In the September 2017 German Federal election, the anti-immigrant populist party Alternative for Germany (AfD) took nearly 13 percent of the vote, earning it 94 legislative seats, making it the 3^rd largest party in Germany’s parliament. Plenty of ink has been spilled to explain AfD’s surprising strength. The most notable and comprehensive example comes from the European think-tank Bruegel. They regress AfD’s vote against an extensive battery of regional statistics: education, population age, immigration-related characteristics, religion, urbanism, and of course communist history. That last point has come to dominate discussions of AfD’s success: it is impossible not to notice that AfD performed incredibly well in the states that once composed communist East Germany.

But there is one powerful factor driving AfD’s strength that will be familiar to American poll-watchers but has been mostly ignored in commentary on Germany: sex. The German government collects demographic data on a stratified random sample of voters, and releases tabulations of that data for public use, published in a way to guarantee anonymity of individual voters. This data enables us to say with some degree of precision exactly who voted for which party. With specific data for each State-Age Group-Sex combination, we don’t need to guess who propelled AfD to victory: we can see it directly.

The first point to note is that many previous analyses of AfD’s performance may have tripped up on the impact of age because they assume age should have a linear effect. But in reality, we can see that AfD performed best among middle-aged people, and worse among the young and old. This pattern repeats itself across virtually all the German Federal states.

But even more striking than age is the sex breakout. AfD performed far worse among women than among men. There is no state and age group pair anywhere in Germany that saw women give a larger share of their votes to AfD than men, and in fact the gaps are reasonably consistent across regions.

AfD didn’t just win men: wherever men made up a relatively large share of a given age cohort in a given state, even women in that state and age cohort were more likely to vote for AfD. An increase in the male-female ratio for a given cohort of 1% is associated with about a 0.5-1.5 percentage point increase in even women’s votes for AfD.

Why do men favor AfD so much more than women? Well, why do American men favor the Republican Party more than women? In the U.S., we often tell it as a story about religious social values. Yet AfD is strongest in the most secular places, and multi-variate regressions show that increased religiosity is actually associated with significantly less support for AfD. So whatever makes German men hostile to immigration (AfD’s defining issue), it isn’t traditional religious values. Rather, it may be that changes in the global economy have systematically disfavored historically male-dominated industries, or that men are more likely to take a protective or defensive view of nationhood.

Or, it may just be that men in some societies are pulled towards more radical politics of many varieties, and just happen to be ticked off at their former political home. This leads to the next vital observation which has been too-often overlooked for Germany: the importance of history. Polling on voter history suggests that at least 420,000 voters switched from the Die Linke party, the successor to the old East German communist party, from 2013 to 2017. While this is fewer than switched from the conservatives—about 1 million—the raw numbers are misleading. The graph below shows what percentage of each party’s 2013 voters switched to AfD in 2017.

As you can see, more than 1 in 10 Die Linke voters switched to AfD. That’s only matched among “other parties,” which includes a huge swing from the ultranationalist National Democratic Party of Germany, which lost over 90 percent of its voters in 2017, presumably with many going to AfD. AfD also appears to have made major gains from various protest and single-issue fringe parties.

But these figures are almost certainly an underestimate, because the same survey says 1.4 million AfD voters also voted for AfD in 2013. The only problem? AfD only received 810,000 votes in the whole country in 2013 in the constituency vote, and about 2 million in the party vote. It’s plausible that 1.4 million of the 2 million party-line voters showed up again, but AfD also massively overperformed in the constituency-level voting. At the constituency level, AfD probably made even bigger gains from other parties, especially Die Linke.

Tellingly, we can also look at the age and sex makeup of Die Linke’s voters in 2013, compared to 2017. These polls show that Die Linke experienced significant losses among middle-aged men, especially in East Germany, the exact category that AfD won with flying colors. Across every metric, the story is clear: the rate at which AfD peeled off voters from the center-right, neoliberal, and center-left parties was quite similar, but the formerly-communist Die Linke, extreme-rightist minor parties, and protest parties hemmhoraged voters to AfD. The story here isn’t of conservatives getting more conservative, but of the extremes of the political spectrum bending back on themselves, and a cluster of East Germany men with weak commitments to stable democratic institutions voting for whichever party most symbolizes protest, anger, and opposition. In the past, that meant Die Linke with its communist ties, the NPD for its ultranationalism, or even the Pirate Party. Today, it means AfD, a loosely-organized political movement based around being a thorn in the side to Germany’s political establishment.

