Grit

What it is

The construct is about duration. Not effort in the moment and not enthusiasm in the moment, but the same interest and the same effort held to over years, through plateaus, boredom and setbacks, in pursuit of a single ultimate concern. A gritty person is not someone who works hard today. It is someone still working on the same thing in five years.

It has two named facets. Consistency of interest, and perseverance of effort. The original theoretical claim was that the two belong together as one higher-order trait and that both do comparable predictive work. That second half is the part the literature has taken apart.

The instrument, in the form the book prints, is ten items scored on a five-point agreement scale and divided into two five-item subscales. The five consistency items are reverse-scored, because each is worded as a statement of the thing grit is not, a new idea displacing an old one, a project abandoned for a more interesting project. The five perseverance items are scored forward: setbacks not discouraging the respondent, work finished once begun. The grit score is the mean of all ten, and each subscale score is the mean of its own five.

Two relatives circulate and are worth keeping apart. The original 2007 instrument is twelve items in two six-item subscales. The Short Grit Scale of 2009 is eight items in two four-item subscales, and Duckworth has since said in print that shortening it may have been unwise, because duration is what distinguishes grit and the subscales separate more cleanly on the longer form.

This publication does not reproduce the items, and does not reproduce the percentile table that accompanies them in the book. Duckworth's own laboratory asks that book authors not reproduce the scales, and the norming sample behind the published percentiles is given no size, no date and no citation anywhere, so no reader can be told what their percentile means.

In effect

The supporting evidence is strong and anyone dismissing grit outright has not read it.

At West Point, the founding study surveyed almost the whole of one entering class in the first days of the academy's brutal summer initiation and found that grit predicted who completed it. A second class replicated it. A third replicated it again in the paper that introduced the short scale. In all three cohorts the academy's own admissions composite failed to predict who dropped out. Then, in 2019, Duckworth herself co-authored the study that ought to settle the retention question: preregistered, data posted, more than eleven thousand cadets across nine cohorts and a full decade, with the significance threshold tightened for multiple comparisons. Grit was the only reliable predictor of completing initiation training. Neither cognitive ability nor physical ability reached the paper's own threshold. That is the founding claim confirmed on a sample nine times the size of the original, by the originator, under preregistration.

At the National Spelling Bee, grit predicted how far finalists went, and on a subset tested for verbal intelligence grit and verbal ability were unrelated, which is the independence claim in its cleanest form. A second study designed with Anders Ericsson found that solitary deliberate practice predicted advancement strongly, that being quizzed by others did not survive the full model, and that reading for pleasure was a flat null, which the book reports honestly against its own interests. In a sales company where more than half the staff had gone by follow-up, grit predicted who stayed and was the only personality variable to reach significance, with conscientiousness falling to nothing once grit was in the model. Among nearly five thousand Chicago school juniors, grit predicted graduation more strongly than caring about school, than conscientiousness about studies, and than feeling safe at school. And on malleability, two independent registered randomised trials in Turkish primary schools moved standardised mathematics scores by about a fifth of a standard deviation, still detectable two and a half years later. That result cuts in Duckworth's favour and belongs in the record.

Three rescopings travel with that evidence every time, because the primary papers contain them and the book does not. In the sales study the largest predictor of retention, by an order of magnitude, was not a trait at all but how long the employee had already been there. In Chicago, standardised achievement test scores predicted graduation far more strongly than grit did, and being female predicted it more strongly too, and the book names the three variables grit beats and neither of the two that beat it. And in the special forces selection study used to ask what distinguished the men who made it through, general intelligence and physical fitness both predicted completion more strongly than grit did. The honest version of that study is that grit added incremental prediction over intelligence and fitness, which is what the paper shows and is still a real result.

The paper the field turned on is Marcus Credé, Michael Tynan and Peter Harms, published in the Journal of Personality and Social Psychology in 2017. It pooled five hundred and eighty-four effect sizes from eighty-eight independent samples drawn from seventy-three studies, representing more than sixty-six thousand individuals, and tested for publication bias and source bias and found neither. Four findings are load-bearing. First, grit and conscientiousness are very nearly the same measurement: the true-score correlation between them sits at a level conventionally taken to mean two names for one thing, and the authors say in their own words that it is so strong as not only to limit the incremental value of grit scores but to put the distinctness of the construct in question. Second, the two-facet structure does not hold: consistency and perseverance correlate substantially but with a credibility interval wide enough that the authors call the relation heavily moderated, and the relation is markedly weaker on the original twelve-item form than on the short one. Third, the facets are not equal partners, and the original paper said they were: against every academic criterion tested, the perseverance facet predicts far more strongly than the consistency facet, and the difference is significant on all four. That asymmetry has since replicated in a 2022 meta-analysis covering one hundred and thirty-seven studies and more than a quarter of a million people. Fourth, and decisively, in the meta-analytic regressions overall grit explains no additional variance in either overall academic performance or school grade point average once conscientiousness is controlled, and a negligible amount in college grade point average, while conscientiousness explains incremental variance over grit. The asymmetry within grit is just as sharp: perseverance of effort does add substantially and holds up controlling both conscientiousness and consistency, while consistency of interest adds essentially nothing and shows a possible suppressor effect. Their own conclusion is that interventions to raise grit may have only weak effects on performance and success, that the construct validity of grit is in question, and that its primary utility may lie in the perseverance facet. One fairness point travels with the meta-analysis: Credé and colleagues note themselves that the retention studies had very high base rates of retention, and treat that as a boundary on the retention results rather than as grounds for dismissing them. The surrounding record points the same way, with a United Kingdom twin study of more than two thousand pairs reporting a genetic correlation between grit and conscientiousness close to unity, and further papers arguing that the overlap is with self-control as much as with conscientiousness and that the two-factor structure is an artefact of mixing positively and negatively worded items.

