Every professional firm imposes interim deadlines on clients, and the usual justification for doing so traces back to one well-known experiment. That experiment is more interesting than the version people quote, and it has since been challenged on precisely the point the justification rests on.
Key Takeaway
The study asked whether people self-impose meaningful, costly deadlines, whether those deadlines improve performance, and whether people set them optimally. The answers were "yes" to the first two questions, and "no" to the third[1]. A subsequent replication concluded that the commonly received wisdom that exogenous spaced deadlines are an effective mechanism to limit procrastination is possibly false, and that it is important to go back to the drawing board[2].
Our Grades For These Claims
Applying the scheme from the first article in this series.
That people will self-impose costly deadlines is Grade B. It is a clean experimental finding and it is not the part under challenge.
That self-imposed deadlines improve performance is Grade B, on the same basis.
That evenly spaced externally imposed deadlines outperform self-imposed ones is Grade C. It is the original's headline result and a subsequent replication reports it may not hold.
Our position: this is the reverse of the usual pattern in this series. Normally the surprising headline survives and the mechanism does not. Here the modest findings look secure and the headline is the part in doubt.
A Note On Method
Everything here is verified to August 2026.
We obtained the published abstract in full from PubMed and the publisher, plus portions of the paper from the lead author's own institutional copy[1][3][4][6][7].
The results ordering reaches us through a document-hosting site carrying an optically scanned copy of the paper, which is legible on the key sentence but garbled elsewhere[5]. We flag it at each use and it is corroborated in substance by the replication's own description of the original.
The replication is a working paper hosted on an author's personal website[2]. We could not establish that it is peer-reviewed, and we say so at each use.
We report no effect sizes or sample sizes for either paper, because we obtained none.
This article reviews behavioural research. It is not tax, filing or practice management advice, and nothing here concerns any statutory deadline.
The Three Questions
The structure of the paper, which is unusually clear.
Ariely and Wertenbroch published Procrastination, deadlines, and performance: Self-control by precommitment in Psychological Science, 13(3), 219–224, in May 2002[1]. The authors were at MIT and INSEAD respectively[4].
The paper opens by observing that people delay writing up their research, repeatedly declare they will start their diets tomorrow, or postpone until next week doing odd jobs around the house, and that people also sometimes attempt to control their procrastination by setting deadlines for themselves[1].
Then the three questions, quoted: "(a) Are people willing to self-impose meaningful (i.e., costly) deadlines to overcome procrastination? (b) Are self-imposed deadlines effective in improving task performance? (c) When self-imposing deadlines, do people set them optimally, for maximum performance enhancement?"[1]
Two observations, ours.
The questions are separable and sequential, which is why the paper is quotable. Each has a clean yes or no, and the answers do not all point the same way.
And note the parenthesis in the first: meaningful, that is, costly. The test is not whether people say they will finish early. It is whether they will accept a penalty for not doing so, which is a considerably harder thing to demonstrate.
The Design
The experiment, from the author's own hosted copy.
Participants had three texts to proofread within a three-week window, under one of three conditions[4].
Evenly spaced deadlines: participants had to submit one of the three texts every 7 days.
End deadline: they had to submit all three texts at the end of 3 weeks (21 days).
Self-imposed deadlines: they had to choose their own deadline for each of the three texts within the 3-week window.
The measures reported were mean errors detected, delays in submissions, and earnings[4].
Three observations, ours.
Everyone had the same total work and the same total time. Only the internal structure of the window differed.
The self-imposed condition permits the optimal answer. A participant could have chosen days 7, 14 and 21 and matched the evenly spaced condition exactly. Nothing stopped them.
And errors detected is an unusually good outcome measure. It is objective, it is not a self-report, and it maps onto something a client would actually care about.
The Answers
The result, quoted.
"A set of studies examined these issues experimentally, showing that the answer is 'yes' to the first two questions, and 'no' to the third. People have self-control problems, they recognize them, and they try to control them by self-imposing costly deadlines. These deadlines help people control procrastination, but they are not as effective as some externally imposed deadlines in improving task performance."[1]
Two observations, ours.
