris does. Interestingly, undecided voters report it being harder to determine the true positions of both Trump and Harris compared to decided voters.
Here are charts, showing the results we got for these two questions:
2️⃣ Feeling like you don’t fit on the political spectrum
Strength of agreement with the statement:
"I feel like I don't fit anywhere on the left-right political spectrum"
was also predictive of being an undecided voter. The more someone felt like they didn't fit on that spectrum, the more likely they were to be undecided.
For these participants, the typical political categories (progressive, conservative, centrist, etc) might not do a good job of capturing their beliefs. For instance, they may have a combination of views that cut across traditional partisan lines, or maybe they view politics through a non-traditional lens - prioritizing issues or values that are not part of the typical national conversation or policy agenda of the main two parties. Perhaps it is for these reasons that these participants find it harder to decide who to vote for.
3️⃣ Closeness to Trump’s views
We found that, the closer participants saw themselves as being to (what they thought were) Trump’s views, on the two-dimensional (fiscal and social) graphs above, the more undecided they were on average. In other words, seeing yourself as close to Trump (socially and fiscally) was linked to more indecision about who to vote for.
Why might this be? Perhaps participants who see themselves as ideologically similar to Trump on the social and fiscal dimensions might still have reservations about other aspects of his candidacy. For example, they may feel aligned with his conservative views on specific policies but struggle with aspects of his character, leadership style, or perceived electability.
4️⃣ Perception of Harris’s social stance, and perception of Trump’s fiscal stance
Perhaps our most baffling finding was that viewing Harris as socially conservative was predictive of being undecided, as was viewing Trump as more fiscally progressive.
Contrarily, while our analysis showed that views about how fiscally conservative Harris is or how socially progressive Trump were were predictive of being an undecided voter only on their own - but once we account for the other factors the effect disappeared.
If you have any theories as to why, in particular, viewing Harris as more socially conservative, and viewing Trump as more fiscally progressive, were predictive of being undecided, please share them with us, as on this finding we are stumped.
Variable | Coefficient (positive means it's predictive of being undecided about who to vote for, negative means it's predictive of having decided already) | p-value (an estimate of the probability that you'd get a coefficient this big if, in fact, this factor was completely useless for predicting) | Individual correlation with being undecided (not controlling for any other factors) |
How difficult the participant feels it is to determine the true positions and values of Donald Trump | 0.2297 | 0.0018 | 0.2097 |
How strongly the participant feels they do not fit on the left-right political spectrum | 0.252 | 0.0019 | 0.3249 |
How socially conservative the participant sees Harris as being | 0.3112 | 0.0038 | 0.1887 |
Participant’s distance from Trump in the 2D space of social vs fiscal conservativeness | -2.4164 | 0.0185 | -0.2249 |
How fiscally conservative the participant sees Trump as being | -0.1853 | 0.044 | -0.1949 |
How difficult the participant feels it is to determine the true positions and values of Kamala Harris | 0.1399 | 0.0742 | 0.2418 |
How socially conservative the participant views Trump as being | 0.1388 | 0.1803 | -0.173 |
Participant’s distance from Harris in the 2D space of social vs fiscal conservativeness | 0.9687 | 0.1973 | 0.0181 |
Participant’s social conservativeness | -0.1332 | 0.1978 | 0.1049 |
Participant’s fiscal conservativeness | 0.0227 | 0.7938 | 0.1613 |
How fiscally conservative the participant sees Harris as being | 0.0202 | 0.8319 | 0.0986 |
What can we learn from this?
We hope you’ve learned a little more about the mysterious undecided voter. Of course, the correlations we’ve found might not be indicative of causation: it’s possible that (for instance) other underlying factors, which we haven’t measured, are influencing both a person’s indecision and their responses to our questions.
Still, if you want to understand the heart or mind of a voter, these results point to some interesting areas to consider. For instance:
1️⃣ Clarity on Trump
Undecided voters are more likely to be unsure about Trump’s true positions and values. Whether they vote for Trump or not could hinge on what conclusion they come to on these questions about him.
2️⃣ The Left-Right Spectrum Doesn't Capture Everything
Many undecided voters feel that they don’t fit neatly on the traditional political spectrum. This could mean that talking about policies in terms of "left" and "right" may not resonate with these folks. Candidates who care about reaching this group may need to focus on what they plan to do regarding specific issues, rather than relying on left or right coded talking points.
3️⃣ Ideological Distance From Each Candidate
It seems that undecided voters are not just focussed on which candidate is closer to them ideologically (on social and fiscal issues). In fact, being closer ideologically to Trump was associated with more likelihood of being undecided.
While our study certainly raises at least as many questions as it answers, we hope that it helps you understand the mysterious undecided voter better than you did before.
If you want to take your thinking about politics further, why not try our Political Bias Test? Learn to spot and overcome your political biases and test your US policy knowledge with this free quiz.
Comments