People who are pro-choice on abortion also tend to be against the death penalty. But are those political beliefs internally connected? Would changing one belief change the other? And do all of our views add up to a coherent belief system? Mark Brandt finds that beliefs that tend to go together in the society as a whole don’t necessarily map onto how we see them fitting together in our own heads. And the internal relationships between our ideas can be better used to predict the dynamics of opinion change. We do have coherent belief systems, but they may not match our societal divisions.

Guest: Mark Brandt, Michigan State

Studies: “Measuring the Belief System of a Person” and “Belief System Networks Can Be Used to Predict Where to Expect Dynamic Constraint


Matt Grossmann: How we connect our political beliefs, this week on the Science of Politics for the Niskanen Center. I’m Matt Grossman. People who are pro-choice on abortion also tend to be against the death penalty, but are those beliefs internally connected? Would changing one belief change the other? And do all of our views add up to a coherent belief system? Despite long debates about the ideological innocence of the American electorate and increasing polarization, we still lack an understanding of how individuals see their political views fitting together.

This week we investigate the psychology of political beliefs. I talked to my Michigan State colleague Mark Bran about his new work on belief system networks. Including “Measuring the belief system of a person” in the Journal of Personality and Social Psychology, and “Belief system networks can be used to predict where to expect dynamic constraint”, in the Journal of Experimental Social Psychology. He finds that beliefs that tend to go together in this society as a whole don’t necessarily map on to how we see them fitting together in our own heads. And the internal relationships between our ideas can be better used to predict the dynamics of opinion change. Here’s our conversation.

So Mark, what is a belief system network and how can it help us understand people’s politics?

Mark Brandt: Yeah, so I mean I think a belief system network is just a convenient label that I give it, when I talk about it or study it. But I think the first question is maybe what’s a belief system to begin with? And when I think about belief systems, and I think a lot of social scientists when they think about belief systems, it’s a sort of connection of attitudes or maybe attitudes and identities related to politics. And it’s characterized by the connections between those things.

So it’s not necessarily liberal or conservative or libertarian or whatever your preferred positions are, but it’s about how they’re structured and interconnected with one another. And it’s this system or interconnections of attitudes, or attitudes and identities, that we call a belief system. And then we think just methodologically and theoretically it’s useful to think about that as a network, because there’s been this whole literature on network science that helps to try to understand the dynamics of things that are interrelated. And that’s what we think we can maybe draw on, some of that, to learn some things.

Matt Grossmann: And here the networks are not between people but between ideas or opinions.

Mark Brandt: Exactly. So typically in political science and psychology, we’ll have studies on social networks, maybe looking to see if people with more similar political attitudes they tend to hang out or how they might influence one another, and this sort of thing. And these are really interesting studies and they’re just not what we’re doing. So instead, we’re looking at these networks, the networks here, the nodes in the network or the key focal parts of the network that we’re looking at, are people’s attitudes or their identities. And then the things that link them together, which are sometimes called edges, where those are, how those things are connected, whether they’re positively connected or negatively connected or stronger or weakly connected.

Matt Grossmann: So as an example, we might think that someone’s opinion on taxes might be related to their opinion on government spending, maybe more than their opinion on abortion or something like that?

Mark Brandt: Yep, that’s right. And then those things might be linked to further attitudes or potentially different values down the road. So it might be that there’s a value of equality that links up a bunch of different attitudes together. Or there might be particular identities, maybe people’s identity, identity as a Republican or a Democrat, that also links to these policies together. And so these links that draw connections and hold these belief systems together.

Matt Grossmann: So you say that typically we have thought of or at least measured these kinds of belief systems with aggregate data in a population, that says which beliefs usually go together. But you think of them as properties of individuals. So explain that distinction, and maybe why we shouldn’t think of a belief system as something that goes with a culture or a political community?

Mark Brandt: So I think there’s, I don’t want to have too strong of a distinction here, in part because I think belief systems are in part influenced by people’s groups and their cultures and the country or political system that they’re in and this sort of thing. But I think what we’re measuring them and trying to get a sense of what’s happening, I think it’s useful to figure out what’s going on within the individual. So I might have a belief system that is connected in one particular way and Umap might have another one and our listeners might all have different variations of them. Now some parts of those belief systems might be similar. So we may all see taxes and inflation going together, but we might vary in how strongly we see those things going together, and we might connect those attitudes to other kinds of identities or values or other attitudes in different ways.

