In 2006, Carol Dweck, a psychologist at Stanford University, published "Mindset: The New Psychology of Success." The book's central argument was elegantly simple: some people believe that their intelligence and abilities are fixed traits, while others believe they can develop through effort. Fixed-mindset people avoid challenges, interpret setbacks as evidence of inadequacy, and plateau. Growth-mindset people embrace difficulty, learn from failure, and keep improving. More importantly, Dweck argued, the mindset itself could be taught. Telling children that their brain was like a muscle that grew stronger with use, or that the word "yet" could follow any "I can't", could shift students from one category to the other and improve their academic performance.
The idea spread with extraordinary speed through the education world. Dweck's research had shown meaningful improvements in student performance after brief mindset interventions, and the message was both inspiring and actionable. By the early 2010s, growth mindset had become educational policy in thousands of schools. Teachers were trained in it. Posters hung in hallways. "Not yet" replaced failing grades in some districts. The appeal was partly the research and partly the story: intelligence wasn't fixed destiny; effort could change outcomes. This was a message that educators, students, and parents wanted to be true.
The replication problems emerged gradually. Individual studies that showed strong effects often used small samples and were conducted in the labs of researchers committed to the theory. When independent researchers tried to reproduce the results in larger, more diverse populations, the effects shrank.
In 2018, Victoria Sisk and colleagues at Case Western Reserve University published a pair of meta-analyses in Psychological Science. The first examined 273 studies and over 365,000 participants, looking at the correlation between growth mindset and academic achievement. The relationship was real but modest, mindset explained roughly one percent of the variance in academic outcomes. The second examined 29 experimental studies with over 57,000 participants who received mindset interventions. The average effect on academic achievement was around 0.10 on a standardized scale, and even this small effect disappeared among students most at risk of academic failure, precisely the population for whom the intervention was most heavily marketed.
A deeper problem emerged from the intervention studies: in trials that checked whether the manipulation had actually changed students' beliefs about intelligence, the academic effect appeared only when the mindset change had failed. When students genuinely adopted a growth mindset, as measured by follow-up questions, there was no improvement in performance. The theory of change didn't survive contact with its own evidence.
What Dweck's original research had likely captured was real but smaller and more contextual than the popularization suggested: students who persist through difficulty do tend to perform better over time. But the belief that intelligence is malleable is not, on its own, a reliable enough lever to change outcomes across diverse populations. Academic achievement is multiply determined by factors, poverty, family stability, school resources, teacher quality, that a brief mindset lesson cannot displace.
Dweck and her collaborators pushed back on the meta-analyses, arguing that many of the interventions reviewed were low-fidelity implementations of the theory rather than the kind of targeted, carefully delivered programs her original work described. That debate remains unresolved. What the evidence settled was the more sweeping claim, that teaching growth mindset reliably and substantially improves academic performance across populations, which is what had been sold to schools.
