As politicians polarize, Americans are also sorting into clearer partisan camps who dislike the other side. What reinforces that cycle? Perhaps both our social relationships and our increasingly unequal society. Tobias Konizter finds that Americans are increasingly selecting spouses based on partisanship and then passing on our political views to our children. But John Kuk finds that economic inequality is increasingly tied to divisions on social and racial issues, which in turn are driving dislike of the other party. Our social ties, resources, and racial views are all dividing us into two partisan sides.

The Niskanen Center’s Political Research Digest features up-and-coming researchers delivering fresh insights on the big trends driving American politics today. Get beyond punditry to data-driven understanding of today’s Washington with host and political scientist Matt Grossmann. Each 20-minute episode covers two new cutting-edge studies and interviews two researchers.

You can subscribe to the Political Research Digest on iTunes here.


Grossmann: This week on Political Research Digest, how marriage and inequality reinforce partisan polarization. From the Niskanen Center, I’m Matt Grossmann. As politicians polarize, Americans are also sorting into clearer partisan camps who dislike the other side. What reinforces that cycle? Perhaps both our social relationships and our increasingly unequal society.

In a new Journal of Politics article, “The Home as a Political Fortress” with Shanto Iyengar and Kent Tedin, Tobias Konizter of PredictWise finds that Americans are increasingly selecting spouses based on partisanship, and then passing on their political views to their children, giving our families better matched views.

But perhaps rising inequality is also contributing. Along with Konitzer, I also talk with John Kuk of Washington University about his new paper, “Inequality and Public Polarization.” He finds that economic inequality is increasingly tied to divisions on social and racial issues, which in turn is driving dislike for the other party.

We’ve long known that partisanship travels in families, but Konizter and colleagues find that spouses now agree more with on another, and that may be just as intended.

Konizter: Eighty-two percent of spouses agree with each other politically. We also present pretty strong evidence that, again, the reason behind it, the causal mechanism, is not one of what we would call spuriousness, so basically the idea that it’s related to some other phenomenon, basically just an observed thing. But it is due to active selection. If I’m a Republican and I go look for a spouse, and I go on the dating market, the political underpinnings of my potential partner become increasingly important in that selection mechanism.

Grossmann: Agreement in marriage used to come more from persuasion as marriages developed, but now it comes from active dating.

Konizter: The early studies that started to be published in the late ’60s and early ’70s always made the point that the agreement rate was already pretty substantive, but the reasons for the agreement rate were very different than active selection, and that sort of hinges back to I think the state of society in the ’60s as well. While you were sort of in the mating process, political underpinnings merit very little, but then after you were married, by a processes of persuasion and political conversion, especially of the female mate, you arrived at these sort of also very high figures of what we would call political homogeneity among spouses.

And then really there was not much else in terms of studies until the early 2000s where we had … people started to get access to all these online dating sites, which allowed us to sort of study the political underpinnings of mating as they occurred. And these studies come down pretty heavily on, yes, politics is a big factor in seeking out mates and seeking out spouses.

Grossmann: Konitzer used voter file data to find millions of spouses.

Konizter: We also are in the unique position where we have massive databases, specifically the American voter file, that we can throw at the problem. And not only is it a one stop voter file, or a single voter file. It is a voter file over time that we can throw at the problem, and look at not a sample of, I don’t know, 10,000 potential couples, but a sample of close to 20 million spouses.

Grossmann: The new data shows spouses are not just in agreement after long marriages.

Konizter: We used to have a homogeneity rate among spouses that were 72 or 73 percent, and now we have one that is 82 percent. Yeah, you know, that’s some increase, but I wonder what the substantive effect of that is. To me, what is more important here is, again, the causal underpinnings of that number have changed. It used to be inadvertent, used to be persuasion after the initial sort of marriage was conducted. The survey data from the ’60s sort of suffered from the fact that those were all spouses who had 17 or 18 year-old kids. So they had had time plenty of time to converge in matters of politics. Very hard to set off the selective aspect. How much does actually political underpinnings matter in the selection process, right? Very hard to study with that data. And we have just much better data available to study the political underpinnings of dating as dating occurred.

Grossmann: Comparing the newly married between 2014 and 2016, they saw limited changes. They were already in agreement before.

Konizter: If we can take multiple voter files that go back a couple of years, we can match individuals across voter files. So if I look at the 2014 file, I can look at potential, prospective spouses, who at this point at least don’t live together. They are not married. But in 2016, they have completed that process, and either they at least live together or they are actually married, or they have the same last name. And we go through different iterations of that, but to us that was sort of the closest that we could get to study mating as it occurred with our massive sample of 20 million spouses, right?

And one of the things that we find, of course, is that even if you look at partisan homogeneity rate among prospective spouses (that’s the people who did not know each other yet, and they hadn’t started dating, at least by every indication that we had), partisan homogeneity was already pretty high. And even outstripped educational or homogeneity in educational attainment.

Grossmann: And now spouses agree on policies issues and attitudes toward social groups, not just partisanship.