So maybe I see my attitudes about abortion are connected to some of these other issues, whereas maybe your attitudes on abortion are disconnected or connected to things in a different way. When we look at these in the typical political psychology, political science sort of way, we don’t really have the tools to do that. So in the typical studies we are often taking just survey data that we have, which often has a lot of strengths to it, the representative samples, the variety of issues that you can look at, and we look to see how they’re correlated.

But that just gives us that estimate for that one survey or that one sample. That’s interesting, but it doesn’t capture any nuance or differences among the population. Instead it just captures what attitudes do people tend to hold together at the same time, which might be because people see them as connected, but it could happen for just a variety of other reasons that we’re not observing or not measuring. I felt like I answered a lot of stuff there. You can cut this out later.

Matt Grossmann: No, it’s good. So your intervention is to connect these different attitudes with people’s ideas about their similarity, either because they are asked how people connect these views in general, or how they themselves do it. So talk about that, that measure what you’re asking people and what you can conclude from it.

Mark Brandt: Yeah, so our way of trying to expand and improve upon this literature on belief systems and belief system structure, and to apply it to the person level, is we have this task that we call the conceptual similarity task. One clear downside is the name is annoying to say, but the idea here is that we essentially take a pair of attitudes, so taxes and inflation, or abortion and same-sex marriage, or pick whatever pairs of attitudes you want. And for each pair of attitudes we have people rate basically how likely they are that they think they go together. And to do this we’ve explored two different versions of this. The results are roughly the same. So if you’re a researcher, you should use whatever you feel like, but one of them asks about the average American. So average American, they hold a position on, they support, say, higher taxes, how likely is it that they are going to support pro-choice abortion policies?

And then you rate whether or not they think those are likely to go together or not. The other one variation that we do it, we ask them to report on their own attitudes. This is a little trickier because of course not everybody supports higher taxes or pro-choice abortion policies. So we have to rephrase this a little bit differently. And so this one we are asking people essentially, imagine that you support higher taxes, what is the likelihood that you would support pro-choice abortion policies? And rate how likely that is to be the case? And we take higher scores on that essentially that they’re more likely to see these things as connected.

So that’s just one pair of attitudes. We do that for, we’ll take around 10 or 11 attitudes and that’s all possible pairs of those attitudes. And if you do the math on that, it turns out that you have to do this between 70 and 100 times, which is not the most exciting thing that our participants have ever done. And I did all these studies myself as well. So I felt like if I’m going to put participants through this, I’m going to do it. But the idea here is that we can then take these ratings after this boring task, and get a sense of how they see things as being connected, or disconnected as the case may be.

Matt Grossmann: So we have direct questions about the connections between attitudes, which is definitely different than just correlating attitudes. But it still seems like people are being asked for an estimate of what things look like in the American population. So are they reporting a connection between beliefs or are they just reporting, “here’s what I see in the world?”

Mark Brandt: Yeah, so it’s probably a mix of both, but I think that’s okay, for a few reasons. So first of all, I should say when I first… I’d been giving talks about this belief system network approach and have some papers using some older methods. And one of the challenges with these older methods is what I explained before, is that we couldn’t get this individual estimate, which is what we really wanted. And people were like, “Well what if you just ask people?” And my response was like, “We can never ask. That would never work.” But then I just got frustrated with my other approaches and took this approach. So I had the same reaction that underpins the question.

But I think it’s okay, I think these work and I think we have some evidence that they work. So one is that we have some data where we’ll do things like controlling for just general political knowledge or political engagement. And the measures seem to be associated with those things. But not only those things, which I think is useful. There’s correlations there as you’d expect. People are more highly political knowledge there, where they see connections are what you’d expect from attitudes in the American public. But that doesn’t explain all the variance and it’s not actually that high of a correlation. I think it’s about 0.3 if I remember.

The other reason is, I think a lot of where some of these connections are coming from are people’s observation about what’s going on and what’s happening in the world. And there’s different theoretical perspectives from sociology and from political science and some from psychology about, part of where our belief systems come from is by observing how things are connected in the world. And you see how these attitudes tend to co-occur maybe among your favorite politicians or your friends or people that you work with. And those inform how you see them connected. And I think that, so if we’re picking up on that, that’s okay. Because I think that’s where some of these connections are probably coming from.

