The Trolley Problem: What Real Voting Data Reveals About Moral Choices

In 1967, philosopher Philippa Foot asked a deceptively simple question: if a runaway trolley is heading toward five people, and you can pull a lever to divert it onto a side track where it will kill one person instead, should you pull it? For most people, the answer feels obvious. Pull the lever. But Foot was not interested in the answer — she was interested in the tension it created when compared to a different case: harvesting organs from a healthy patient to save five dying ones. Same arithmetic. Completely different intuition.

The lever: what the real data shows

Across decades of academic surveys, the lever version generates one of the most consistent findings in moral psychology: between 80% and 90% of respondents say they would pull the lever. On SplitVote, where the trolley dilemma is one of the most-voted scenarios, real user data tells a similar story. The minority who refuse are not confused — they often express a deontological intuition that active intervention to cause a death, even to save more lives, is categorically different from allowing the original trajectory to continue.

The lever variant works as a near-consensus case because it maintains moral distance. The person on the side track dies as a side-effect of redirecting the trolley, not as the means of stopping it. You are not using their death — you are accepting it as a cost of changing the trolley's path.

Step onto the bridge: the footbridge variant

Judith Jarvis Thomson introduced the footbridge (fat-man) variant in 1985. You are on a bridge above the tracks, standing next to a large stranger. Pushing them off would stop the trolley and save five people — they would die on impact. Jumping yourself would not stop it. The arithmetic is identical to the original case. But acceptance rates collapse to roughly 20-30%.

This is not noise. It is one of the most studied divergences in experimental philosophy. The gap between the lever (80-90% yes) and the footbridge (20-30% yes) reveals something structural about human moral cognition that neither pure consequentialism nor pure deontology fully predicts.

The doing-allowing distinction and Philippa Foot's real question

Foot was primarily interested in what scholars call the doing-versus-allowing distinction. There is a widely shared intuition that doing harm — even to produce a better outcome — is morally different from allowing harm to occur. The trolley lever sits in a grey zone: the harm to the one person is a foreseen but unintended side-effect of changing the trolley's path. In the footbridge case, the death is the instrument of rescue — you need the person's body to stop the trolley. The distinction is explored in depth in academic literature including scholarship on Foot's original contribution (Cambridge University Press, The Trolley Problem: The Doing-Allowing Distinction). Scholars continue to debate whether this distinction holds as a moral principle or reflects cognitive bias — but the empirical reality of the gap is not in dispute.

The doctrine of double effect

A second framework explaining the lever-footbridge gap is the doctrine of double effect, developed in scholastic moral theology and refined by 20th-century philosophers. It holds that an action with both good and bad effects is permissible if the bad effect is not the means to the good effect — it must be a side-effect. The lever satisfies this: one death is a side-effect of diverting the trolley, not the means of saving five. The footbridge violates it: stopping the trolley requires using the person as a physical obstacle. Their death is the means, not the side-effect.

What the gap between lever and footbridge reveals about you

If you would pull the lever but not push the stranger off the bridge, you are not being inconsistent. You are applying different moral reasoning to structurally different cases — and that reasoning is philosophically defensible. If you would do both, or neither, you are applying a coherent framework: either consistent consequentialism (outcomes determine rightness, mechanisms are irrelevant) or consistent deontological refusal (using any person as a means to an end is wrong regardless of outcomes). The trolley problem is uniquely useful because it isolates these frameworks and makes the differences visible.

What SplitVote data shows across demographics

On SplitVote, users who vote on both the classic trolley and the footbridge variant show a consistent pattern: the footbridge generates a significantly more even split. Younger users show slightly higher acceptance of the footbridge — possibly reflecting more exposure to utilitarian frameworks in education. The split also varies across regions, with users from countries with stronger collective-welfare cultural norms slightly more likely to push. These are trends, not conclusions: SplitVote data is self-selected and not a representative population sample.

Frequently asked questions

What percentage of people would pull the lever in the trolley problem?

Between 80% and 90% in controlled academic studies. Real-world platform data from SplitVote shows a similar majority, with regional and demographic variation.

What is the footbridge (fat-man) variant?

Introduced by Judith Jarvis Thomson in 1985: you stand on a bridge next to a large stranger. Pushing them off stops the trolley and saves five people. Most people refuse even though the arithmetic is identical to pulling the lever.

Why do answers differ between the lever and the footbridge?

The doing-allowing distinction: in the lever case, the death is a side-effect of redirecting the trolley. In the footbridge case, the death is the instrument of stopping it. The doctrine of double effect formalises this distinction. Joshua Greene's dual-process research adds the role of physical contact in triggering stronger emotional responses.

What was Philippa Foot's view on the trolley problem?

Foot's interest was the doing-versus-allowing distinction, not the lever result per se. She was asking why some harmful actions feel permissible and others do not. The fat-man variant was added by Thomson in 1985, not by Foot.

This article is for educational purposes. SplitVote voting data reflects anonymous aggregate user trends and does not constitute scientifically certified research. For peer-reviewed analysis, consult the academic literature on experimental moral philosophy.