Last night I gave a talk about growth mindset research at the Global Math Department. They just posted the recording, here, and I previously shared links to some of the research we discussed last night here.
My main point from the talk was that the world of teaching for a growth mindset is talking about something quite different than what mindset researchers talk about. Dweck’s Mindset and Boaler’s Mathematical Mindsets are not talking about same thing.
For instance, YouCubed and Jo Boaler’s Mathematical Mindsets lays out a comprehensive vision of what “growth mindset” classrooms look like. This includes things like ending timed work, using more visual representations, ending tracking, plus many more things besides.
This has very little direct impact on mindset, though. This becomes especially clear after a careful look at what psych researchers themselves are doing with mindset research.
The original work comes from Dweck in a series of papers in the 70s and ‘80s. She noticed that, when presented with a challenge, some students would behave helplessly (e.g. giving up, “I’m bored,” getting upset, etc.) while others seem motivated by the challenge. What could explain these different patterns of behavior?
Dweck retreated to goals to explain behavior. Different people react to challenges differently because they have different goals. Some people have learning goals, while others have performance goals — basically a desire to look good.
OK, but then why do people have different goals?
That’s when Dweck lands on implicit theories of intelligence, i.e. mindset. People who have fixed mindsets believe that intelligence is essentially something you’re stuck with, hence they don’t have learning goals and are only left with looking good. (Presumably everyone likes looking good, it’s just that if you believe that intelligence is malleable you believe in learning.)
(As an aside, what does it mean to believe that intelligence is malleable other than believing in learning? In other words, is Dweck’s theory tautological?)
(Another question: why stop at mindset? Why doesn’t Dweck explain where mindsets come from? I guess, from the interventions they’ve designed, we can get a sense for one place that mindsets can come from.)
This leads mindset researchers to essentially make two predictions:
- Mindset matters. Having a growth mindset leads to good things (e.g. academic achievement, good relationships, professional success, etc.)
- We can do something about it — we can turn some of those bad outcomes into good outcomes with our interventions.
Those interventions can be very brief, as short as 45 minutes, and they can be done using completely standardized online materials. Pretty much every intervention has the same basic structure: science tells us that the brain can grow, that challenge is good, and hard work can lead to good things happening. Then there are some reflective prompts for discussion or writing.
(Some really good intervention materials from mindset researchers devoted to making this stuff free can be found on PERTS.)
One last question to round off the researcher version of growth mindset: how could such a brief intervention have any significant impact at all? It seems like magic. Not so, say Yeager and Walton in “Social-Psychological Interventions in Education: They’re Not Magic.” The key is recursive effects, positive feedback loops where the intervention marginally increases mindset, which then marginally promotes good learning, which then further effects mindset, which then &c., &c.
So the claim is that growth mindset interventions can have major effects. But just how major? That’s the question of the first of the recent papers we talked about, the 2018 meta-analysis of existing research.
How important is growth mindset for academic achievements? When the researchers pooled a comprehensive selection of results together, they found a very modest average effect. (Check out the paper and you’ll see some discussion of a fairly high gap between some studies that find fairly strong effects and others that don’t, more heterogeneity than you might expect. They aren’t able to easily find an explanation.)
And what about mindset interventions? The main point is that they find extraordinarily weak average effects for mindset interventions, as shown in this forest plot below. (The diamond represents the mean effect. Note just how many interventions are touching or less than 0 in their effect sizes.)
And these results seem roughly in line with what Dweck and her recent collaborators at PERTS are finding in their top-notch mindset experiments. They’ve been trying to address criticisms of mindset research while doing work that would prepare mindset research to scale and reach many students.
This move of the focus of mindset experiments to scale has quietly involved a number of other shifts as well.
- Away from testing in just one context
- Less researcher involvement in the mindset treatment
- Standardized materials
- Big studies
Correspondingly, there have been two changes in the sorts of results that the PERTS team have been publishing with Dweck.
- Moving away from claims that mindset interventions have an impact on everybody, now a focus on at-risk students.
- Much more modest effects being claimed.
You can see this in the figures from the 2015 pilot study. There was really not very much of an effect at all from the growth-mindset intervention for the general population. This is in line with the meta-analytic results.
There is an effect — again, a modest one — from the growth mindset intervention for at risk students on their GPA.
But these results came from what was supposed to be a trial run before a much larger experiment. And those results were briefly released earlier this year in draft version. The official version hasn’t been published, but the study was widely praised for its rigor and again found modest results, mostly for at-risk students.
Here’s where I want to start putting the pieces together. What I want to ask is what these results should entail for advocates of teaching for a growth mindset. My answer is: not much, because the relationship of all this research to what goes on in teaching was never clear.
Here’s what I think happened. Dweck’s research begins with descriptions of helpless behavior from students, and she points to motivational problems. All of this resonates deeply with teachers, so we associate motivational concerns with growth mindset. When Boaler talks about teaching for a growth mindset, usually she means motivational teaching, e.g. if you teach with visual representations more students will experience success and they’ll be more motivated to continue engaging.
But mindset researchers were only ever trying to impact the very bottom of this causal chain, the mindset stuff. And maybe, it turns out, that wasn’t as important as they thought it was, and all the stuff in the middle (i.e. teaching) is really important.
When we dialogue with research, there are always a couple different games we can play. One game we can play is “how can this be wrong?” There’s a time for that. Another is “how can this be right?” and there’s a place for that too, but my favorite is “how can both of these things be right?”
I think it’s entirely possible that some mindset interventions had huge successes, but I also think that when we try to get clear enough about why these worked, the generalizations don’t hold particularly well. And that’s because we’re working with motivation, fickle motivation, and it depends on so many different things — what the kids happen to need, who’s running the intervention, how the kids happen to hear the message, which of their other adults were involved, etc. It seems to be very hard to standardize motivation.
But teachers who advocate for growth mindset can be right — mostly right — also, because they were never talking about just changing a student’s theory of intelligence. They were talking about the entire causal chain of motivation, including social safety, helping kids see success, making the curriculum more accessible, &c., a million things that have to do with motivation but that are waaaay beyond the purview of mindset research.
I don’t mean to be too wishy washy. Mindset is not a revolution — that’s not a hypothesis that finds a ton of support in 2018. And teaching for a growth mindset doesn’t have all that much to do with mindset, and I’d prefer that we talk about motivation rather than mindset — I think that’s what we really mean to say.
I’ll end with two questions:
What other research would we be curious to see in 2019 or any other year, relating to mindset?
What are the narrower, more specific, less potentially revoluationary but more potentially connected to the classroom, questions that we can ask about motivation? Because revolutions like Dweck’s are consistently overrated at the outset, and in education we tend to place teaching on these boats that end up being much more rickety than they at first seem.