Matt Grossmann: So you also combine some things that political science usually keep separately, like attitudes toward parties, specific policy views, and maybe broader goals like environmental protection. So how do you see these things fitting together, and why do you treat them as just all just nodes in a network?

Mark Brandt: Well, so we could just do policy attitudes for example, and that would be pretty consistent with some approaches. I think a downside there, though, is that then you might find that a particular policy is maybe central in the network. That’s cool and maybe interesting. But we know from a variety of other work that it seems like partisanship is pretty important to people, matters a lot to people, and that if we’re not including that in our networks, then we’re going to have this kind of… We might see connections where there aren’t connections or we might see something as being central that’s actually, it’s not central so much as it’s just a proxy for partisanship.

And so the [inaudible 00:12:16], by having partisanship in there and these other potential values, we can more clearly get a sense of what’s important and central and holding these different attitudes together. I personally don’t like drawing firm boundaries at where these belief systems are, because I think we can then accidentally define ourselves out of the important and strong, and the most important constructs in these belief systems. And maybe some people might want to say, “Okay, only the policies count as belief systems and this other stuff separately as identities,” and they might make that conceptual argument and that’s fine. But I think it’s so useful to study how they’re connected, and we can just label it differently, that’s okay. I don’t mind.

Matt Grossmann: So tell us about the main things that you’re able to find in comparing how people rate these connections with their own preferences and their personal attributes.

Mark Brandt: So there’s a few things that we can do with these networks that I think are useful. So one is, we have some evidence that they do seem to at least impart structure than people’s actually expressed attitudes. So the networks that we’re getting for these different individuals, these are just estimated how related these attitudes are. It’s not whether they’re supporting an issue or opposing it or whatever. But the idea with this perspective is that these connections do influence the positions people actually take. So if you see tax policy and abortion policy strong early related, if you have the liberal position on the tax policy, we’d expect that to follow through and have you have the liberal position on abortion policy. And so if that’s the case, we should have see attitudes are more strongly connected, they should also be kind of attitudes that people hold more similar to each other so they can have similar positions on them, whether that’s liberal or conservative or in between. And that’s what we find. So we have some evidence that when you see attitudes being connected, you hold similar attitudes on that. We also have nice evidence… So let me back up one step. There is this a nice paper by Sarah Stein and Amir Goldberg who are sociologists and their argument is that part of where kind of our polarization and separation of… Yeah, the kind of polarization we see in society isn’t so much because of our social networks being only people that we agree with. But a big part of that is that we see other people as having these different attitudes that go together and that informs how we connect our attitudes and then this pushes us and these new connections push us in different directions.

One implication of their model, and this gets deep in the weeds, but one implication of their model is that people who have these similar belief system networks is they seek things connected in a similar way. They’ll respond to attitude issues and policy issues in a similar way with a similar logic. Now I don’t mean that they’re all will share the same view, but rather they see the same attitudes as kind of going together as you should have either the liberal position on taxes and abortion or the conservative issue on both of those. But those kind of people who have that would share a logic because they see of them as going together or not going together. And what we find in our data set, so with people who’ve never really even met each other, that people who have a similarly structured belief system have a similar logic in how they’re responding to attitude issues answered separately on our survey. So suggest that there’s some degree of organization that’s going on with between this kind of belief system network part and the attitudes that people are endorsing.

All right, so then the other kind of key piece of evidence here is we measure people’s… The extent that they’re strongly liberal or strongly conservative. We measure their political engagement. So just how interested they’re in politics, how much they follow it, and we measure their political knowledge. And what you find is that people who are more politically knowledgeable, more strongly identified with their party or ideological label and who are more politically engaged, they see attitudes that go together within the American public. So liberal issues with liberal issues or conservative issues with conservative issues as going together, whereas those that are kind of a liberal issue and a conservative issue are seen as being more dissimilar to each other. So it suggests that people have more highly organized belief systems, at least according to the kind of American way of organizing stuff when they are more politically knowledgeable, engaged, have these stronger political identities and these models you can control for all these things. And they all kind of come up independently. So it’s not all just a proxy for political knowledge or engagement, but there seems to be something unique about all of them.