That is three findings in one sentence, and they are usually collapsed into the last one. The first two are about self-knowledge and are not in dispute.
And the final clause is carefully worded: not as effective as some externally imposed deadlines. Not all. The comparison is specifically with the evenly spaced condition, and the end-deadline condition was also externally imposed and did worst of the three.
The Ordering
How the three conditions ranked.
An optically scanned copy of the paper reads: "As predicted the number of errors correctly detected was highest in the evenly-spaced-deadlines condition, followed by the self-imposed-deadlines condition, with the lowest performance in the end-deadline condition." It adds that results were similar for participants' delays in submitting their proofreading work, in which case shorter delays produced higher payoffs, and that earnings reflected a combination of error detection and delay[5].
Our source for that sentence is a scanned document with visible optical character recognition errors elsewhere in the file. The sentence itself is legible and is corroborated in substance by the replication's summary of the original[2], but we did not read it in a clean copy of the paper.
Two observations, ours.
Self-imposed deadlines sat in the middle. Better than nothing, worse than a good external structure, which is exactly what the three answers describe.
And all three measures moved together. This is not a case where quality improved and speed suffered, or where a single measure was cherry-picked.
What The First Yes Means
The finding that is least quoted and, on our reading, the most interesting. This section is our own analysis.
People were willing to accept costly deadlines on themselves. Not aspirational ones. Ones with a penalty attached.
Three consequences.
It establishes that the problem is not a failure of self-knowledge. Participants correctly predicted that they would delay, and correctly predicted that a penalty would help.
So the usual explanation for procrastination, that people underestimate how long things take, is not what this study found. That is the planning fallacy, which this publication has covered separately, and it is a different failure.
And commercially, it means a client who says "just chase me" is giving you accurate information about themselves. They are not being self-deprecating; they have diagnosed the problem and are asking you to supply the mechanism.
What The No Means
The third answer, and it is the uncomfortable one. Ours.
People self-imposed deadlines, the deadlines helped, and they still did not set them well. The optimal schedule was available and unchosen.
Two consequences.
Whatever is going wrong is not ignorance and not indifference. People diagnose the problem correctly, pay real money to address it, and still under-treat it.
Which means the practical gap is not in motivation or awareness but in calibration: knowing how much structure you need, rather than knowing that you need some.
That is a considerably more specific and more useful conclusion than "people procrastinate", and it points at a particular intervention, which the replication then complicates.
The Replication
The challenge, and it targets the headline.
A replication study reports that the original showed that a substantial number of people are willing to self-impose deadlines on themselves and that the deadlines were partially effective, and that their main result, however, is that performance is highest when evenly spaced deadlines are exogenously imposed on subjects[2].
It then states: "On the contrary, two studies conducted subsequently to Ariely and Wertenbroch (2002) provide" contrary evidence, with our source truncating[2].
And its own conclusion: the replication "implies that the commonly received wisdom that exogenous spaced deadlines are an effective mechanism to limit procrastination is possibly false. This leads us to believe that it is important to go back to the drawing board to develop new testable hypotheses for why deadlines" work, with our source truncating again[2].
Its three conditions were endogenous, being self-imposed; exogenous, being evenly-spaced intermediate deadlines; and no deadlines[2].
What It Targets And What It Leaves
Reading the challenge precisely. This section is our own.
The replication's stated conclusion concerns exogenous spaced deadlines. That is the original's third answer and its headline.
Three consequences.
The first two answers, that people will self-impose costly deadlines and that these help, are described by the replication itself as what the original showed, and its challenge is not directed at them.
So the finding at risk is the ranking, not the phenomenon. Whether imposed structure beats self-chosen structure is in doubt; whether either beats nothing is less so.
And that matters because the ranking is the commercially load-bearing part. A firm imposing interim deadlines on clients is acting on the claim that imposed beats self-chosen, which is precisely the claim the replication questions.
