What Makes People Persist When Things Get Difficult?

Persistence is predicted better by structure than by character. The most famous answer to this question, grit, is a real finding that was heavily oversold: the largest meta-analysis of it found that grit correlates so strongly with conscientiousness that it may be redundant with it, and that it predicts performance only weakly. What predicts persistence more reliably is a plan attached to a situation, an environment arranged to make continuing easy, and confidence built from having actually done the thing before.

What Duckworth actually reported

Angela Duckworth, Christopher Peterson, Michael Matthews and Dennis Kelly introduced grit in the Journal of Personality and Social Psychology in 2007 (volume 92, issue 6, pages 1087 to 1101), defining it as perseverance and passion for long-term goals. The paper is careful and its samples are substantial: two adult samples of 1,545 and 690 for educational attainment, 138 Ivy League undergraduates for grade point average, two West Point classes of 1,218 and 1,308 for retention, and 175 competitors at the National Spelling Bee for final ranking.

Grit predicted outcomes over and above measures of talent. It accounted for an average of about 4 percent of the variance in success outcomes.

That last number is in the original paper. It is not a criticism smuggled in later. Four percent of variance is a real effect for a single self-report questionnaire and a modest one, and the distance between "explains 4 percent of variance" and the cultural claim that grit is the key to achievement was created almost entirely downstream of the research, in a book, a TED talk and a decade of school district programming. Duckworth herself has publicly opposed using grit scores for school accountability, which is worth stating because critiques of grit often imply she did not.

The meta-analysis

Marcus Crede, Michael Tynan and Peter Harms published the systematic review in the Journal of Personality and Social Psychology in 2017 (volume 113, issue 3, pages 492 to 511), under the title "Much ado about grit." It is frequently miscited as a Psychological Bulletin paper; it is not. They pooled 584 effect sizes from 88 independent samples covering 66,807 individuals.

Four findings matter.

The higher-order structure of grit was not confirmed. Grit is supposed to be a single trait built from two facets, perseverance of effort and consistency of interest. The data do not support treating them as one thing.

Grit is very strongly correlated with conscientiousness, an established personality dimension with decades of validation behind it. The construct validity question this raises is not rhetorical: if a new scale measures something an old scale already measured, the new name adds nothing.

Grit is only moderately correlated with performance and retention. The authors note directly that interventions designed to raise grit are likely to have only weak effects on success.

Perseverance of effort carries substantially more predictive weight than consistency of interest, and it explains variance in academic performance even after controlling for conscientiousness. Their conclusion is that whatever utility the construct has probably lives in the perseverance facet alone, which is to say in the half that most resembles ordinary diligence and least resembles the passion story.

The counter-evidence, presented fairly

Duckworth and colleagues published a large longitudinal study in PNAS in 2019 (volume 116, issue 47, pages 23499 to 23504), covering 11,258 cadets entering West Point across a full decade. Cognitive ability was the strongest predictor of academic and military grades. Noncognitive attributes, grit among them, were more predictive of two different outcomes: completing the brutal summer initiation training, and graduating within four years.

That result cuts both ways and should be read both ways. Grit did predict something real, and specifically the thing you would expect a persistence construct to predict, which is not quitting. It did not predict the outcomes institutions actually grade. And the sample is a self-selected elite, which restricts range on precisely the variables involved.

The defensible reading, holding both papers, is that grit is not nothing. It is smaller than advertised, largely captured by an existing personality dimension, most useful in its perseverance half, and predictive of attrition rather than achievement. That is a legitimate contribution. It is not a theory of success.

It is also worth noticing what a trait explanation cannot do for you. Conscientiousness is substantially heritable and stable in adulthood. If persistence is mostly conscientiousness, then telling someone to be grittier is telling them to have a different personality, which is advice with no available action attached.

What actually predicts persistence, ranked by evidence

1. A plan that names the situation. Gollwitzer and Sheeran's meta-analysis, published in Advances in Experimental Social Psychology in 2006 (volume 38, pages 69 to 119), pooled 94 independent tests across more than 8,000 participants and found a medium-to-large effect, d = 0.65, on goal attainment. Crucially, implementation intentions were found to work not only on initiation but on shielding an ongoing pursuit from interference and on disengaging from failing approaches. That is persistence, mechanically specified, in the research on if-then planning.
2. An environment that makes continuing the default. Structural interventions outperform motivational ones consistently, and they keep working on days when you do not feel like the kind of person the plan assumed.
3. Self-efficacy built from mastery experience. Albert Bandura's 1977 paper in Psychological Review (volume 84, issue 2, pages 191 to 215) identified four sources of efficacy belief and ranked mastery experience first by a wide margin: performing the behavior and succeeding. The ordering matters more than the list. Doing produces the belief; the belief does not have to come first, which is the practical consequence drawn out in the work on where confidence actually originates.
4. Reducing the aversiveness of the task itself. Task aversiveness is one of the strongest documented predictors of procrastination. Making a behavior less unpleasant is an intervention. Deciding to mind it less is not.
5. A pre-decided response to lapsing. In a 2021 megastudy of 61,293 gym members across 53 four-week programs, published in Nature, the single best-performing intervention was a small reward for returning to the gym after a missed workout. Not for streaks. For recovering from breaking one.
6. Trait grit. Last, because even granting the full effect, it is a stable disposition you cannot alter this month. Knowing your score tells you something. It does not give you anything to do.

The uncomfortable summary

The interventions with the best evidence are things you install once, not virtues you cultivate. A plan with a named cue. An environment arranged so the behavior is the path of least resistance. A first attempt small enough to succeed at, because succeeding is what builds the belief. A decided-in-advance response to the day you miss.

None of that is inspiring, and that is close to the point. The grit narrative is appealing partly because it locates the problem in character, which flatters people who succeed and quietly blames people who do not. The meta-analytic picture relocates it into circumstances and structures, which is less satisfying as a story and considerably more actionable.

Where character explanations do have a place is at the edges, on the long-horizon commitments no cue can carry. Even there, the more reliable move is usually to change what your future self is able to do rather than what they are willing to do, which is the logic behind binding yourself in advance.