A growth mindset quiz is usually three or four sentences long. Intelligence is something basic about a person that cannot be changed very much. You can learn new things but you cannot really change how intelligent you are. Agree or disagree, on a scale.
That is the whole instrument. Everything anyone claims about mindset rests on a handful of agree-or-disagree statements about whether ability is the kind of thing that moves.
3 to 4
items in the scale most mindset research uses
1%
share of variance in academic achievement that mindset accounted for across 273 studies
0.08
average effect of mindset interventions on achievement, in standard deviations
97,672
participants in the meta-analysis that found the evidence base compromised
What the score is a score of
The measure is a belief, not a capacity. Someone who scores high is reporting that they think ability is developable. Nothing in the questionnaire observes whether their ability actually develops, how hard they work, or how they behave after a bad result.
This distinction gets lost constantly, and losing it is what produces the disappointed reader. A high score is not evidence of resilience. It is evidence that a person endorses a proposition about resilience, which is a much cheaper thing to endorse than to enact.
Carol Dweck and Ellen Leggett built the framework in 1988 around what happens after failure rather than around performance itself. The prediction was about interpretation: a person holding an entity theory reads a bad result as information about their ceiling, a person holding an incremental theory reads it as information about their method. That prediction is narrower and more interesting than the version that reached the general public, which is roughly that believing in yourself raises your grades.
How large the effect turned out to be
Large enough to be real, small enough that the promotional version was never defensible.
Sisk and colleagues pooled the literature twice in 2018. The first pool asked how strongly mindset correlates with academic achievement across 273 effect sizes and 365,915 people, and the answer was that mindset accounted for about one percent of the variance. The second pool asked what happens when a mindset intervention is delivered, across 43 effects and 57,155 students.
- All students: 0.08d
- Academically at risk: 0.19d
- Low socioeconomic status: 0.34d
The shape of that chart is the whole argument. Averaged over everyone, the effect is small enough to be inside the noise of ordinary schooling. Delivered to students who are struggling or poor, it is several times larger and worth taking seriously, which is the version of the theory that survived.
The largest single trial agrees. Yeager and colleagues ran a short online intervention with a nationally representative sample of more than twelve thousand American ninth graders in 2019 and reported a gain of 0.10 grade points in core subjects among lower-achieving students, alongside increased enrolment in advanced mathematics. A tenth of a grade point is not a transformation. It is also not nothing, from an intervention lasting under an hour.
Then Macnamara and Burgoyne went through 63 intervention studies covering 97,672 participants and found that the design quality was poor often enough to explain the results on its own. Ninety-four percent of the interventions carried confounds, higher-quality studies were less likely to find a benefit, and authors with a financial interest in the outcome were substantially more likely to report one. Their conclusion was that the apparent effects are likely attributable to inadequate design, reporting flaws and bias.
Why the number does not travel well to one person
A meta-analytic effect size describes the average movement of a large group. It is not a forecast for an individual, and the gap between those two things is where almost all of the frustration with this literature lives.
There is also a measurement problem specific to this construct. The items announce their own preferred answer. Nobody misses which end of intelligence cannot really be changed is the one that sounds better, and self-report handles that badly. A score is therefore partly a reading of the belief and partly a reading of how the respondent wanted to present themselves on the afternoon they answered.
The scale is short for a good reason and pays a price for it. Three items give any single careless answer a large share of the total, so two results a month apart can differ without anything underneath having moved. The same arithmetic that makes a percentile readable makes a three-item score fragile, and it is worth knowing which kind of number is on the screen.
| What the quiz reports | What it does not report |
|---|---|
| A stated belief about whether ability is fixed | How that belief holds up after a genuine failure |
| A belief about one domain, usually intelligence | The same belief about relationships, athletics or character |
| A position relative to other respondents | An amount of anything, since the scale has no natural zero |
| A reading taken today | A stable trait, since these scores move with recent experience |
What it is still good for
The useful reading is diagnostic rather than predictive. A fixed-leaning score is a prompt to look at the sentences a person uses after a setback, because those sentences are the thing the theory was originally about and they are observable without any questionnaire.
Someone who says a presentation went badly because they are not a natural presenter has made a claim about a fixed quantity. Someone who says it went badly because they wrote it the night before has made a claim about a method. The first sentence closes an inquiry and the second opens one. That difference is small, it repeats hundreds of times a year, and it is the only part of the mindset literature that a person can act on directly.
It is also the part that does not require believing the promotional version. Whether or not endorsing developability raises anyone's grades, noticing which of those two sentences comes out first is genuinely informative about how a setback is being processed.
Traits themselves do move, slowly and in a consistent direction across adulthood, which is a separate finding from anything in the mindset literature and better established than most of it. Why a personality result changes between sittings goes through what actually shifts over years. Believing that change is possible is not what causes it, and the research does not claim otherwise once it is read closely.
Sources
- Dweck, C. S., Leggett, E. L. (1988). A social-cognitive approach to motivation and personality. Psychological Review, 95(2).
- Sisk, V. F., Burgoyne, A. P., Sun, J., Butler, J. L., Macnamara, B. N. (2018). To what extent and under which circumstances are growth mind-sets important to academic achievement? Two meta-analyses. Psychological Science, 29(4).
- Yeager, D. S., et al. (2019). A national experiment reveals where a growth mindset improves achievement. Nature, 573.
- Macnamara, B. N., Burgoyne, A. P. (2023). Do growth mindset interventions impact students' academic achievement? A systematic review and meta-analysis with recommendations for best practices. Psychological Bulletin, 149(3-4).
- Roberts, B. W., Walton, K. E., Viechtbauer, W. (2006). Patterns of mean-level change in personality traits across the life course. Psychological Bulletin, 132(1).