The Gospel of Grit
In 2013, Angela Duckworth stepped onto the TED stage with a message the world was ready to buy: after studying cadets at West Point, finalists at the National Spelling Bee, and rookie teachers in tough schools, she announced that one trait predicted success above all others, and it was not IQ, social intelligence, looks, or health, but grit, passion and perseverance for long-term goals. Viewed tens of millions of times, the talk made her a MacArthur grant winner and bestselling author, and schools across America built curricula and accountability tests around the idea that success was learnable rather than talented.
It was also, according to a meta-analysis in the Journal of Personality and Social Psychology, mostly wrong.
An Audit of 88 Studies
In 2017, Marcus Credé of Iowa State University and two colleagues audited the literature: 584 effect sizes from 88 independent samples, 66,807 people in all, asking what the TED talk had skipped, how strongly grit predicts performance and whether it measures anything distinct from conscientiousness. Published after peer review, the answers were deflating.
The Effect Is Modest, and the Famous 99 Percent Was Three Points
Overall, grit's pooled correlation with performance was 0.18, a real effect but a modest one, explaining about 3 percent of the variation in performance, while NPR's coverage noted the correlation between SAT scores and college grades runs roughly 0.50; Duckworth did not dispute the number, telling NPR her findings sit in the "small-to-medium" range for personality psychology, an odd framing beside a TED talk that had presented grit as the trait above all others.
Credé also caught statistical framing that misled thousands of readers. A 2009 paper reported that West Point cadets scoring one standard deviation above average on grit were "99% more likely" to complete Beast Barracks, which sounds transformative until you see the underlying reality: 95 percent of cadets complete Beast Barracks compared with 98 percent of the grittiest cadets, so the odds improved 99 percent while the completion rate moved three points. When NPR's Anya Kamenetz asked about it, Duckworth conceded that "Crede is right that I should have said [it that way]," adding that the tables were correct and the intent was not to mislead.
Grit Is Conscientiousness in a New Outfit
Most damaging was the overlap. Grit correlated with conscientiousness at 0.84 after correcting for measurement error, and here is a calculation the paper invites but never runs: square 0.84 and you get 71 percent shared variance, while two different established measures of conscientiousness correlate with each other at only 0.63, which squares to 40 percent shared variance, so grit resembles conscientiousness more closely than conscientiousness resembles itself across questionnaires; the perseverance-of-effort facet alone correlated 0.89 with conscientiousness, psychologists call this the jangle fallacy: an old construct in a shiny new package.
What matters is incremental validity. Controlling for conscientiousness, overall grit explained essentially zero additional variance in overall academic performance, two-tenths of a percent in high school GPA, and four-tenths of a percent in college GPA. It knows almost nothing the conscientiousness score does not already know.
Its structure failed too: grit is supposed to fuse perseverance of effort with consistency of interests, yet the two facets barely hung together, the "passion" half predicted almost nothing on its own, and adding it to perseverance reduced predictive power.
The One Facet That Earns Its Keep
Fairness requires the other half of the story. Perseverance of effort, alone, added real variance beyond conscientiousness: 4.0 percent for overall academic performance, 8.5 percent for high school GPA, 2.3 percent for college GPA, respectable numbers in personality psychology, and Credé's own verdict was that "the primary utility of the grit construct may lie in the perseverance facet."
The Strongest Counterargument
Duckworth's defense deserves its full weight. Start with measurement: the meta-analysis pooled the validated Grit Scale with brief ad hoc measures of unknown quality, and averaging careful and sloppy measurement can smear a real signal; the 0.84 correlation may partly reflect questionnaires that were conscientiousness items wearing grit labels from the start.
Then the finish line: grit was built to explain survival in extreme, long-horizon selection funnels like Beast Barracks and the spelling bee finals, not semester GPA among already-selected college students, and range restriction, which Credé lists as a limitation in his own paper, shrinks every correlation in such samples; the passion facet cannot demonstrate its value in a four-month window by construction, and might even be counterproductive where abandoning a failing approach is the right move.
Deepest is the conceptual objection, and Duckworth eventually made it herself: in a 2021 commentary she and her coauthors conceded the factor-analysis critique outright, admitting the supposed higher-order structure of the Grit Scale was a mathematical artifact they had misread; but they argued the construct question cannot be settled by questionnaire math, since staying committed to one ambitious goal for years may be genuinely different from being organized and dutiful even if no questionnaire separates the two cleanly.
A bad map does not prove the territory is empty.
She also has a record of candor: in 2015 she told NPR that "the enthusiasm is getting ahead of the science," and with David Yeager she warned schools against using grit scores for high-stakes accountability.
What We Didn't Prove
We did not prove grit is useless. The perseverance facet predicts performance beyond conscientiousness, and in high-stakes selection a few points of explained variance can matter.
We did not test grit over decades. The pooled literature is dominated by concurrent designs, a limitation Credé flags himself because it can inflate correlations; the long-horizon test the theory demands has never been run at scale.
We did not prove grit cannot be taught. The meta-analysis tested whether grit scores predict performance, not whether interventions change grit or outcomes, and the intervention evidence is thinner and more mixed.
We did not settle whether the questionnaires capture the right thing. Self-reported grit is fakable: as Credé told Inside Higher Ed, students know how to present themselves as gritty, and a flawed instrument can hide a real trait.
The Bottom Line
The grit gospel promised a newly discovered trait that beats talent. Available evidence says the questionnaires mostly measure conscientiousness, a trait described in the 1930s, while the genuinely novel half of the theory predicts almost nothing. What survives the audit is not nothing: grinding through setbacks helps. We just already had a word for it.
What You Can Do
If you run a school, do not buy a grit curriculum as a distinct product; the evidence favors study skills, deliberate-practice routines, and attendance habits, the unglamorous machinery of conscientiousness.
If you hire, know that a grit score tells you nothing a conscientiousness measure does not, and both are easy to fake. Work samples and structured interviews carry more signal per minute.
If you are a parent, keep the message and drop the label: telling a child not to quit when it gets hard is good advice, and it does not need a trademarked trait to work. If the TED talk moved you, keep what was true in it. Sustained effort matters enormously, even if it was never a new discovery.