Links and Commentary 7/18/25

Did I Leave Out Growth Mindset?, Viewpoint Diversity Woes, Latino Realignment, The End of the DOE?, Will AI Slow Scientific Progress?

Author

James Shanahan
July 18, 2025

Welcome to Scholastic Alchemy! I’m James and I write mostly about education. I find it fascinating and at the same time maddening. Scholastic Alchemy is my attempt to make sense of and explain the perpetual oddities around education, as well as to share my thoughts on related topics. On Wednesdays I post a long-ish dive into a topic of my choosing. On Fridays I post some links I’ve encountered that week and some commentary about what I’m sharing. Scholastic Alchemy will remain free for the foreseeable future but if you like my work and want to support me, please consider a paid subscription. If you have objections to Substack as a platform, I maintain a parallel version using BeeHiiv and you can subscribe there.

Did I Leave Growth Mindset Out of my Skeptical Series?

On Wednesday I posted a reflection on the five posts in the I’m Skeptical series that I’d been writing. The purpose of the posts was to think about potential future failed reforms, interventions, or “big ideas” that resembled a previous failure, learning styles. As I was reading around this week, I came across some discussion of Growth Mindset, the theory that students who believe they can grow and change do better in school than students who believe their capabilities are unchangeable. This one neat trick really is supposed to fix all kinds of problems in education, especially math education, where many of the studies showed kids with a growth mindset would take on more challenging problems and be successful. This belief was further popularized in the 2010s by people in tech circles who argued that failure was not just an important part of startup culture but was essential to learning.

This whole endeavor seems primed to trip my skepticism alarms. There’s a persuasive story being told and some branding that associates growth mindset with economically prosperous and innovative industries. That’s my 5th educational skeptic principle in action! But growth mindset did not appear to violate my first four principles. It was well defined and research showed a plausible causal relationship between growth mindset and desirable education outcomes. The measures used in the studies likewise seemed well thought-out if sometimes a bit… proprietary. Most of all, mindset researchers published studies that took place in actual schools, often in dozens of them with tens of thousands of students. They had a claim to scale and it all seemed like the real deal. So, as skeptical as I was, I don’t think I could say in good faith that I expected growth mindset to be an expensive waste of time for schools that would rival learning styles. Imagine my surprise this week when I came across a 2023 meta-analysis of mindset research that really threw a bomb at the mindset crowd.

When examining all studies (63 studies, N = 97,672), we found major shortcomings in study design, analysis, and reporting, and suggestions of researcher and publication bias: Authors with a financial incentive to report positive findings published significantly larger effects than authors without this incentive. Across all studies, we observed a small overall effect: d¯ = 0.05, 95% CI = [0.02, 0.09], which was nonsignificant after correcting for potential publication bias. No theoretically meaningful moderators were significant. When examining only studies demonstrating the intervention influenced students’ mindsets as intended (13 studies, N = 18,355), the effect was nonsignificant: d¯ = 0.04, 95% CI = [−0.01, 0.10]. When examining the highest-quality evidence (6 studies, N = 13,571), the effect was nonsignificant: d¯ = 0.02, 95% CI = [−0.06, 0.10]. We conclude that apparent effects of growth mindset interventions on academic achievement are likely attributable to inadequate study design, reporting flaws, and bias.

Yikes! Researchers don’t take that harsh of a tone unless they’re confident about the criticism. It seems like my skepticism may have been warranted but I was overly credulous of the research being published. It’s notable that the authors point to publication bias as one of the key drivers of growth mindset’s positive outcomes. So, yeah, maybe in a few years we’ll have people everywhere asking why we spent so much of teachers’ time training them to help kids develop a growth mindset.

Higher Ed Viewpoint Diversity Woes

Two contrasting statements about the state of higher education popped up on the Stacks today and I don’t quite know what to think of them. In the first, Dan Drezner, pushes back against provocateur Chris Rufo’s Manhattan Statement.

This leads me, and me alone, to issue a Medford Statement on Higher Education:

Universities are imperfect institutions, and should always strive to be better at fostering an open intellectual culture and a more dynamic research and teaching environment. However, universities do not need to kowtow to advice tendered from those who:

  • Proffer piss-poor, monocausal explanations for what ails higher education;

  • Talk about universities receiving government funds as a gift without recognizing the spectacular rate of return that state investments into higher education yield;

  • Describe universities as dominated by the humanities when even a cursory knowledge of the higher ed sector exposes the absurdity of such a claim;

  • Fail to criticize political correctness emanating from campus conservatives with the same fervor as that emanating from campus leftists;

  • Refuse to acknowledge that conservative arguments about structural bias in academia mirror progressive arguments about other parts of society;

  • Advance state control over private universities for partisan purposes.