An Attrition Problem
Something visible in the replication's own text, reported because it cuts against the replication rather than for it.
The replication records that a number of subjects never logged in after the initial lab session, and that an even larger number, 32, logged in at least once to see the first page of the first task, of whom 19 completed at least one page of the first task and 15 of them found at least one error, averaging 9.8, with a minimum of 0 and a maximum of 28[2].
Two observations, ours.
That describes substantial attrition and very low engagement in an online replication. A study in which many participants never started the task is measuring something different from one in which everyone did.
And that is a reason to hold the replication loosely too, in both directions. A failure to find an effect in a sample where a large share of people did not engage is weaker evidence than a failure in a fully engaged sample. We are not dismissing it; we are noting that neither paper settles the matter.
The Status Of That Challenge
Being explicit about our sourcing, because it determines how much weight the challenge should carry.
The replication reaches us as a PDF hosted on an academic author's personal website[2].
Three things we could not establish.
Whether it is peer-reviewed or published.
Its sample size, its effect sizes, or the full results.
And the identity of the two subsequent studies it says provide contrary evidence, since our source truncates before naming them.
We report it because a replication of a famous finding, reaching a negative conclusion in its own authors' words, is material information for anyone acting on the original. But a working paper is not a settled refutation, and this series has been careful about that distinction throughout.
Where We Land
Our own reading.
Three propositions we think the evidence supports.
People know they procrastinate and will pay to prevent it. Not in dispute between the two papers.
Structure of some kind beats no structure. Also not the target of the challenge.
And whether imposed structure beats self-chosen structure is genuinely open. The original says yes and is well-known; a replication says possibly not and is a working paper with visible engagement problems.
Our summary: the case for having deadlines is stronger than the case for choosing them for somebody else, and a firm that has been justifying the second by citing this literature has been leaning on the shakiest part of it.
The Client Who Arrives In March
The application, and it is our own extension to a setting the research did not examine.
The pattern is universal in professional practice. Records arrive late, in a rush, and the client is not surprised by this.
Three observations that follow from the parts of the finding not under challenge.
The client's self-diagnosis is likely accurate. On the first answer, people correctly recognise their own self-control problems, so a client who predicts they will be late is providing good information rather than an excuse.
The client would plausibly accept a costly commitment, since that is what the first answer establishes people are willing to do. A fee structure, a scheduling penalty or a stated consequence is not necessarily unwelcome.
And on the third answer, if it holds, whatever schedule the client proposes will probably be too loose. Not absent, just insufficient.
We flag that the research concerned proofreading tasks by students and executive education participants, not clients assembling a year of records, and that this section is extrapolation.
The Interim Deadline Question
The practical question this article was written to address, answered honestly. Ours.
Should a firm impose interim deadlines on clients?
Three points.
The original research says yes, and evenly spaced ones. That is the finding under challenge.
The replication says the received wisdom on that is possibly false and calls for new hypotheses. It does not say interim deadlines are harmful; it says the evidence for their superiority is weaker than believed.
And the parts not under challenge still support asking the client to commit to something, and to something with a consequence attached, because that is what people are demonstrably willing to accept.
Our own view: the honest position is that imposing structure is a reasonable practice with weaker evidential support than commonly claimed, and it is cheap enough that the evidential gap probably does not change the decision. That is a different justification from the one usually offered, and it is the one the evidence actually supports.
The Word Costly
The detail most likely to be dropped in application. Ours.
The first question specifies meaningful, that is, costly deadlines[1]. The finding is about deadlines with a penalty.
Three consequences.
A deadline with no consequence is not the thing that was studied. Asking a client to send records by a date, with nothing following if they do not, is outside the finding entirely.
Which means most interim deadlines in professional practice are not instances of this research at all, and citing it for them is a category error.
And the consequence does not have to be punitive. A stated and honoured effect on scheduling, on turnaround, or on what can be guaranteed is a real cost without being a penalty in any adversarial sense.
