If you have read one piece of advice about deadlines, it probably came from this study. Break the big deadline into smaller ones. Space them evenly. If nobody sets them for you, set them yourself; it helps, though not as much. Books, talks and productivity apps have repeated it since 2002.
The study behind it is "Procrastination, deadlines, and performance: self-control by precommitment" by Dan Ariely and Klaus Wertenbroch. On 2 September 2026 the journal that published it, Psychological Science, retracted it. We planned this article as a plain summary of what the experiment found. It has to be a different article now, and a more useful one: what can you still say about deadlines you set for yourself?
What the 2002 paper claimed
The paper had two studies.
Study 1, a classroom. Two sections of a course each wrote three short papers, with a 1% grade penalty for each day late. One section was given evenly spaced due dates. In the other, students chose their own dates at the start of term and were held to them. The paper reported that most students did choose dates before the last day, that the section with fixed spaced dates got better grades, and that students who spread their own dates out did nearly as well as that section.
Study 2, proofreading. Sixty people had three weeks to proofread three texts, paid 10 cents for each error found and docked a dollar for each day late. A third had a deadline every seven days, a third set their own, a third had one deadline at the end. The reported order was clear: spaced deadlines best, self-set in the middle, one final deadline worst.
From this came three ideas that travelled: people know they procrastinate and will pay for a commitment; self-set deadlines help; evenly spaced ones help most.
What went wrong
The replication. Kyle Hyndman and Alberto Bisin ran the proofreading study again, pre-registered, with 124 participants (about twice the original). The deadline arrangement made a negligible difference to errors found, days late and pay. Spaced deadlines did not stand out. A near-identical run with 120 people a year earlier gave the same answer. Their paper appeared in Psychological Science in July 2026.
The data. The research-integrity blog Data Colada examined spreadsheets of the original data that had been emailed from Ariely's account to Hyndman in 2006. For Study 2 they reported, among other things, that 18 of the 20 people in the end-deadline group had an exact twin elsewhere in the same group, with identical scores on all three texts. For Study 1 they reported that 13 of 49 grades in the choose-your-own section differed from what the grade components add up to, almost all in the direction that favoured the paper's claim. Their conclusion was that the data had been tampered with or fabricated.
The response. Wertenbroch said he never had access to the data and asked the editor to retract the paper. Ariely said the data contained serious anomalies, that his records and memory could not explain them, and that he would cooperate with a retraction. The journal retracted the paper on 2 September 2026.
A retraction means the paper can no longer be used as evidence. It does not prove the opposite. So the next question is what the other experiments on deadlines found.
What other deadline experiments found
People do want deadlines. This part held up. In the 2025 replication, about 93% of people who were allowed to set their own deadlines for the first two texts set them earlier than they had to, on average about twelve and eight days before the end. Bisin and Hyndman found the same strong demand for commitment in a 2020 field experiment with students. We know we put things off and we reach for a date to stop ourselves.
The dates did not make them do better. In that 2020 experiment, deadlines did not raise the share of students who finished; with hard deadlines some did worse, because they left the work until close to the date and then ran out of time.
Interim deadlines can backfire. Nicholas Burger, Gary Charness and John Lynham (2011) paid students $95 to log 75 hours of study over five weeks. Half had only the final target. Half also had to hit about twelve hours in each of the first four weeks. More than 60% of the first group finished. About 40% of the group with weekly targets did. One bad week ended it for them.
A far-off deadline may be worse than none. Stephen Knowles and colleagues (2022) invited people to fill in a short survey that triggered a charity donation, with one week, one month or no deadline. Responses were lowest with one month. With one week, or with no date at all, many people did it right away.
Setting a date feels like progress. The experiments say it is a wish until something else changes.
What to take from it
None of these studies settles the matter, and each used a different task. Taken together, they point in one direction: a date on its own does little. What seems to matter is how close it is, whether anything real hangs on it, and whether it leaves room for a bad day.