Matt Grossmann: So one alternative that people in political science or are used to hearing political science will think about I think is just the ideological spectrum. And political scientists have found that most people’s views can be placed along an ideological spectrum. And that doesn’t mean we’re all liberals or conservatives, but there’s a lot of people that are in the middle of that ideological spectrum. And there are some people whose views don’t really fit the ideological spectrum, but that’s pretty small compared to the people who are just not quite on the fully conservative or liberal end of this spectrum. So kind of compare that view with the kinds of views that you’ve come up with here.

Mark Brandt: I actually think that that view is really consistent with the approach that I’m taking here. So if you think about the kind of standard way that political scientists and psychologists have studied belief system structure, we get a correlations between attitudes in a single sample. And what you find is often the case, not always but often, is that those attitudes tend to be all positively correlated. Maybe not particularly strongly, but they’re positively correlated. And actually the dynamics of that kind of belief system would push everybody to have a consistently liberal or consistently conservative belief system. But as you say, that doesn’t seem to happen for everybody. Happens for some people for sure, but not for everybody. One thing that you need to… If you buy that the connections between attitudes make a difference for the actual attitudes people endorse in the end to have these kind of more unique things that pop up, whether it’s like the classical liberal libertarian attitudes or the kind of socially conservative high social welfare supporting ideologies that you find parties for in other countries. And there’s some Americans who support that combination too.

You need a differently structure of belief systems where those things are connected in a different way that kind of push those things to happen to co-occur. So you need to have a way of getting a sense of whether there are people who see maybe stronger social welfare policies as going together with their socially conservative policies on things like gay marriage and abortion and so on. And the kind of classical methods wash that out. Whereas we are able to pick out some of that nuance and in theory should be able to identify the belief system structures that ought to lead to these different combinations of beliefs.

Matt Grossmann: So there are different combinations, but you still are mostly asking people which side are you on between a set of sides. And I guess what I would say is political science usually finds most people are in the middle of those sides. And so part of the difference might be that whether you would expect someone to have an environmental protection pro attitude and a pro spending attitude, but it might be that there are some people who are in favor of those things. How do you deal with that?

Mark Brandt: Yeah, so in a separate paper, we have simulations that kind of take the ideas of these models seriously and then just you can throw a bunch of different parameters at it, see where things shake out and where the people with the… You can’t see, but I’m using air quotes everybody. People in these simulations, they’re moderate when they have a couple of different things happening. So one thing that can have the sort of moderate positions or mix of positions is when the belief system has a mix of positive connections. So people see the attitudes as going together and negative connections. So in political science this has sometimes been talked about as a form of cross pressure where the different connections between attitudes are pushing against each other and that’s something that’s possible and it shows up a fair number of times with our participants in this sample.

So I think that’s consistent there. The other place the simulations show that you get this moderation, it’s just a different kind of cross pressure, but it’s more from maybe your… The demographic groups that you belong to or identify with kind of pushing different directions. So the rich religious person who pushes them in one direction, but maybe the rich secular person gets pushed in a more moderate direction on average because they want the lower taxes and maybe more socially liberal sorts of policies. And so these kind of cross pressures that maybe depends on how you conceptualize them or maybe more outside the belief system ish, those also push people towards those moderate directions even if the rest of the belief system would be positively connected and suggest more a clear ideological point of view.

Matt Grossmann: So you also use partisan frames to adjust people’s belief systems. What do you learn from that?

Mark Brandt: So this one paper where a lot of these studies that I’m referring to are in with the goal of this paper is just to see does this measure work as you might expect it to? And it’s just in some ways it’s a boring paper, but I think it’s a necessary paper in that we need to identify just if our measure works. And so one thing that we think based on a variety of prior theories is that people are getting these connections in their belief systems just because of what they observe out in the world and they see how they see things going together in the world. And so one way to kind of induce that is with a partisan frame. You take a relatively neutral attitude that people don’t know anything about, which is the case in our study. They know what these things were and we say this farm subsidy policy, Democrats really like it or Republicans really like it or we don’t tell them any information.