And The Firm Itself
The application inward, which is where we would start. Ours.
Everything above applies to a practice's own internal work at least as much as to clients.
Three observations.
A firm with one deadline at the end for internal work is running the condition that performed worst of the three in the original study.
Partners and staff will self-impose schedules, and on the third answer, if it holds, those will be too loose. Not absent.
And the firm has something a client does not: the ability to observe the pattern over years. Whether work distributed across a period produces fewer errors than work concentrated at the end is answerable from a firm's own records, and that answer is worth more to that firm than either published study.
What To Do
Take a client's self-diagnosis seriously. On the finding not under challenge, people correctly recognise their own self-control problems, so "just chase me" is accurate information rather than a joke.
Note that a deadline without a consequence is not what was studied. The research concerns costly deadlines, and most interim dates in practice have nothing attached.
Expect self-chosen schedules to be too loose rather than absent. That is the third answer, and it points at calibration rather than motivation.
Do not cite this literature as settled support for imposing deadlines. A replication concludes the received wisdom on exogenous spaced deadlines is possibly false and calls for going back to the drawing board.
Do not treat the replication as settled either. It is a working paper we could not confirm is peer-reviewed, with substantial attrition visible in its own text.
Separate the planning fallacy from procrastination. One is misestimating duration and the other is knowing and delaying anyway, and they call for different responses.
Apply it internally first. A single end deadline for internal work is the condition that performed worst in the original.
Check your own records. Whether spreading work reduces errors in your practice is answerable from data you already hold, and it beats both papers for your purposes.
The Limits Of This Analysis
Several caveats matter. This article reviews behavioural research and is not tax, filing or practice management advice; nothing in it concerns any statutory deadline. Everything is verified to August 2026. We obtained the original's published abstract in full and portions of the paper, but not a clean full text, and report no effect sizes and no sample sizes for it. The results ordering reaches us through an optically scanned copy with visible recognition errors elsewhere in the file; the key sentence is legible and corroborated in substance by the replication's description, but we did not read it in a clean copy. The replication is a working paper hosted on an author's personal website; we could not establish whether it is peer-reviewed or published, could not obtain its sample size or results, and its source truncates before naming the two subsequent studies it says provide contrary evidence. We report the replication's own attrition figures because they cut against it, and neither paper settles the question. The research concerned proofreading tasks by students and executive education participants; every application to clients or professional firms is our own extension to settings it did not examine. The distinction between the parts of the finding under challenge and those not, the analysis of what "costly" excludes, and all practical sections are our own reasoning, not findings. Roughly twenty-three years of subsequent literature was not reviewed.
Frequently Asked Questions
What did the original study find?
What is the significance of "costly"?
Has it replicated?
Should I trust the replication?
So should we still set interim deadlines for clients?
Is this the same as the planning fallacy?
