School choice via U-Haul

Earlier this week, Alex Hernandez argued that 43 percent of children in the United States are benefiting from school choice. In The 74, he wrote:

School districts assign children to public schools based on where they live. Then, the world’s largest game of musical chairs begins. The families of 24 million students find ways around their forced assignments to get the schools they want for their kids.

The biggest school choice group is families who move to a new neighborhood to change their assigned school. They represent an estimated 8.2 million students, about one third of all school choosers. To be clear: The most popular way to choose a school in America is to literally move your family.

This led me to wonder: what kinds of families are able to choose a school for their child by moving? Thankfully, Hernandez shows his work in a Medium post, revealing the specific NCES reports he used to generate his estimates. His data comes from the NCES National Household Education Surveys Program, specifically the Parent and Family Involvement in Education (PFI) survey.

Earlier this month, I was at an SDP workshop and heard from Alex Bowers on the kinds of research you can do with publicly-available education data, including NCES surveys.

I took this coincidence as a sign to take the advice of one Alex (Bowers) and extend the work of another Alex (Hernandez). All of the analysis below was performed using this publicly available data and the code is available on GitHub.

Who are the movers?

Moving isn’t cheap. Families with more financial resources are more able to absorb the costs associated with moving to a neighborhood with a more desirable zoned school. Generally, as household income rises, students are more likely to have moved to their neighborhood specifically for the school. The following graph shows how frequently parents responding to the NCES survey decided to move neighborhoods for a better school, broken down by household income.

inc_move_pct

The racial demographics of each income bracket vary. Higher-income movers are less racially diverse than lower-income movers.

race_in_move

Splitting the chart above by racial group, we can see how likely a family is to move given their race and their incomes. We can see that the groups most likely to move neighborhoods for better schools are white and Asian families with a six-figure household income.

race_in_move_pct

The public NCES survey data doesn’t include precise location data, but it does provide the type of neighborhood where these families live. It’s not surprising to see that most movers head for the suburbs. For clarification on what separates a “suburb, large” from a “town, distant” check out their definitions on the NCES website.

zip_move

This chart shows the volume of movers by their chosen destination and their household income. It’s pretty clear that affluent families overwhelmingly choose to move to either suburbs or “rural, fringe” areas, defined as:

Census-defined rural territory that is less than or equal to 5 miles from an urbanized area, as well as rural territory that is less than or equal to 2.5 miles from an urban cluster

inc_move_zip

For lower-income families, suburbs are still a popular destination, but they are much more likely to choose cities or towns.

inc_move_zip_pct

Mover destinations by race reveal a consistent preference for suburbs, with varying degrees of preference for other destination types.

zip_race_move_pct

I’ve only scratched the surface of what we can learn from this particular survey. It has a lot more information I’d like to investigate, including parents’ views on school quality, if their students are in their first-choice school, and more. The survey has also been administered in 2007, 2005, and 2003, so it would be interesting to track some of these patterns on both sides of the Great Recession.

Publicly-available education data is pretty cool.


IBM Is Counting on Its Bet on Watson, and Paying Big Money for It

Nowadays, Watson isn’t just playing Jeopardy:

At the University of North Carolina School of Medicine, Watson was tested on 1,000 cancer diagnoses made by human experts. In 99 percent of them, Watson recommended the same treatment as the oncologists.

In 30 percent of the cases, Watson also found a treatment option the human doctors missed. Some treatments were based on research papers that the doctors had not read — more than 160,000 cancer research papers are published a year. Other treatment options might have surfaced in a new clinical trial the oncologists had not yet seen announced on the web.

But Watson read it all.

It’s pretty cool to see the support Artificial Intelligence (AI) could provide to oncologists. As the research and knowldege base in medicine expands and increasing rates, it’s impossible to expect every researcher to be up-to-date on the every study, technique, or drug released. AI can help doctors make better decisions by recommending treatments based on a significantly larger base of knowledge.

If Watson can help doctors, why couldn’t it also help teachers?

In my opinion, it’s not a question of if, but when teachers will get the same support from AI. Changing classroom practice based on educaiton research has always been a challenge. Organizations like Deans for Impact are doing great work to promote good instructional practice based on cognitive science literature, but it’s really hard to scale.

A “Watson for teachers” could help lower the barriers between high-quality cognitive science, education research, and instructional practice. It may still be years away, but it will certainly help more teachers make better decisions to support student growth.


Book Notes - Why Knowledge Matters

E.D. Hirsch is a self-described “hedgehog,” viewing the world through the lens of one big idea:

“[T]he achievement gap is chiefly a knowledge gap and a language gap. It can be greatly ameliorated by knowledge-based schooling.”

The inverse of Hirsch’s “big idea” is the skills-based approach to education that drives instruction in the vast majority of American schools. The Schools We Need & Why We Don’t Have Them” was my first introduction to the “knowledge vs. skills” debate. This book re-introduces Hirsch’s basic case and extends it based on new evidence from the world of cognitive science and the experience of French schools over the past several decades.

