Marcus Credé

Who they are

Grit was proposed as perseverance of effort combined with consistency of interest over long periods, and as a predictor of achievement over and above the traits already known to predict it. The 2017 meta-analysis with Michael Tynan and Peter Harms tested those claims against the whole accumulated literature: 584 effect sizes from 88 independent samples drawn from 73 studies, representing 66,807 individuals.

Four results came out of it and the fourth is the one that moved the field. Grit and conscientiousness correlate at a level conventionally taken to indicate that the two constructs are redundant. The proposed two-facet higher-order structure is not confirmed. Perseverance of effort predicts academic outcomes considerably more strongly than consistency of interest does, which contradicts the original claim that both predict equally well. And in the meta-analytic regressions, overall grit explains no additional variance in academic performance once conscientiousness is controlled, while conscientiousness explains variance over and above grit.

Two features of his work must travel with it and are usually dropped. He tested for publication bias and for source bias and found neither, and he reported separate estimates for peer-reviewed and non-peer-reviewed studies which came out very close to one another. That is unusually careful and it strengthens the result considerably. And he supplies his own fairness caveat, noting that the retention studies had very high base rates of retention, which is a boundary on the retention results rather than a dismissal of them. That caveat has to travel with any quotation of him against grit.

His own conclusion is narrower than his reception. He holds that interventions to raise grit may have only weak effects, that the construct validity of grit is in question, and that the primary utility of grit may lie in the perseverance facet. He does not say grit is nothing, and this site does not say it on his behalf.

One further fact is recorded for completeness with no motive characterised. No reply from Angela Duckworth accompanied either the 2017 meta-analysis or the 2018 review, and none appeared in the journal that carried either. What she has published since is a series of concessions elsewhere. The single point she pushed back on publicly was the distance between his findings and the language of an accompanying press release, and that is recorded here because a page giving only one side of a disagreement is not doing its job.

He belongs in the same category on this site as Scott Lilienfeld and Michael Inzlicht: people inside the discipline who did the checking. That framing matters, because it is the fairest version of the replication story. The field corrected itself, and the correctors were insiders.

Key papers

Credé, M., Tynan, M. C., & Harms, P. D. (2017). "Much Ado About Grit: A Meta-Analytic Synthesis of the Grit Literature." Journal of Personality and Social Psychology, 113(3), 492-511. First author, and this vault's primary source on grit, read in full from the paper's own text. Five hundred and eighty-four effect sizes from 88 independent samples across 73 studies, representing 66,807 individuals. Grit correlates with conscientiousness at a level indicating construct redundancy; the two-facet higher-order structure is not confirmed; perseverance of effort predicts academic outcomes far more strongly than consistency of interest; and grit explains no additional variance in academic performance once conscientiousness is controlled. Publication bias and source bias were both tested for and neither was found.

Credé, M. (2018). "What Shall We Do About Grit? A Critical Review of What We Know and What We Don't Know." Educational Researcher. The follow-up review. Note the journal: it is Educational Researcher, not Educational Research, and the mistaken form appears in at least one popular secondary source. Verified 2026-09-04.

Key ideas

Construct redundancy. Where a new measure correlates with an established one at the level grit correlates with conscientiousness, the burden is on the new measure to show it is measuring something else.

Incremental validity is the test that matters. A predictor earns its place by explaining variance the existing predictors do not, and on the pooled evidence grit does not do so for academic performance.

The facets are not equal. Perseverance of effort carries the prediction; consistency of interest does not, which is contrary to the original claim.

The jingle-jangle problem in one case. Two names for one thing is as much a measurement error as two things under one name.

His caveats are part of his result. The publication-bias tests strengthen it, and the retention base-rate caveat bounds it, and neither may be dropped when he is quoted.

---