Steven Levitt ran an experiment with more than 20,000 coin tosses over decisions people could not make, and the ones the coin told to make the change were happier six months later: his conclusion is that people may be excessively cautious about life-changing choices

The advice about moving on is almost always emotional. Let go, make peace, close the chapter. It treats staying too long as a feeling problem, something that resolves when you have processed enough.

The research treats it as a decision problem, and the decision research is considerably less flattering. What keeps people in a job, a product line, a company or a relationship past the point of usefulness is not usually sentiment. It is that leaving would mean conceding the earlier choice was wrong, and the person who made that choice was you.

There is one genuinely randomised study on this, which is unusual for a question this personal, and it is worth starting there.

What the coin toss experiment actually did

Steven Levitt ran a large field experiment, published as an NBER working paper in 2016 and later in The Review of Economic Studies in 2021. People who could not decide something significant, with quitting a job and ending a relationship among the examples the paper gives, were invited to settle it with a coin toss. In most cases heads meant make the change and tails meant stay. More than 20,000 usable tosses came in.

Nobody was bound by the result. That is the point of the design: the coin does not force anyone’s hand, it just nudges, which lets Levitt use the toss as an instrument and get at causation rather than correlation.

Two findings came out of it. People who made a change, whichever way the coin landed, reported being substantially happier two months and six months later. That part is only correlational, as Levitt says plainly, because the people who change are not a random group. The causal part is the second finding: those whose coin came up heads were more likely to actually go through with it, and were happier six months on than those told to stay put.

His stated conclusion is careful and still striking. In his words, the results suggest that people may be excessively cautious when facing life-changing choices.

Why this is not a licence to quit everything

The design has real limits, and they matter more than the headline usually admits.

Everyone in it was undecided enough to go looking for a website that would flip a coin for them. That is not a cross-section of humanity. It is a self-selected group already circling the exit, and the finding applies to that group. The outcome measure is self-reported happiness, which is what people say rather than what happened to them, and roughly two in five did not answer the follow-up at all.

Compliance was also partial. Nobody obeyed the coin as a rule, and heads-flippers were only about a quarter more likely to make the change than tails-flippers. The effect is real and it is not enormous.

None of this makes the result less interesting. It does mean the finding is that a particular set of people, already looking for a push and then given one, tended not to regret it. That is different from a rule that leaving is generally right.

The reason leaving is hard has a name and a date

The mechanism was pinned down forty years earlier, and the experiment is still one of the cleanest in the field.

In 1976 Barry Staw, then at Northwestern, published Knee-Deep in the Big Muddy in Organizational Behavior and Human Performance. He gave 240 business undergraduates at the University of Illinois a fictional company and 10 million dollars of research funding to award to one of its two divisions. Five years on in the scenario, with another 20 million to hand out and free rein to split it, they saw how their pick had performed.

Half had made the original choice themselves. Half were told a previous financial officer had made it.

The students who had personally chosen a division and then watched it decline put an average of 13.07 million of the second 20 million back into it. That was more than any other group in the study. Choosing badly yourself did not make people retreat from the failing option. It made them double the bet on it.

Staw opens the paper with a memo George Ball sent Lyndon Johnson in July 1965, quoted from the Pentagon Papers, warning that once large casualties were taken the United States could not stop short of its objectives without national humiliation. Ball’s word for what would trap them was not strategy or cost. It was humiliation. The paper’s argument is that the same machinery runs in a business school classroom with imaginary money, and that what people are protecting when they refuse to move on is the appearance that they were right.

Hal Arkes and Catherine Blumer put a slightly different frame on it in The Psychology of Sunk Cost, in Organizational Behavior and Human Decision Processes in 1985. In one of their field studies, the first 60 people to buy season tickets to the Ohio University Theater were randomly handed an unannounced discount, or no discount at all. Over the first half of the season the full-price group attended an average of 4.11 plays, against roughly 3.3 for the discounted ones, for the same performances. Their reading was that the driver is a desire not to appear wasteful.

Between them the two papers describe most of what makes moving on difficult, and neither has much to do with attachment. One is about not looking wrong. The other is about not looking wasteful.

Disengagement is only half of it

Here is where the popular version of this advice goes thin. It stops at letting go, as though the difficulty ends once you have released the thing.

Carsten Wrosch, Michael Scheier, Gregory Miller, Richard Schulz and Charles Carver looked at both halves in a 2003 paper in Personality and Social Psychology Bulletin. They separated the capacity to let go of a goal that has become unreachable from the capacity to commit to a different one, and reported that the two interact: the combination is associated with wellbeing in a way that neither fully accounts for on its own.

Take that carefully. It is three studies with 115, 120 and 45 participants, it is correlational, and it measures self-reported wellbeing. The result is a finding from these samples, not a universal rule about everyone. But the structural point survives the caveats, because it is really a point about what the word covers.

Moving on is two moves. Everything in the culture, and almost everything in the advice, is about the first one.

What this leaves you with

The useful synthesis is smaller than the phrase suggests, and more specific.

If you are already looking for a push, the one randomised study points towards change rather than towards staying, for people in roughly that position. If you cannot move at all, Staw and Arkes give a better explanation than sentiment does, and it is worth asking whether you are defending the decision or the person who made it. And if you have already let go and it has not helped, the disengagement literature suggests the missing piece is not more letting go.

What we keep noticing about all of this is how badly the language fits. Moving on sounds like a single act with a clean edge, something that either happened or did not. The evidence describes something more like a transfer, where the first half is easy to name and the second half is the part that decides how it goes.

The people who seem to do it well are rarely the ones who were best at letting go. They are the ones who already had somewhere for the effort to land.