The Big Idea

Reading, critical thinking, and problem solving are not general skills that can be developed and applied to any context. Like most human skills, they are domain-specific, becoming stronger or weaker based on a person’s background knowledge of the topic. Hirsch recommends providing all students with a curriculum in the early grades that focuses on developing knowledge, not skills, to promote equity and achievement among all students.

Key Passages

I’ve heard many well-intentioned people, including educators, state that in the age of Google, it’s less important for students to learn a bunch of information they could just look up. Their belief is that we need to equip students with the critical thinking and problem solving skills to apply these tools to the problems they are tackling.

Hirsch provides a great deal of evidence to support the need focusing on building knowledge instead of generalizable skills, but the most powerful rebuttal to this line of thinking is his explainiation of the “Matthew Effect” in education:

Reference works, inlcluding those available on the Internet, are immensely valuable to already knowledgeable people. Google is not an equal-opportunity fact-finder; it rewards those already in the know. Instead of being an agent of equality, Google rewards cognitive insiders. It consolidates educational inequity. Just as it takes money to make money, it takes knowledge to gain knowldege.

Throughout the book, Hirsch shows how the shift from a universal knowledge-based curriculum in France to a locally-determined and skills-focused approach in 1990 led to disastrous results:

The new localizing and indivudalizing of the schools had been proposed as a way of overcoming the reproduction of social stratification. But the educational results went the opposite way, intensifying inequalities. As the French data from 1987 to 2007 show, the new 1990 arrangements greatly intensified the social reproduction they were supposed to diminish. They systematically deprived poor children of the enabling knowledge that rich children had acquired from their home environments.

Hirsch’s proposed solution, replacing a indivudal-focused reading curriculum based on “leveled readers” with a knowledge-centered communal curriculum, is paradoxically a better way to encourage students to develop as individuals:

It must not be assumed that a different approach - whole-class instruction of real books and subject matters - means grey uniformity. On the contrary, real knowledge breeds real interest and individuality.

This principle is extended later in the book:

It’s in relation to the community that the individual develops. Only by mastering the shared knowledge and norms of the community can one diverge from those norms effectively.

The Core Knowledge curriculum, developed by Hirsch, is one approach to empowering students with a foundation of communal knowledge. He acknowledges that there can be some differentiation in early grade curricula based on local values while still building a strong foundation of knowledge:

Rival elementary curriculums may reflect rival values but all responsible ones will exhibit a great deal of overlap - out of a sense of obligation to the children who will need knowledge represented by those words in order to function well in American society.

Actionable Items

Hirsch makes a convincing case for the power of a content-rich, coherent, and cumulative curriculum to build a strong foundation of literacy. While research has shown the strong impact of a high-quality curriculum on student achievement, there’s a paucity of publicly available data on school curriculum. In some cases, districts may not even know which textbooks are being used in their classrooms.

One can agree or disagree with the case Hirsch makes, but we should all be able to agree that curricular choices play an important role in student achievement. It would make sense to make information about those choices more transparent so that:

  1. Educators would know which schools/districts are using similar/identical approaches, creating opportunities for highly relevant cross-school/district collaboration.
  2. Researchers could evaluate the relative strength/weakness of curriculum implemented in public schools.
  3. School boards and local school councils would have significantly more information to inform adoption/renewal of curriculum.

None of this is possible without transparent data on curriculum. Collecting information about such an important component of public education shouldn’t require a FOIA request.


Blame the Elites for the Trump Phenomenon

Intelligence Squared hosts Oxford-style debates, with two teams supporting or opposing a motion. The motion of their most recent debate: blame the elites for the Trump phenomenon.

Personally, I’m for the motion, but I would encourage you to watch and decide for yourself.


The World Turned Upside Down

After 11 years of litigation, Judge Thomas Moukawsher issued a seismic decision in the CCJEF school finance lawsuit. In the decision, Moukawsher not only eviscerated Connecticut’s school finance system, he also detailed flaws in the state’s approach to teacher evaluation, teacher compensation, the lack of clear definitions of success in elementary and secondary education, and gave the state 180 days to come up with a solution to all of it.

There’s a lot to digest, but here are my initial thoughts:

  • It’s good to see the ruling bring more attention to the distribution of funding than the overall allocation. As I argued nearly two years ago, actual appropriations are not driven by the ECS formula, leading to huge funding disparities (positive and negative) that affect thousands of students.
  • I’m sympathetic with the desire for better approaches to teacher evaluation and compensation, I’m not sure the order in this ruling is the best way to actually improve how those systems operate. To paraphrase Rick Hess, while judicial orders can make people do things, it cannot make people do those things well.
  • The timeline is super-aggressive. Staffers at the Connecticut State Department of Education are going to have a busy winter.
  • Will the state’s proposed solution include magnets, charters, and/or other choice options? The state already spends a significant amount of funding on these schools and the ruling sounds like a mandate for a comprehensive solution. It will be interesting to see how they address this question.

There’s much more to digest in this decision. While the plaintiffs got a partial victory, they’re about to learn an important lesson:

Winning is easy. Governing’s harder.