Duckworth's published response is concession rather than rebuttal, and recording it at full strength is a matter of fairness, because the concessions are the most important facts about the construct. In a peer-reviewed commentary in 2021 she and her original co-author conceded the higher-order structure outright, stating that the two-factor hierarchical model cannot be distinguished from a model with two correlated subfactors and no higher-order factor at all, and that their argument for grit as a compound of related but distinct dispositions should not have rested so heavily on the optimal factor solution for a questionnaire. In a review chapter on grit at work in 2020 she and her co-authors cite Credé and colleagues by name, acknowledge that the novelty of grit has been questioned, and answer by conceding the frame and narrowing the claim. Her words are that they conceive grit as a facet of conscientiousness. What she retains is the narrower point that grit alone specifies the compound of passion and perseverance sustained over months and years, and the same piece concedes that shortening the scale may have been unwise. And the most deflationary sentence about grit anywhere in the peer-reviewed record is hers: the 2019 West Point paper, having found grit the only reliable predictor of completing initiation training, reports that the whole battery of cognitive and non-cognitive attributes plus demographics explained under four per cent of the variance in graduating four years later, and adds that there may be severe limits to how well any set of personal attributes can forecast an individual's destiny.

So the verdict is that grit is a real and modestly predictive disposition that sits inside conscientiousness rather than beside it. The perseverance half does useful work, especially on the question of who stays and who quits in high-attrition settings, where it is repeatedly and prospectively confirmed. The passion half does very little. The claim that grit is a distinct trait outperforming conscientiousness is not supported by the meta-analytic record, and Duckworth's own published position since 2020 no longer asserts it.

One standing rule governs any use of the instrument, and it is to Duckworth's credit. The Grit Scale is a research and self-reflection instrument. This publication never describes it as a selection tool and never cites the book as licence for such use, because Duckworth has stated a clear, early, public and repeated position against exactly that. Her laboratory's own distribution page discourages use of these scales in high-stakes settings where faking is a concern, naming admissions and hiring decisions. Writing in a newspaper in 2016 she said she worried she had contributed inadvertently to an idea she vigorously opposes, high-stakes character assessment, and that schools should not be rewarded or punished on these measures. In a peer-reviewed article with David Yeager the year before, she and her co-author concluded that no available measure of personal qualities is suitable for between-school accountability judgements. The book itself contains no such caution and reports three selection contexts approvingly, and both facts are stated together every time.

What it does not say

It does not say grit outperforms conscientiousness. On the meta-analytic evidence it adds nothing over conscientiousness for academic performance, while conscientiousness adds over it, and the originator's own published position since 2020 conceives grit as a facet of conscientiousness rather than a rival to it.

It does not say the two facets are equal. Perseverance of effort carries the predictive work. Consistency of interest adds essentially nothing and may act as a suppressor, and that asymmetry has now replicated on a sample of more than a quarter of a million people.

It does not say the supporting evidence was wrong. The West Point retention finding is prospective, preregistered on more than eleven thousand cadets, and confirmed. What it supports is narrower than the book's framing, and the supportable statement is that talent composites predict who performs while grit predicts who stays. In the same founding paper the admissions composite predicts military performance and academic grades far better than grit does, and with the composite controlled grit does not predict academic grades at all.

It does not say a personal quality forecasts a life. Duckworth's own preregistered paper reports that a whole battery of cognitive and non-cognitive attributes plus demographics explained under four per cent of the variance in graduating four years later, and says in terms that there may be severe limits on how well any set of attributes can do this.

It does not licence using the Grit Scale to select people. Duckworth has published against that use, her laboratory's distribution page discourages it in high-stakes settings where faking is a concern, naming admissions and hiring, and she has said in public that she worries she contributed inadvertently to an idea she opposes.

And it does not let a reader of the book alone evaluate the dispute. The book states no effect size, no correlation, no odds ratio and no coefficient anywhere, for any grit prediction, across more than three hundred pages of main text, while making comparative claims against conscientiousness. The entire external argument is about precisely those magnitudes.

---