Konizter: You go back to the ’60s estimate, there was correspondence on issue beliefs or policy preferences was not very high. Whereas we find that the agreement rate among policy preferences for spouses is consistently above 8 percent, so that in fact is a massive increase.

Grossmann: Geographic polarization is also increasing, but they were able to rule out the possibility that it was just local dating markets comprised only Democrats or Republicans.

Konizter: We are able to group spouses by the political makeup of the zip codes in which they reside, and we are able to look at zip codes that are extremely heterogeneous, you know where you have an abundant supply of Democrats and Republicans, and this is by party registration, here.

Then we ask, “Do spouses that live in these very heterogeneous zip codes are even census blocks, differ from spouses who live in areas in which the partner market is very constrained?” Meaning if you’re a Republican who married a Republican, but you also happened to live in an area that has 90 percent of registered Republicans, that might not really be causal, right?

And what we do find is that in these more heterogeneous areas, be it neighborhoods, as in census blocks, or zip codes, the homogeneity rate among spouses goes down a little bit, but it never really dips below 80 percent. And so, yes there is definitely some moderating effect here of geographic heterogeneity, but we find it to be very slim.

Grossmann: And political matches between spouses were even higher than those based on religion or other social factors.

Konizter: Well, you know, if you’re an Evangelical Christian you would seek out another Evangelical Christian, chances are you’re both Republican, but that was actually not the driver behind your decision to partner with that prospective spouse….that is in some ways the hardest causal thread to debunk.

We tried to do that with our survey data, and just basically backing out estimates of what was your religious agreement on the day you got married? What was your partisan agreement on the day you got married? And how do these estimates compare? And we find that partisan agreement was higher.

But there is one other thing that we try to test, which is to say that if that alternative is true. Where it’s really about religious affiliation, or it’s really about authoritarianism, especially as it relates to the issues in child rearing.

Grossmann: But it could still be the case that partisanship measures a broader level of agreement on values between partners.

Konizter: It’s very, very hard to disentangle those two, but I think it is also that these alternative causal things have become pretty politicized as well. I think one of the reasons that we find this really, really partisan homogeneity among spouses is that partisan homogeneity, or partisanship, is a heuristic basically for the value frame of your prospective spouse.

Grossmann: Married people pass on their partisanship to their children, just like always, but more so now with enhanced agreement and politicization.

Konizter: We confirmed the socialization literature that also stems from the late ’60s to early ’70s, which is to say that successful transmission, which in that case means that your children agree with you politically, hinges on two facts. It hinges on the consistency of cues, right? So if you have two very progressive parents, that means that that transmission process is obviously going to be more successful, and it also hinges on the frequency of cues. So if politics are discussed on a day to day level in the household, that transmission, or that consistency of political beliefs from one generation to the other is going to be much higher.

And that is exactly what we find.

Grossmann: The results are bad for increasing negative political attitudes, but they may be better for the marriages.

Konizter: We do find that among spouses who politically agree, levels of affective polarization are much higher, and that is especially relevant for out party feelings. So if you’re Republican and you’re married to a Republican, that means that you “hate” that Democratic nominee for president, or candidate for president much more than the average Republican.

So on the one hand, we do see that it fosters this sort of more classically defined affective polarization. Now on the other hand, I also believe that it is a consequence of more efficient sorting, right? And that’s something that I said before, which is that in the old days you did not have as much information about your prospective spouse. Now I think partisanship really works so that you can judge your potential spouse alongside many, many different dimensions, and you can do that pretty cheaply with just partisan affiliation.

Grossmann: John Kuk is trying to connect two troubling trends, rising economic inequality and political polarization.

Kuk: In politics we have seen increasing polarization among politicians, and in American economy we have seen growing economic divide between rich and the poor. And it’s not that these two trends are separate. McCarty, Poole, and Rosenthal showed these two trends are really intertwined to each other.

Grossmann: He finds that growing economic differences are tied to social polarization rather than rising differences on economic views.

Kuk: Growing economic differences are dividing our country, more specifically our voters, in racial, social, cultural issues. And even in their emotions.

Grossmann: The connection between economic differences in the public and political differences in Congress has been hard to draw. Prior research has assumed that inequality should be tied to economic attitudes, since it reflects economic circumstances.

Kuk: The one level is about the mass public. The other one is politicians. So we need a connection between those two. But that connection was really unclear. And there was some conjecture about trying to understand the connection, and in this case existing studies said it’s the economic issues, like issues about tax or welfare are dividing us. It’s not the other ones.

Grossmann: Kuk looked at survey data since 1980, tracking Americans’ attitudes along two dimensions and finding a clear distinction between questions that asked about economic attitudes and those that asked about social and racial views.

Kuk: I wanted to compute ideology scores in two dimensions. Politicians think about politics as liberal or conservative, one dimension. But lots of political science studies show actually voters are not thinking in this one dimension.

So I needed a model that can compute ideology in two different dimensions, and I assumed that dimension would be, one, economic dimension and the other one being non-economic dimension, racial issues or social issues.