In closing, the Manhattan Statement on Higher Education is an unhelpful collection of lies and half-truths designed to subvert and not build up U.S. universities. Anyone telling you differently is selling you something.

But then the very next thing in my inbox is Matt Yglesias saying that, actually, experts in academia are too cloistered and too politically radical. He says climate researchers are too progressive, as are scientists in general, and it undermines both public trust in their expertise and decreases their ability to make meaningful contributions to science. This left-ness of academic scientists, according to Yglesias, (and, again, he ain't talking humanities here) is even the cause of the Great Awokening because they all went on Twitter and spouted off about anti-racism. He closes with a call for viewpoint diversity and throughout the post he seems to be adopting many of the things Rufo is saying.

I think political leaders should start talking in a calm, non-judgmental way about the fact that strong ideological sorting across professions is not a positive trend for society. Having robust disagreements about values among professionals with the same areas of factual expertise can clarify our options and the tradeoffs they entail. This sort of diversity also makes it easier for expert communities to communicate with the broad public in a way that’s credible and compelling, rather than relying on arguments from authority that increasingly fail in our endlessly fragmented ecosystem.

If you know me of have followed my writing, you know that I am inclined to agree with Drezner here but it also makes me wonder what’s going on with Yglesias. Why is he so ready to take on the same positions as the bad-faith activist Rufo? I’ll have to think about this and see if I can’t sus out whether there is a valid critique here.

But, contra-Yglesias, it’s not initially clear to me that having more conservatives in, say, educational psychology departments, would necessarily yield better results. Look at the links above! Growth mindset theories come primarily from the work of Carol Dweck who is a professor at Stanford. You know, Stanford, the home of the conservative Hoover Institution. The Hoover Institution does extensive work on education. How did the viewpoint diversity at Stanford help exactly?

Hispanic Political Realignment and China Shock

This link strays farther from education than usual but I think it’s interesting nonetheless. As you may have heard, Hispanic and Latino voters have been swinging to the right and voted for Trump in greater numbers than they had previously. Here’s the bit where schools come in:

Rio Grande City is run not so much by parties, Barrera explained, but by rival factions with a strong resemblance to the machines of old. School-board elections are officially nonpartisan, but the voting is organized around competing candidate slates. The slates are like parallel political parties, but able to endorse across party lines for partisan races. These factional operations are far more sophisticated than the formal party structures. Candidates for the statehouse in Austin will simply pay these slates to serve as their get-out-the-vote operation, forgoing traditional campaign activity.

Why do the slates matter so much? In many of the poorest counties in the nation, with little private industry, the No. 1 employer is the local school district. And whoever wins the school-board elections decides who gets the relatively well-paying patronage jobs that come with those seats. That means the school-board races are uniquely high-stakes; incumbents will go to extreme lengths to safeguard their power.

I’ve experienced something similar. School jobs are steady and pay well in economically depressed areas where opportunities for gainful employment my not exist or may only be in harsh conditions such as coal mining or agricultural work. This means there are patronage machines attached to whoever runs schools that can benefit friends and families and raises the direct financial stakes of school board elections. As a result, anti-patronage populism and support for Trump grows among those out of power who do not benefit from the old system.

As these old partisan ties begin to weaken, it’s worth remembering that something similar has happened before, when the white working class’s status as the bulwark of the old Democratic Party began to unravel in the 1960s. That was also a time of rapid social change, when a politics once focused on meeting the material needs of the working class instead started to revolve around questions more abstract: of war and peace, of race and sex. And on key points, the working class—meaning the white working class early on and a more diverse group today—was not on board with the Democrats’ growing cultural liberalism.

The realignment of the working class, which helped Trump win in 2016, would not stop with white voters. In 2020 and 2024, the realignment came for nonwhite voters. A basic tenet of the Democratic Party—that of being a group-interest-based coalition—was abandoned as the party’s ideologically moderate and conservative nonwhite adherents began to peel off in a mass re-sorting of the electorate. The Democratic analyst David Shor estimates that Democrats went from winning 81 percent of Hispanic moderates in 2016 to just 58 percent in 2024. And these voters were now voting exactly how you would expect them to, given their ideologies: conservatives for the party on the right, moderates split closer to either party.

But let’s make an important connection here: economics and international trade, especially with China. It turns out, opening trade with China and the incipient “China Shock” was felt unequally by various communities in the US. Most of the focus has been on geographic differences in where the shock was felt, e.g. rustbelt cities. Some new research suggests, however, that the effects fell more harshly on Hispanics/Latinos than on other minority groups.

