I have had a couple of people tweeting this study today and I can understand why:

• This study is a preregistered test which means that the researchers give their plans free beforehand so they can’t alter anything afterwards.
• The study has a huge sample (12,542),
• There was a field setting (65 U.S. public schools),
• They also used independent research consultancy for many aspects of the research process,
• …

To sum it up:

But what are the results?

The good:

While this may seem small, do note, this is actually not bad for a very low-cost intervention. But it’s more nuanced as the researchers state:

…the growth mindset intervention effect was heterogeneous in predictable ways. Some sub-groups of students (lower-achievers) and schools (those with supportive behavioral norms) showed appreciably larger increases in grades.

But what do we learn from this, another favorite tweep sums it up as follows:

And also this question remains:

You can check the preprint here from which this is the abstract:

A pressing global challenge is to identify interventions that improve adolescents’ developmental trajectories. But no intervention will work for all young people everywhere. It is critical then to study the heterogeneity of intervention effects in a way that is generalizable and replicable. In the National Study of Learning Mindsets (N = 12,542) researchers randomly assigned 9th grade students in a representative sample of 65 U.S. public schools to a growth mindset intervention, which conveyed that intellectual abilities are not fixed but can be developed. The brief (~50-minute), scalable and low-cost intervention reduced by 3 percentage points the rate at which adolescents in the U.S. were off-track for graduation at the end of the year, corresponding to an estimated benefit of approximately 100,000 adolescents per year. This is the first experimental evidence that an intervention can improve adolescents’ educational trajectories in a national probability sample. Yet the growth mindset intervention effect was heterogeneous in predictable ways. Some sub-groups of students (lower-achievers) and schools (those with supportive behavioral norms) showed appreciably larger increases in grades. Heterogeneity findings were reproduced in a conservative Bayesian “sum-of-regression-trees” analysis, which guards against false discoveries. These findings lead to novel hypotheses about ways to enhance intervention effects and target public policies. Findings also illustrate the power of even slight adjustments in motivational priorities to create enduring change during adolescence.