So with this model, I used a survey data. So in the economic dimension, most of the survey questions  were about federal spending (e.g. “Do you want to increase or decrease federal spending?) was loaded in the economic dimension.

And the social issues/racial issues dimension are all loaded with issues regarding race and culture. For example, school prayer, does the society need to provide more jobs for blacks? Do we need more equal rights for society? All these related questions.

Grossmann: But partisan differences were rising alongside inequality only on social and racial views.

Kuk: When I wanted to see the pattern from 1980s to 2012 and see the correlation with inequality, I found the pattern of inequality and polarization in the social/racial issues dimension are highly correlated. Both two trends are going up. But the pattern of polarization in the economic dimension is pretty flat, and there’s no relationship with inequality.

Grossmann: The polarization does not reflect more extreme views, just the lining up of social issue attitudes and partisanship.

Kuk: It doesn’t necessarily mean that my finding suggests partisans are having more extreme view. This result can also suggest that Democrats and Republicans are sorted to the party that fits with their views about these issues.

Grossmann: Kuk found that racial resentment was the big correlate of these social views.

Kuk: When I ran a model to see which is highly correlated with the social/racial issues dimension, it was the racial resentment question, and actually it was three times stronger than the other factors.

Grossmann: But economic circumstances might still be the culprit for changes in these racial views.

Kuk: For individuals who used to have a higher social/racial hierarchy, when they are facing more economic hardship and economic distress, they want to prevent losing their position in this hierarchy. So developing more racist attitudes is one way to preserve their position. And also situations related to economic distress, for example, a threat to themselves or their group, higher uncertainty, or need for more self esteem leads individuals to develop more strong ingroup solidarity and outgroup deridation. So when individuals face more economic stress, they tend to have more worldview of dividing the ingroups and outgroups.

Grossmann: Social issue polarization also caused increases in affective polarization.

Kuk: I ran a statistical model to understand how individual partisan emotional difference between my party versus the other party are related with these two ideology scores. And I wanted to see how this is changing over time. And what I found is that over time the correlation between the social/racial issues dimension and partisan affect is getting stronger over time. And the correlation between the economic dimension and partisan effect is actually decreasing over time.

Grossmann: Connecting a mass trend, economic inequality, with an elite trend, Congressional polarization, is still not obvious. But Kuk thinks the path likely runs through changes in the social and racial views of partisans.

Kuk: For income inequality, it’s about the mass public, the economic differences among mass public.

And without connecting these two, these two are just intertwined by just coincidence. So I wanted to understand why these patterns from different parts of society are connected. One way to understand that is understanding polarization among the public.

Grossmann: Nolan McCarty, Keith Poole, and Howard Rosenthal were the first to draw attention to these connections, but they found the impetus in rising immigration. Kuk says that attention is deserved, but not because it changes redistributive views.

Kuk: What’s happening here is not about redistribution in general. It’s more about how racially and culturally they perceive about foreign-born population immigrants. So I would say it was correct to point the rise of foreign born population, but it was more about racial or cultural attitudes about these perceived outgroups.

Grossmann: Konitzer says increasing inequality and polarization may both be aligned with social sorting in marriage, but he doesn’t think that’s the main explanation.

Konizter: What we call assortative mating sort of lies on the same dimension as inequality to a certain extent, because they both touch on more efficient sorting, right? Although, I will say that the increased rates of partisan homogeneity are true across educational attainment. They differ, but they’re definitely true across educational attainment.

Grossmann: Kuk’s next step is to find out why politicians led voters in this direction. He thinks it has to do with the Republican Party’s incentives to avoid becoming a minority, economic elite party.

Kuk:  The rise of racial and social issues are actually driven by their need for increasing inequality. They want to keep the voters 50/50. And appealing to this agenda is really effective. Plus, I believe regions that are experiencing more economic hardship are more likely to be prone to these new agendas. And I don’t think it’s only nationalistic. Others, like social and racial issues, are all together in this bigger strategy. When there is more increasing inequality, especially for the Republican Party, they need support from voters who are experiencing stagnating wages and more difficult economic situations. So the way how the Republican Party, especially politicians from regions that are going through economic stagnation, they found these new agendas, like especially racial and social agendas that can actually make voters more divided, and also keep them supporting the Republican Party.

Grossmann: Konizter has moved from academia to practitioner. And he sees the next step as using new sources of private sector data to continue tracking these trends.

Konizter: Studying American public opinion and insights into American preferences, preferences of American value frames of Americans, becomes more and more important. And it becomes even more important (now that I’ve changed sides from the academic perspective to the practitioner’s perspective) that we make sure that these data collection efforts are ongoing and in progress, and are not one-shot things. They’re not cross-sectional data collections.

Grossmann: There’s a lot more to learn. Political Research Digest is available biweekly from the Niskanen Center and on iTunes. I’m your host Matt Grossmann. Thanks to John Kuk and Tobias Konizter for joining me. Join us next time for a special one-year anniversary edition, where I’ll be joined by my “Asymmetric Politics” co-author David Hopkins to discuss how the Democratic and Republican Parties are changing.