But if weak commitment to institutions is a driving factor, then we may want to explore German history from before 1989, or even 1945. Political institutions can be quite durable. A well-known map of Poland’s electoral outcomes shows that the formerly Prussian-controlled part of Poland reliably votes for somewhat less populist parties in Polish elections. Other research has shown that the spread of the Reformation through Germany left a durable political imprint to the present day. Across the border in Austria, right-wing populists have also seen recent victories. The interesting thing about Austria, however, is that, like Germany, it experienced communist occupation in part of the country. For ten years after WWII, the USSR occupied a large portion of northeastern Austria. Economic research has shown that this occupation created economic and demographic dislocations with real effects persisting to the present day.

So might there be a deeper origin to which parts of Germany have a weaker commitment to stable democratic institutions? Virtually every analysis of the 2017 election notes the blindingly obvious fact that East Germany was a hotbed of AfD support. This is written down to the legacy of communism, which seems plausible. Conservative populism has proven appealing in virtually every post-Soviet country from East Germany to the Russian Far East.

One thing we could do is to look at German governance before Germany was a unified state and see if it predicts modern political choices. The borders of German states pre-unification were different from modern borders in many cases, so can help explain district-level variation within and across German states. I have coded every election district in Germany based on its pre-unification governance, and tested their statistical associations across a wide range of models. Across almost every test I could define, one factor kept popping up: the longer a given electoral district had been governed by Prussia, the less it favored AfD given almost any set of controls.

Different places saw different amounts of time controlled by Prussia. But broadly speaking, areas with Prussian dominance are associated with between 0.5 and 5 percentage points lower AfD vote after controlling for age, sex, sex ratio, and other key factors, with that range mostly determined by variation in the duration of Prussian rule. This helps explain AfD’s great strength in Saxony and Thuringia, yet relative weakness in Brandenburg, all three of which were dominated by communism. It also helps explain why AfD made greater gains in southern Germany, never directly dominated by Prussia, than North-Rhine Westphalia or the Rhineland, where Prussia was the dominant power for over a century. Just as a more free market rightist party did better in Prussian-dominated parts of Poland, so the German neoliberal FDP party made major gains in formerly-Prussian parts of Germany as well.

Determining causality is challenging and must be left to more formal academic study. But it should be noted again that this is a very similar effect as observed in Poland: Prussian dominance versus some other polity, be it the Electorate of Saxony or Russia, is associated with less support for populist or extremist movements in recent years. It should be noted as well that East Germany’s distinctiveness and radicalism is not new. Many German commentators have mourned the rightward turn in Saxony because Saxony was once a pioneering progressive state. But that misses the point: Radicalism, the appeal of revolutionary protest politics, is less about coherent policy platforms, and more about the appeal of mob, tribe, and movement. For all its legacy of militarism and authoritarianism, Prussia also left a legacy of education and stable bureaucracies.  Moreover, East Germany’s borders were not quite so new as we may think. East Germany was clearly a very different political world than West Germany well before communist domination, as can be in the electoral maps of 1933, 1920 ,1912, 1898, and 1893. To the extent that communist history is associated with a major political break, it may be partly tracing far more ancient divisions than the Cold War.

Indeed, from October of 2017 to October of 2018, the Protestant world is celebrating one of Europe’s oldest movement-based, mass-media-fed, anti-institutional, protest-oriented populist social movements: the 500^thanniversary of the Reformation. The Reformation began, of course, in Saxony, the AfD heartland. It is not too much of a stretch to suggest that populist revolt has been a crucial element of East German politics for at least 500 years, especially among middle-aged men dissatisfied with their lot in life, their treatment by formal transnational institutions, and their options in conventional politics.

Postscript: Much has been made of AfD’s mobilization of nonvoters. However, every party made net gains among nonvoters, that is, picked up more 2013 nonvoters than it lost to nonparticipation. Some parties did better than others of course, but two parties stand out for having over net participation changes account for over 10% of their 2017 vote: net participation increases account for 23% of AfD’s constituency-level voting or 20% of their party-list voting, while it accounts for 22% of FDP’s constituency-level voting and 14% of their party-list voting. In other words, AfD did slightly better than FDP at getting nonparticipants to show up and vote. Overall, FDP picked up more CDU/CSU voters than AfD as well, as well as more Greens, first-time voters, and new residents of Germany. The only category where AfD truly excelled was in taking voters from Die Linke.