One way to read the graph is that there are basically no countries with very low levels of education that have managed to be democratic over the long term, and almost every country with a high level of education has remained a stable democracy.
Thomas Jefferson wrote that “if a nation expects to be ignorant and free, in a state of civilization, it expects what never was and never will be.” In 1960, 36 nations had less than 1.74 years of schooling (which happens to be the level that Afghanistan has today). Of those 36 countries, only two — India and Botswana — managed to have average democracy scores above 4.2.
Out of the 19 countries in this sample with more than 5.3 years of schooling (the current level in Iran) in 1960, 17 have average democracy scores above 7.9. Fifteen of these have been perfectly democratic, at least by the standards of Polity IV. Only Poland and Hungary were dictatorships, and one can certainly argue that those places would have been democracies in 1960s if it were not for Soviet troops.
But in the middle ranges of education, between two and five years on average, almost anything goes.
Some places, like Costa Rica and Italy, have been extremely democratic, while others, like Kuwait and Paraguay, have not. Iraq falls into this category today, which suggests a fair amount of uncertainty about that country’s political future.
Why do I think that the chain of causality runs from education to democracy rather than the reverse? Democracy in 1960 is essentially uncorrelated with subsequent growth in the levels of education. Education in 1960, on the other hand, does an extremely good job of predicting increases in democracy.
The ability of education to predict the durability of democracy is well illustrated by the paths of former Communist bloc countries. Initially well-educated places, like the Czech Republic and Poland, have managed to transition toward being well-governed republics. Poorly educated places have not.
Why is there a connection between human capital and freedom?
Giacomo Ponzetto, Andrei Shleifer and I have argued that the connection reflects the ability of educated people to organize and fight collaboratively.
Dictators provide strong incentives for the ruling clique; democracies provide more modest benefits for everyone else. For democracy to beat dictatorship, the dispersed population needs to have the skills and motivation to work collaboratively to defeat dictatorial coups and executive aggrandizement.
Education teaches skills, like reading and writing, that enable people to work collaboratively. At younger grades, teachers spend a lot of time teaching children how to get along. In the United States, education is strongly linked to civic engagement and membership in social groups. The ability to work together enables the defense of democracy.
Update, I had a busy morning, so I could not add my own two cents until now, but I am not entirely convinced abut the chain of causality. What interests me is whether countries who are experiencing vast expansions in education are finding their democracies stabilizing. Anecdotally for Glaeser's example, Argentina, the answer appears to be no. The differences between Argentina's and Brazil's democracies are quite stark. Argentina's democracy is, to my mind, immature, while Brazil, after an unfortunate period of dictatorship has developed quite a mature and stable democracy. In these data, Argentina is well above Brazil in 1960 in terms of average years of education and remains so today though both have doubled. So perhaps it has more to do with INSTITUTIONS (of which education is a part) is what is really important. Perhaps good education is just a proxy for good institutions (laws, courts, bureaucracies, etc.)