And what we find is that consist… Well, we replicate past work. So when you do that, people who are Democrats who hear the Democrats support the policy, they’re like, awesome, love the new policy, let’s get it. Republicans who hear the Republican, they do the same thing, just they respond to their in group party here. That’s what everybody does. What we’re able to then do here is we also then measured how they thought these things were connected. And what we find is that the partisan queue does seem to basically take a link where there wasn’t one and create one in the belief system between people’s partisan identity and this new issue that they didn’t know much about. So this for us was a nice way to validate that people are responding to this outside information and creating links in their belief systems like we’d expect from prior theories.

Matt Grossmann: So you also have used just aggregate data to try to look at connections between viewpoints, but you say that they mainly sort of reflect these stable differences between some people who have these views and some other people’s who have these views. So explain that and why it’s kind of the impetus for your method.

Mark Brandt: So this is annoyingly technical I think, but maybe if you’re listening to this podcast, that’s your thing. I don’t know. So basically when you have just a single survey, so this would be typical polling data, typical survey that you see in social science research, cross-sectional survey. In these surveys, it’s capturing people at one point in time. It’s when they open survey on their browser or for the 2% that answer their phone, it’s answering their phones and so on. But that’s just where they’re at in one time and space and we don’t really know how they got there. So did they get there because they’re like… Did we happen to get them on this particular attitude as they were moving between different attitudinal points and maybe they’re really variable and they move up and down all the time. Or it could be that they’re just always at that space. If you ask them this question and another time, maybe the same question all the time. And this has implications for then what we can infer and what these kind of correlations mean when we’re looking at cross-sectional data.

Because if it’s just capturing stable stuff, that’s useful, but it’s different than if it’s capturing where people are moving in this grand space of time and space and they’re going about their lives and so on. So we try to separate those out and so we have some longitudinal data so we can see how much people are moving and how much people are staying the same. We can separate out their movement over time from how much they stay the same. And what we find is that those cross-sectional correlations are really, really similar to where people are just at continuously. So if you take their average S score from a 10 year period, it’s really similar to the correlations between when we just take their cross-sectional scores. What that means then is actually these cross-section correlations, they’re not capturing really how things are maybe connected for a person which is in the person’s head as they move on one attitude, they’re going to move on another.

Instead, it’s capturing just how where the divisions are in society. So if you think about a big correlation between attitudes, not so much that one attitude is causing the other, but rather it’s just that people who are on one attitudinal team or political team happen to have these sets of attitudes and people on the other team happen to have these other sets of attitudes and a correlation’s more that this is a big dividing line between people and in society rather than that that’s necessarily how they’re connected between for a person inside their head.

And that’s important because then if this is all the case, then we can’t really use these cross-sectional correlations to infer where we might see a causal effect between attitudes, nor can we use that to infer where we might see where for a particular person, how they would move if they changed their mind on one issue or another. And that’s I think a pretty big shortcoming in a sense because I think a lot of the ideas about belief systems and when we’re thinking about attitudes being correlated, even if we’re not saying it in our papers, we’re guessing that that might at least be the case. And I thinking about them instead is like here, this is where the dividing lines are in society, I think is a way that more accurately captures what we’re actually measuring and is actually still interesting. It’s just interesting in a different way.

Matt Grossmann: So you do find that you can show connections between attitude changes by actually changing people’s attitudes on one issue if you have this information about how they connect their attitudes.

Mark Brandt: Yes.

Matt Grossmann: How do you do that and what do you take from it?

Mark Brandt: What we’ve done here is we run these kind of complex experiments where we have multiple waves of data. In the first wave, we try to estimate these belief systems for people and then, at a second wave, we give them persuasive messages and we try to change their minds on a target issue and then we measure their attitudes towards a bunch of different issues that are included in the belief system. There’s some perspectives that would suggest that if you change one attitude, you should basically never change these other attitudes. They’re all independent of each other and it’s just kind of happenstance that they look connected in any kind of way.

Other perspectives suggest, well, if they’re in this belief system where things are connected, at least under some circumstances, you ought to see changes on other attitudes. It’s a little bit more complicated than that, unfortunately. We’ve done simulations on this. If you assume things are connected and influencing each other, just changing one doesn’t mean that everything will change because things kind of hold each other in place, but there are cases where you can find this and in some studies we’ve now done this where we measure belief systems, give people persuasive messages, we change the targeted attitude, which is good, it means that our manipulations worked and we see changes on closely related other attitudes.

