Twitter was angry with me after my Tucker Carlson interview.  The most common accusation was roughly, “Of course Fox News loves The Case Against Education.  Undermining education is a plot to help the Republicans by spreading ignorance.”  

I could take umbrage, but I see where the accusers are coming from.  In the last election, the education gap between Trump supporters and opponents was indeed enormous.  Trump – or at least his public persona – is proudly ignorant.  Wouldn’t it help Trump’s successors if education levels fell?

I doubt it.  While it’s possible that Trump has sparked an historic party realignment, the safe bet says that (a) Trump is an outlier, and (b) we’ll return to the long-run trend.  And over the long-run, there’s surprisingly little connection between education and partisanship.  As I explain in my book:

In the data, the well-educated are only microscopically more liberal.  In the General Social Survey, people place themselves on a seven-step scale, where 1 is “extremely liberal,” 4 is “moderate,” and 7 is “extremely conservative.”  An extra year of education seems to make people .014 steps more liberal.[1]  Taken literally, over seventy years of education are required to shift ideology a single step.  Statistical corrections make education’s impact on ideology look stronger, but it stays weak.[2]

If the effect on ideology is slight, the effect on partisanship is slightly perverse.  As education rises, people grow less Democratic.  The General Social Survey’s respondents place themselves on a seven-step scale, where 0 is “strong Democrat,” 3 is “independent” and 6 is “strong Republican.”  An extra year of education seems to make people .071 steps more Republican.[3]  Statistical corrections makes this effect look weaker, but education still appears to mildly boost support for the party that teachers and professors disfavor.[4]

Furthermore, whatever pattern exists likely reflects relative education rather than absolute
education.  After all, education has risen dramatically over the last
century, but America’s partisan composition has been quite stable.  If
education matters, then, it matters because – regardless of average
national education – people with
similar education levels
politically cluster together.  In technical terms, education largely
works via peer effects – and peer effects are inherently double-edged:

To isolate education’s influence on society, however, you must unpack how education sways students.  Is the mechanism “leadership” – planting teachers’ ideas in students’ heads?  Then education remolds society.  Is the mechanism “peer effects” – sorting kids into distinct groups?  Then education mainly reshuffles society without remolding it.[5]
Suppose you funnel an extra kid into college.  His peer group seismically shifts.  Given human conformity, the freshman will likely try to blend in with his new peers.  College youths are less religious, for example, so one would expect the student to veer in a secular direction.  This does not imply, however, that college makes society less religious.  The existence of college splits kids into two subcultures with opposing peer effects.  If college kids are less religious than average kids, then non-college kids must be more religious than average kids.  Members of each subculture adjust their behavior to locally fit in.  Religious conformity pressure in the non-college pool offsets secular pressure in the college pool.  Net effect on society’s religiosity: unclear, even if college demonstrably makes students less religious.
Voter turnout is a clean example:

Voter turnout rises sharply with education. Substantial effects of education on turnout usually linger after statistically correcting for income, demographics, intelligence, and so on.[6]  Despite some thoughtful naysayers, limited experimental data also show extra education boosts turnout.[7]
The catch: Education has sharply risen over the last century, but turnout has gently fallen.  This could mean offsetting factors masked education’s pro-voting effect.[8]  But several prominent researchers instead conclude that turnout depends on relative education.[9]  People don’t vote because they’re educated, but because they’re more educated than others.  This once again suggests peer effects: The longer you stay in school, the more politically active your social circle, and the more politically active you become to fit in. 

But what about the effect of education on issue views?  That, too, seems largely driven by peer effects.

Bottom line: I’m delighted to receive favorable coverage from Fox (or any major news outlet, for that matter).  But even if The Case Against Education became the blueprint for radical education reform, there’s little reason to think it would reshape the electorate.


1 Results from regressing GSS variable identifier POLVIEWS on a constant and years of education.
2 Chief problem with the simple approach: The well-educated are richer, and the rich are more conservative.  As a result, income conceals some of education’s effect.  If you regress POLVIEWS on a constant, years of education, and log family income, one year of education makes you .028 units more liberal – double the estimate from the simple approach.  Further correcting for race, sex, age, and year, one year of education makes you .024 units more liberal.
3 Results from regressing GSS variable identifier PARTYID on a constant and years of education, excluding respondents who support third parties.
4 If you regress PARTYID on a constant, years of education, log family income, race, sex, age, and year, one year of education makes you .029 units more Republican.
5 To surmount this “zero-sum” problem, peers must have non-linear effects in the right direction.  As Burke and Sass 2013, p.58 remark, “[P]olicy can hope to generate aggregate achievement gains only if peer effects are nonlinear and therefore non-zero-sum in their impact on achievement.”  See also Lavy and Schlosser 2011, p.4.  Hoxby 2002 discusses the complex empirics of non-linear peer effects. 
6 For overviews of the research and some basic results, see e.g. Burden 2009, Nagler 1991, and Powell 1986.
7 The most notable naysayers: Kam and Palmer 2008, p.612 reports higher education has no effect on turnout after fully accounting for “preadult experiences and influences in place during the senior year of high school.”  Tenn 2007 finds immediately after gaining an extra year of education, individuals are no more likely to vote than they were in the previous year.  Sondheimer and Green 2010 examines three sets of experimental evidence on education and turnout.
8 Burden 2009 reviews the contrast between micro- and macro-level evidence, and summarizes the top contending “offsetting factors.” 
9 The leading defenses of the relative education theory are Tenn 2005, and Nie et al. 1996.