In a 2005 study of 140 eighth graders, self-discipline measured months in advance predicted final grades about twice as strongly as IQ did, one clue to why comfort and long-term performance often pull in different directions

In 2005, the psychologists Angela Duckworth and Martin Seligman published a study in Psychological Science of 140 eighth graders that has been cited more than a thousand times since. Self-discipline, measured months before the school year ended through self-report, parent report, teacher report, and a task in which students chose between a smaller reward now or a larger one later, predicted final grades, school attendance, and selection into a competitive high school programme the following spring. A replication with 164 more students added a behavioural delay-of-gratification test and a group IQ test. Self-discipline accounted for more than twice the variance in final grades that IQ did, and the effect held even after controlling for the students’ grades from earlier in the year.

This is one pair of studies from one research group, working with adolescents in a specific school setting, not a universal law about effort and comfort. It is still one of the more carefully built cases in psychology for a claim that gets asserted constantly and demonstrated rarely: that the willingness to do the harder, less comfortable thing now shapes outcomes more than raw ability does.

What the study actually measured

Duckworth and Seligman were not testing willpower in the abstract. Their self-discipline measure combined how students, their parents, and their teachers described the students’ behaviour with an actual choice: take a dollar today or wait a week for two. Students who consistently chose to wait, who reported doing homework before watching television rather than after, and whom teachers and parents independently described as disciplined, tended to finish the year with better grades regardless of how they scored on an IQ test.

The paper’s own framing is blunt about what this does and does not show. It suggests, in the authors’ words, a major reason students fall short of their intellectual potential: not a lack of ability, but a failure to exercise self-discipline in the ordinary, unglamorous moments when comfort and effort pull in different directions. It does not show that discipline can be installed like software, or that every student who struggles is simply choosing comfort over reward.

Why comfort is a poor guide to what is working

A separate line of research helps explain why the more comfortable option so often feels like the better one, even when it is not. The psychologists Robert Bjork and Elizabeth Bjork have spent decades studying what they call desirable difficulties: conditions such as spacing out practice, mixing up topics rather than blocking them, and testing yourself rather than rereading, all of which slow down apparent progress while improving what is actually retained.

Their research repeatedly finds that people misjudge their own learning because ease of performance in the moment feels like evidence of mastery. Massed, repetitive practice produces fast, comfortable gains that fade quickly. Spaced and varied practice feels harder and slower, and lasts. In one of their studies, participants who had just performed measurably better under harder, spaced conditions still said they believed the easier, blocked method had worked better for them, and said they would choose it again. Comfort, in other words, is not simply a lower-effort path to the same result. It can actively mislead the person choosing it about which path is working.

Where the story gets more complicated

It is worth being careful here about which construct is being discussed. Self-discipline, as Duckworth and Seligman measured it in 2005, is a fairly narrow behavioural pattern: delaying gratification, following through on routine tasks. A few years later, Duckworth introduced a related but broader idea, grit, defined as sustained passion and perseverance toward long-term goals, which became far more widely known through her books and talks.

Grit’s public reputation outran what a closer look at the evidence supports. A 2017 meta-analysis in the Journal of Personality and Social Psychology by Marcus Credé, Michael Tynan, and Peter Harms, pooling 584 effect sizes from 88 samples and more than 66,000 people, found that grit correlates only moderately with performance and retention, correlates very strongly with the existing personality trait of conscientiousness, and that its proposed structure as a distinct higher-order trait was not well supported by the data. The authors concluded that efforts to boost grit specifically are likely to have only weak effects on success. That does not overturn the original 2005 self-discipline findings, which measured something more specific and were not the subject of that meta-analysis, but it is a useful caution against letting one appealing result expand into a broader claim than the evidence can carry.

There are other limits worth naming plainly. Both Duckworth and Seligman’s samples were adolescents in American schools, not adults choosing how to spend a working afternoon. The measure combined several kinds of self-report with one behavioural task, and self-report is not the same as observed behaviour over time. A correlation, however striking, between fall self-discipline scores and spring grades does not rule out some third factor, a stable home environment or a particular teacher, doing some of the work. The paper is transparent about this. Later researchers built on it precisely because a single well-designed study, however useful, is a starting point rather than an answer.

None of this is an argument for discomfort as an end in itself. Bjork and Bjork are careful to distinguish desirable difficulties, ones a learner can actually work through and recover from, from difficulties that are simply too hard and produce no learning at all. A task that is uncomfortable because it is poorly designed, or because someone lacks the background to attempt it, is not made more valuable by its discomfort.

What this suggests for people who work under their own direction

None of this converts into a formula. The 2005 study measured school-age adolescents in a specific academic setting over a single year, using self-report and parent report alongside a lab task, and correlation is not the same as proof that discipline alone caused better grades. But the pattern it documents, that the easier-feeling option in the moment is a poor predictor of the better outcome later, shows up again in the separate literature on learning and practice, measured a different way, in a different population, by a different research group. That kind of convergence across otherwise unconnected studies is a more useful signal than any single paper on its own.

For anyone deciding how to spend an unsupervised hour, on a task, a skill, or a body of knowledge, the two literatures point in the same direction without quite proving the same thing: the version of the work that feels easiest to do right now is not a reliable guide to which version is actually paying off.