What actually goes with procrastination (across 691 studies)

If you have been told you put things off because you are a perfectionist, the largest synthesis of this research is a strange kind of relief: it does not hold up. In a 2007 meta-analysis¹1 A meta-analysis pools the results of many earlier studies into one overall estimate, so it can say something none of them could alone. ↩ pooling 691 correlations from 216 separate works, perfectionism came out at about zero for starting late, and procrastinators turned out to intend to work about as much as everybody else. What did go with putting things off was more ordinary: how unpleasant the task feels, how far off the payoff is, and how easily you are pulled toward something nearer.

One paper, 691 correlations

The paper is Piers Steel's review in Psychological Bulletin: 691 independent correlations²2 A correlation is a number from -1 to 1 saying how closely two things move together. Around .1 is faint, .3 is noticeable, .5 and up is strong for this kind of research. ↩ from 216 works, double-coded and pooled (Steel, 2007, p. 73).

What that method shows is which things move together, not which moved first. Nearly every study in it measured procrastination by questionnaire, and most of the people asked were university students (p. 73). So read all of it as "this goes with that, across a lot of samples", and not as "this makes that happen".

The usual suspects

Perfectionism is the explanation most articles reach for, and it fares worst. Across 24 studies and nearly 3,900 people, its correlation with procrastination was −.03 (Table 3, p. 76).

Two qualifications, because leaving them out would swap one myth for another. That −.03 covers perfectionism aimed at yourself and at other people. The third kind, believing that other people have set the standard, sits at .18 and Steel counts it with the fear-of-failure measures rather than with perfectionism (p. 76, and Table 3 note b). And the whole result concerns delayed starting: someone holding themselves to a very high standard may begin on time and still be late finishing, because it is never good enough to hand over (p. 81).

Anxiety fares a little better and not much: fear of failure .18, irrational beliefs .17, neuroticism .24 (Table 3, p. 76). Weak rather than nothing. The interesting part is underneath. When the paper split neuroticism by what its measures were made of, the impulsive-flavoured ones averaged .33 and the rest .16 (p. 76), and its reading is that procrastination "does not appear to be anxiety related" once impulsiveness is accounted for, with fear of failure getting what relationship it has through low self-confidence (p. 81).

Show the data

The explanations that do not hold up

Label	Correlation with procrastination (observed)
Neuroticism	0.24
Fear of failure	0.18
Irrational beliefs	0.17
Intelligence or aptitude	0.03
Perfectionism	-0.03
Intention to work	-0.03

Adapted from Steel (2007), Tables 3 and 4, pp. 76 to 77

Observed correlations, uncorrected, with the number of studies behind each: neuroticism .24 (59 studies, N = 10,720), fear of failure .18 (57 studies, N = 10,785), irrational beliefs .17 (71 studies, N = 13,137), intelligence .03, perfectionism -.03 (24 studies, N = 3,884), intention -.03 (8 studies, N = 1,017). Anything under about .2 is weak. The perfectionism figure excludes socially prescribed perfectionism, which sits at .18.

What does go with it

The things that do track with it are duller and more useful. The strongest cluster is about self-regulation: conscientiousness at −.62 across 20 studies, self-control at −.58, distractibility at .45, and being organised at −.36 across 25 studies (Table 5, p. 78). Impulsiveness sits in the middle of it at .41 (Table 4, p. 77). The negative signs mean more of that thing goes with less procrastinating.

Then the plainest finding in the paper: the more unpleasant a task is, the more it gets put off, at .40 across 8 studies (Table 2, p. 75). Asked to pick from 26 possible reasons they might put off a term paper, the one students endorsed most, 45 percent of them, was disliking writing term papers (p. 75). The task is horrible and so it waits.

Show the data

What actually tracks with procrastinating

Label	Correlation, corrected for measurement error
Conscientiousness	-0.75
Self-control	-0.73
Distractibility	0.59
Impulsiveness	0.52
Task feels unpleasant	0.46
Self-efficacy	-0.46
Being organised	-0.45

Adapted from Steel (2007), Tables 2, 4 and 5, pp. 75 to 78

Corrected correlations, which adjust for the imprecision of the questionnaires and so run higher than the raw ones. Observed values, with studies behind each: conscientiousness -.62 (20 studies), self-control -.58 (21), distractibility .45 (13), impulsiveness .41 (22), task aversiveness .40 (10 for the trait measure), self-efficacy -.38 (39), organisation -.36 (25). Negative means more of that goes with less procrastinating.

Wanting was never the problem

Across eight studies, the correlation between how much people intended to work and how much they procrastinated was −.03. Across sixteen, the correlation between procrastination and a second questionnaire about putting things off was .52 (Table 5, p. 78). Both sides of that second number are the same person filling in two self-report scales, so it is not intention measured against observed behaviour (p. 79). It still says something: what people reported intending had no relationship with procrastinating, and what they reported doing had a strong one.

In the final hour, it is the procrastinator who is doing more work than intended.

People who put things off do not report wanting it less. The gap opens after the wanting, and widest when the deadline is furthest away: the paper reports the distance between intention and action growing the further ahead you plan, then shrinking and reversing as the deadline arrives (p. 79). Which is why the work happens the night before, and why it feels, at the time, like you suddenly care.

Where to look instead

None of this says that making a task less unpleasant will make you start sooner. It says where to look first, and what to try on yourself.

1. Make the first action smaller than feels sensible. The unpleasantness that goes with delay attaches to the task as you have described it. "Write the chapter" is unpleasant. "Open the file and write one bad sentence" mostly is not.
2. Bring an ending into view. The paper's model has motivation staying low while a reward is distant and climbing as it approaches (pp. 71 to 72). Anything that puts a finish line nearby, a timer included, is working on that rather than on your character.
3. Move what is nearer, not your willpower. Distractibility at .45 is one of the larger numbers here. Changing what is within reach is easier than out-arguing it.

What the paper does not license is dismissing the feelings. It says the opposite: that traits like fear of failure should not be dismissed but investigated as more distant causes, possibly working through self-confidence (p. 83). A later account by Sirois and Pychyl builds on ground Steel himself examined and found inconsistent, reading procrastination as short-term mood repair (p. 79)(Sirois & Pychyl, 2013). If that reading is right, the lever is how the task feels in the moment rather than how far off the payoff is.

Where Yappie fits

We make Yappie, a focus app with a pet beside the timer, so read this as coming from an interested party. It is built around breaking a task into steps small enough to begin and putting a visible ending on the near horizon, which is the pair of things this research points at. Whether that helps you is a question this paper cannot answer, and neither can we.

References

1. Steel, P. (2007). The nature of procrastination: A meta-analytic and theoretical review of quintessential self-regulatory failure. Psychological Bulletin, 133(1), 65 to 94. https://doi.org/10.1037/0033-2909.133.1.65↩
2. Sirois, F., & Pychyl, T. (2013). Procrastination and the priority of short-term mood regulation: Consequences for future self. Social and Personality Psychology Compass, 7(2), 115 to 127. https://doi.org/10.1111/spc3.12011↩