Now these other changes are not big, so don’t think that this is some really easily observed finding. There are secondary effects and persuasive effects tend to not be big anyway, so it’s a secondary effect on an already not big effect, but we do find that the changes on these secondary attitudes that are connected. The more closely connected ones will have a bigger change than ones that are more distantly connected in these networks, suggesting that when we think about persuasion, we think about messaging, these sorts of things and changing people’s minds, we’re not just changing people’s minds necessarily on one issue, but we might be having these knock on effects and secondary effects on other attitudes and issues.

Matt Grossmann: Just as an example, if I say that attitudes generally between gun control and capital punishment are pretty closely related, but they’re not that related to things like taxes, this would mean I can change somebody’s opinion on gun control and that might actually lead to a change in opinion, a smaller one, on capital punishment?

Mark Brandt: Yep, that’s the idea. These have gone by other different names and different disciplines. For example, in the famous Converse 1964 book chapter and work following from that, they talk about dynamic constraint. In psychology, which is actually embarrassingly, I found this late because I also came from it from the Converse angle originally, but we’ve talked about secondary transfer effects as kind of the idea from there. There’s a variety of works from multiple disciplines on this trying to figure out exactly when you’ll see these effects and when you won’t and that is kind of up in the air, but the general idea is that we might be able to see, in time at least, predict when we’re going to most likely see these knock on effects, these secondary effects and when maybe they’ll be more muted because the connections may be less strong or these other attitudes are held in place by other things, maybe other identities or other attitudes that are more distant from the manipulation in the network. These are things that people are actively working on trying to figure out how we can best understand these.

Matt Grossmann: The way that political scientists usually talk about how people get their views on things like policies that they may not know that much about, is that they get them from the parties, but sometimes we’ll talk about it more as just if they’re watching Fox News, they’re going to hear one thing against immigration, they’re also going to hear one thing pro gun control. I don’t know if we assume that they’re making some kind of an internal connection between them, we just usually talk about it as sort of repeating what you hear. Talk about what might be at stake, I guess, in that distinction.

Mark Brandt: Yeah, so first of all, I think it’s useful to know that when I think about somebody’s position on an issue, I would follow the standard that I think also people are getting their positions from other places. It might be what they watch on the news, it might be, if there’s personality effects on attitudes or whatever it might be, could be their position in life, so on. My idea is that we should focus not only on that, but also what these other connections are and that people’s attitudes might come from that as well, so it’s rather than an either or, I think it’s a both hand perspective.

That said, I think with the example of things like Fox News or even political parties, part of what they’re getting is maybe, hey, this is the key thing. I should think this about the Ukraine War now that I watched Fox News, but you’re also learning that the pundit that you heard talk about the Ukraine War, they also talked about CRT immediately before that and they are going to talk about the upcoming election and the politicians you should support later on in the episode. They’re also learning those batch of things go together. That’s what we’re thinking about, is how do people think about these things as going together?

In our partisan framing experiment, for example, the focal part was the link between partisanship and this new neutral policy of farm subsidies, but what we found that was interesting was that the partisan queue not only changed how partisanship was seen as being connected to the farm subsidies, but also how farm subsidies was seen as being connected to almost all the other attitudes that we measured. It seemed like this partisan queue not only created the connection between the focal parts of the queue, but also created these other connections as well, as people saw this new policy in a light that was maybe more consistent with the partisan vent of these other issues.

It’s more than just one connection was created, but many were and I think that’s happening when people watch Fox News or they listen to their favorite politician give a speech. They’re not just learning the main thing, but they’re also learning that hey, these things, they should be packaged together and should go together.

Matt Grossmann: Usually when we’re talking about belief systems in political science, we’re thinking of these intermediary views that might be above individual policy positions, things especially like the self-reported ideology on the ideological spectrum. Also, sometimes things like core values like egalitarianism or social traditionalism and I know that psychologists often talk about moral foundations or other kinds of summary measures. Where do those fit into these belief systems? Are they just other attitudes or is there some kind of overarching view that is not just connected to the others, but is kind of a summary of the others?