References
- Ariely, D., & Wertenbroch, K. (2002). Procrastination, deadlines, and performance: self-control by precommitment. Psychological Science, 13(3), 219–224. DOI 10.1111/1467-9280.00441, published abstract via PubMed, on procrastination being familiar to most people who delay writing up their research, repeatedly declare they will start their diets tomorrow, or postpone until next week doing odd jobs around the house; on people also sometimes attempting to control their procrastination by setting deadlines for themselves; on the three questions posed, being whether people are willing to self-impose meaningful, that is costly, deadlines to overcome procrastination, whether self-imposed deadlines are effective in improving task performance, and whether when self-imposing deadlines people set them optimally for maximum performance enhancement; on the answer being yes to the first two questions and no to the third; and on people having self-control problems, recognising them, and trying to control them by self-imposing costly deadlines, which help people control procrastination but are not as effective as some externally imposed deadlines in improving task performance. Note: we obtained the published abstract in full; we did not obtain a clean full text and report no effect sizes or sample sizes. pubmed.ncbi.nlm.nih.gov
- Replication study of Ariely and Wertenbroch (2002), working paper hosted on an academic author's personal website, on the original having opened with three general observations, that most of us procrastinate, that deadlines are a common feature of life, and that many people seek commitment devices; on the original having shown that a substantial number of people are willing to self-impose deadlines on themselves and that the deadlines were partially effective; on the original's main result being that performance is highest when evenly spaced deadlines are exogenously imposed on subjects; on two studies conducted subsequently to Ariely and Wertenbroch providing contrary evidence, our source truncating before naming them; on the replication's three conditions being endogenous, exogenous and no deadlines; on the replication implying that the commonly received wisdom that exogenous spaced deadlines are an effective mechanism to limit procrastination is possibly false, leading the authors to believe it is important to go back to the drawing board to develop new testable hypotheses; and on a number of subjects never having logged in after the initial lab session, with 32 logging in at least once to see the first page of the first task, of whom 19 completed at least one page and 15 found at least one error, averaging 9.8 with a range of 0 to 28. Note: a working paper on a personal website. We could not establish whether it is peer-reviewed or published, could not obtain its sample size or full results, and report its attrition figures because they cut against its own conclusion. kylehyndman.com
- Publisher record for Ariely and Wertenbroch (2002), reproducing the abstract and the three questions identically, and confirming the citation, volume, issue, pages and date; and recording the corresponding address for the lead author at the Massachusetts Institute of Technology. Note: the publisher's own record, used as independent corroboration of the abstract. journals.sagepub.com
- Lead author's institutional copy of Ariely and Wertenbroch (2002), on the authors' affiliations at the Massachusetts Institute of Technology and INSEAD, Fontainebleau, France; on the evenly-spaced-deadlines condition requiring participants to submit one of the three texts every 7 days; on the end-deadline condition requiring all three texts at the end of 3 weeks, being 21 days; on the self-imposed-deadlines condition requiring participants to choose their own deadline for each of the three texts within the 3-week window; and on Study 2 reporting mean errors detected, delays in submissions, and earnings across the three conditions. Note: an author-hosted PDF; we obtained portions. web.mit.edu
- Document-hosting site carrying an optically scanned copy of Ariely and Wertenbroch (2002), on the number of errors correctly detected having been highest in the evenly-spaced-deadlines condition, followed by the self-imposed-deadlines condition, with the lowest performance in the end-deadline condition; on results having been similar for participants' delays in submitting their proofreading work, where shorter delays resulted in higher payoffs; and on participants' earnings reflecting a combination of error detection and delay. Note: a scanned copy with visible optical character recognition errors elsewhere in the file. The quoted sentence is legible, and is corroborated in substance by the replication's description of the original at reference 2, but we did not read it in a clean copy of the paper. scribd.com
- Semantic Scholar record for Ariely and Wertenbroch (2002), confirming the citation and reproducing the summary that the answer is yes to the first two questions and no to the third, showing that people have self-control problems, recognise them, and try to control them by self-imposing costly deadlines; and carrying indexed descriptions of related literature, including that procrastination is defined as postponing a task while anticipating adverse consequences in the future, and that a contrasting concept termed precrastination was introduced in 2014. Note: a bibliographic record with third-party indexed summaries; we obtained none of the related work described. semanticscholar.org
- Scholars@Duke institutional publication record for Ariely and Wertenbroch (2002), confirming the citation, journal, volume, issue, pages and date, and reproducing the concluding sentence that these deadlines help people control procrastination but are not as effective as some externally imposed deadlines in improving task performance. Note: an institutional repository record, used as a further independent confirmation of the citation and concluding claim. scholars.duke.edu
This article reviews behavioural research and is not tax, filing or practice management advice, and nothing in it concerns any statutory deadline. The original study was not obtained in a clean full text and no effect sizes or sample sizes are reported. Its results ordering reaches this article through an optically scanned copy with recognition errors elsewhere in the file. The replication is a working paper on a personal website which could not be confirmed as peer-reviewed, and whose own attrition figures are reported here because they weaken its conclusion. Every application to clients or professional firms is an extension to settings this research did not examine.