Minority workers tend to be disproportionately harmed by negative economic shocks. Indeed, we show that Hispanic populations experienced worse employment losses due to import competition from China, relative to whites, largely due to lower education levels. In contrast, Black-white employment and wage gaps actually narrowed due to relative growth in non-manufacturing sectors. We show that Black workers were less attached to manufacturing by 2000, compared to whites, and were therefore more poised to take advantage of China shock induced reallocation to services. The lasting negative impacts of the China shock on exposed communities were primarily driven by white workers.

The story this could tell is one in which Latino communities in the US were harmed most by opening trade with China and economic precarity began to creep in. As good work became harder and harder to find, the inequitable systems that doled out what good jobs did exist were exposed and caused the realignment. Seems plausible.

Is This the End of the DOE?

I’ve said before that I have my doubts about whether or not the Trump administration will actually close the Department of Education.

Let’s say you’re a devoted evangelical christian and you believe that all public schools should have prayer in school, display the ten commandments, and follow an abstinence only sexual education curriculum. Maybe you feel that Diversity, Equity, and Inclusion programs are contrary to your christian beliefs and promote a Marxist-atheist worldview. Your state of, let’s say, Louisiana already does most of this but you’re not happy because a liberal state like Illinois or New York will not adopt those rules. Vouchers aren’t enough! Not every kid will attend a properly christianized school. Millions of children are being educated under a regime that you think is going to morally ruin those kids. You would prefer that the federal government enact some kind of policy to get those states to go along with your preferred vision of what education should look like. This might mean that you actually want the Department of Education to exist and to promote those policies.

But this last, the Supreme Court gave the go-ahead for the Trump Administration to begin firing federal employees and this week they specifically said the president could lay off DOE workers, too. Following the layoffs, the administration is moving some key functions over to other departments. For example, the Department of Labor will inherit $2.6 billion in funding for training programs for “adult education, family literacy programs and career and technical education.” Meanwhile, the $1 trillion in student loans has yet to be reassigned anywhere. Overseeing protections for students with special needs may go to the Health and Human Services department under RFK Jr., who has some pretty simplistic views of people with disabilities. The Justice Department may get to manage overseeing civil rights protections.

Still, I am skeptical that this means the DOE will be closed for good. For one, the OBBBA that passed earlier this month directs the DOE to set up and manage a federal school voucher system. I also continue to believe that the DOE is needed by conservative and religious activists in the Trump administration to enforce ideological conformity in blue states. For example, the DOE is investigating schools’ DEI practices. While this could be taken over by the Justice Department, there don’t appear to be plans in place to do so. What I expect is that every legitimate function of the DOE will cease or be spun off to another department where it will also die off. In its place we will have the prerogative state, which exists to bend and break the law in service of the president’s ideological agenda.

Will AI Slow Scientific Progress?

Since we’re actively defunding research taking place within higher education while simultaneously pressuring them to hire more conservatives who disagree with scientists about climate change, many have turned their hopes to AI. Maybe sufficiently advanced LLMs will generate scientific breakthroughs and make up for the loss of so much scientific capacity? Sayash Kapoor and Arvind Narayanan say not so fast.

Over the last decade, scientists have been in a headlong rush to adopt AI. The speed has come at the expense of any ability to adapt slow-moving scientific institutional norms to maintain quality control and identify and preserve what is essentially human about science. As a result, the trend is likely to worsen the production-progress paradox, accelerating paper publishing but only digging us deeper into the hole with regard to true scientific progress.

The number of papers that use AI quadrupled across 20 fields between 2012 and 2022 — even before the adoption of large language models. Figure by Duede et al.

The point they make here is an important one! Quality control is mentioned in the sense that papers have to be doing good science, following proper methodologies, and going through rigorous error checking. They point out that the way cutting edge AI currently works means that relying on AI might mean “prolonged reliance on flawed theories.”

In a recent commentary in the journal Nature, we illustrated this with an analogy to the geocentric model of the Universe in astronomy. The geocentric model of the Universe—the model of the Universe with the Earth at the center—was very accurate at predicting the motion of planets. Workarounds like "epicycles" made these predictions accurate. (Epicycles were the small circles added to the planet's trajectory around the Earth).

AI-based modeling is no doubt helpful in improving predictive accuracy. But it doesn't lend itself to an improved understanding of these phenomena. AI might be fantastic at producing the equivalents of epicycles across fields, leading to the prediction-explanation fallacy.

In other words, if AI allows us to make better predictions from incorrect theories, it might slow down scientific progress if this results in researchers using flawed theories for longer. In the extreme case, fields would be stuck in an intellectual rut even as they excel at improving predictive accuracy within existing paradigms.

Yikes!

Thanks for reading!