1. Put dates on things that have one
The delivery is Thursday. The inspection is on the 14th. The client needs the quote before their Monday meeting. Those dates carry weight because someone else is waiting. If every task on a list has a made-up date, the real ones are lost among them and the overdue count stops meaning anything.
2. If you invent a date, make it soon and tell someone
The survey experiment suggests a date a month off is an invitation to wait. If a job has no outside deadline and you want one, pick a date this week and say it to a person who will ask: a partner, a client, the crew lead. A date only you know about is the weakest kind. (The experiments here do not test telling someone; this is our reading of why outside dates hold and private ones do not.)
3. Date the next part, not the whole job
"New menu, due 30th" gives you three weeks of not starting. "Cost the six mains, Friday" is a thing you can do or fail to do this week. This is what was sensible in the old advice about spacing, and it does not need the retracted study: a near date on a small, clear piece is easier to act on than a far date on a large one. It is also how you find out early that the whole job is running late (see the planning fallacy).
4. Leave slack: a missed checkpoint is not a failed job
In the study-hours experiment, weekly targets hurt because missing one meant losing everything. On your own list, the equivalent is the task that goes red, then gets ignored because it is already late. When a checkpoint slips, move it the same day and write why. A list of honest new dates is worth more than a list of dead ones.
5. Pair the date with a start
A deadline says when to finish. Most delay is about not beginning. For anything with a date, write the first small step and when you will do it: "After Tuesday's lunch service, I price the fish." That kind of if-then plan has much firmer evidence behind it than deadlines do; it is covered in why you put things off.
On a task list
Due date: only where the date is real or promised to someone.
Checklist: the parts of the job, so the next part is always visible.
Note: the first step and when you will take it.
Weekly review: look at every overdue task and either do it, give it a new date with a reason, or take the date off. Ten minutes; see one list and a weekly review.
On a team board: a due date plus one owner. A date with nobody's name on it is nobody's deadline.
What the research does not show
It doesn't show that deadlines are useless. External deadlines with real consequences obviously move people; these experiments are about extra, optional or interim ones.
It doesn't show that spaced deadlines never help. The one study that said they did has been withdrawn. The replication found no benefit for a proofreading task among students. Other kinds of work have not been tested this carefully.
The remaining studies are small and specific. Students logging study hours, people answering a survey, students doing lab tasks for small payments. A kitchen, a building job or a freelance contract may behave differently.
The retraction is about one paper. The journal's notice says the replication and the reanalyses called the findings into question. It makes no ruling on who did what, and neither do we.
Put a due date on the next part of the job, add the first step to its note, and check the overdue ones once a week.
Open tick-tock-doneSources
Ariely, D. & Wertenbroch, K. (2002). Procrastination, deadlines, and performance: self-control by precommitment. Psychological Science 13(3), 219-224. Retracted 2 September 2026: retraction notice
Hyndman, K. & Bisin, A. (2026). Replication of "Procrastination, deadlines, and performance: self-control by precommitment". Psychological Science. Working paper (PDF)
Simonsohn, U., Simmons, J. & Nelson, L. (2026). Artificial deadlines, part 1 and part 2. Data Colada 138 and 139
Retraction Watch (3 September 2026). Procrastination study retracted after sleuths find signs of data tampering
Ariely, D. (2026). Statement on the 2002 procrastination study
Bisin, A. & Hyndman, K. (2020). Present-bias, procrastination and deadlines in a field experiment. Games and Economic Behavior 119, 339-357. NBER working paper 19874
Burger, N., Charness, G. & Lynham, J. (2011). Field and online experiments on self-control. Journal of Economic Behavior & Organization 77(3), 393-404 (figures as summarised by Hyndman and Bisin).
Knowles, S., Servátka, M., Sullivan, T. & Genç, M. (2022). Procrastination and the non-monotonic effect of deadlines on task completion. Economic Inquiry 60(2), 706-720. Working paper
Cover: clipart from Openclipart, released under CC0 (public domain).