One at a time, fifth graders were walked from their classroom to an empty one, sat down with a researcher, and handed ten picture puzzles. After four minutes the researcher scored them: “Wow, you did very well on these problems.” A number. “That’s a really high score.”
Every child in the study heard that, whatever they had actually solved. Every child was told they had solved at least 80 percent. Only the next sentence changed. Some heard, “You must be smart at these problems.” Others heard, “You must have worked hard at these problems.” A third group heard nothing more.
Then the children were asked what they would like to work on later. Some options promised a chance to shine, such as “problems that are pretty easy, so I’ll do well.” One promised something else: “problems that I’ll learn a lot from, even if I won’t look so smart.”
In the first study, 67 percent of the children praised for being smart picked a chance to shine. Of the children praised for effort, 92 percent chose to learn.
What the praise taught
Claudia Mueller and Carol Dweck, psychologists at Columbia, published the studies in 1998. Nothing seems kinder than telling a child they’re smart. In a poll they cited, 85 percent of parents thought praising a child’s intelligence after a success was necessary to make the child feel smart.
The kindness had a cost once the puzzles got hard. After a much tougher set, and the news that they had done “a lot worse,” the children praised for intelligence were less willing to keep going. In one study, when they wrote to children in another state about the puzzles, 38 percent of them reported a higher score than they had earned. Among children praised for effort, 13 percent did.
The children’s own answers offered a clue. Those praised for intelligence were more likely to describe it as a fixed trait. Those praised for hard work more often said it could improve.
Believe ability can grow, and a hard problem looks like practice instead of a verdict. In the theory, believing it is fixed turns a hard problem into a test of who you are, and makes the easy problems the safer bet.
Dweck’s research called these beliefs “entity” and “incremental” theories of intelligence. Her 2006 book Mindset took them to a much wider audience as the fixed and the growth mindset.
Smaller than the slogan
The idea spread into classrooms around the world, and then researchers measured how much it actually moved.
In 2018 a team led by Victoria Sisk ran two meta-analyses. One, covering more than 365,000 participants, looked at the link between mindset and academic achievement. The other, covering more than 57,000, looked at mindset interventions. Overall effects were weak in both. Some results did support a narrower claim: that students with low socioeconomic status, or who were academically at risk, might benefit from the interventions.
The largest test came in 2019. Researchers led by David Yeager randomly assigned 12,490 ninth graders, in a nationally representative sample of 65 US public high schools, to an online growth mindset lesson of less than an hour or to a control condition. Lower-achieving students who got the mindset lesson ended the year with core-class grades 0.10 points higher, on a 4.3 scale, than similar students in the control group. Among higher-achieving students there was no significant effect. The gains were larger in schools where peer norms supported taking on challenges. A separate team, testing 624 children aged ten to twelve, found that praise for intelligence had no significant effect on their cognitive performance.
Dweck herself has warned against the shallow version. “Perhaps the most common misconception,” she wrote in 2015, “is simply equating the growth mindset with effort.” Praising effort when a student isn’t learning, she wrote, can hide a learning gap instead of closing it.
The report she asked to keep
Imagine you manage a new analyst who has turned in three excellent sales reports. Each time, you have told her she’s a natural. Now there’s a pricing model nobody has cracked, and you offer it to her. She asks if she can keep doing the reports.
The 1998 study can’t tell you that your praise caused this. What it shows is how the same choice went for fifth graders: praised as smart, most of them picked the problems that would keep them looking smart.
So the next time she finishes something, the comment could be about the work: the query she rewrote, the outlier she chased down. When she gets stuck on the model, the line Dweck suggested for teachers carries over to a desk: “Let’s talk about what you’ve tried, and what you can try next.”
Her first draft of the pricing model comes back with nine cells flagged in red. This time the question on the table is which one to fix first.