Mark Brandt: Yeah, I think we really know the answer to this question. The way that we’ve looked at this empirically is in some studies that we’ve already done, we have people’s views about the parties embedded within this conceptual similarity task and we have then estimates for these individual networks for each of our participants in our studies. If people’s partisanship, for example, is helping to serve as some sort of summary function for their policy views or if their policy views are causing their partisanship, either way, you’d expect that partisanship would just have a lot of connections with all the other attitudes. Abby Casar is a PhD student in the psychology department. She’s been looking at this actually over the last few weeks and what you find is that the partisanship parts of the network, they are more, for most people, not everybody, they are the most interconnected attitude in the network, so they’re connected with all these different policies and pretty strongly, but that’s not true for everybody.

I think it’s about 40% of the sample, they have other attitudes that are much more highly connected and so maybe are less summary-ish for them. We haven’t done this with values, although this is part of things we’re thinking about for the future and I think that’s going to be a similar thing with the values, is that people just see that as being connected to a bunch of stuff. When you ask people later on about their values and you say, are you more traditional or more progressive or whatever, that might be a summary judgment that they’re using, but at least how we’re thinking about them in their networks, I’m thinking about this, at least, as just being many more connections with other stuff and so that values or other kinds of broad political identities are broad and they’re abstract in part because they are able to be connected to all these other more specific policies.

That’s kind of how I’m thinking about them. Now whether that means they’re a summary or just another attitude, I don’t know. I think that almost becomes a little semantic, but this is how I’ve been thinking about them.

Matt Grossmann: We have been talking about attitudes as if they either come from partisanship or from individual policies, but there’s another alternative in both of our worlds that I know you’ve done a lot of work on, which is social identities. That is people might identify with a group, they might perceive that group as having certain kinds of views or they might perceive the out group as benefiting from certain policies. Fit your findings into that world, where we’re really motivated more by our social group identities and our views of groups.

Mark Brandt: You can think about it in a few different ways. I think of partisanship as just one of these identities personally and so when we’ve done these studies and included partisanship, to me, I’m like, oh yeah, that’s what we’ve done and so we think that these are potentially highly related. For scholars who are interested in how people’s identities are related, at least within this kind of way we’ve been thinking about it. I think our task, it’s totally possible to just swap out different kinds of identities, at least in some cases or views of groups or this sort of thing. It’s just not something that we’ve had the chance to do, but I think it’s totally plausible.

I think then what this can do is then help us answer different research questions. I have a hunch, but I’m not theoretically committed to any of these, but it could be that, well maybe people’s policies are really important and then the group stuff is kind of peripheral and it’s not so important. Other people would suggest that, hey, group identity is hugely important, big factor in our politics. If that’s the case, we should see them connected to a bunch of different policies and to be influential on those policies. I kind of think that’s what we’d find, but that’s something that’s definitely testable with this method.

When I think about belief systems, it’s these connections between these types of beliefs and I have a pretty broad view of what I would consider acceptable to be tested within that thing and I think people’s identities are, in some ways, an attitude about a group. It might just be a group that you belong to and your place in that group. I think that’s great and I think we can try to integrate that and see how that shakes out within this belief system type approach that we’ve been working on.

Matt Grossmann: In the original Converse conceptualization, there are some people who think about politics in terms of belief systems, but there are some that are better regarded as thinking about which side is trying to benefit what groups in society. There are others who are just thinking about how things are going right now and who is in charge and then there are some people who have no real understanding of politics. To what extent do you buy that what you’re doing should just apply to some portion of the population and really for some other set of people in society, belief system is not the way that they think about politics?

Mark Brandt: Yeah. The way I’ve been thinking about this is that it’s just kind of an empirical question whether or not somebody has a belief system. I wouldn’t say that everybody has a belief system or an organized belief system or anything like that. In our studies, there are people who basically respond such that there’s very little connection between things and other people who respond, it’s like everything is strongly connected all the time. Those are maybe our ideologues from the Converse stuff. I think the first are kind of more, who cares about politics people and so I think this is more just an open question. What’s nice, I think about our methods, is that if you don’t care about politics or you think nothing is connected, you can respond in ways that indicate that and I think that’s useful at least.

We haven’t looked to see if we get similar percentages to what Converse would expect, but we could. I think that would be interesting. But yeah, that’s what I think is nice about this is we’re not forcing people to have a particularly structured belief system or that’s not forced upon our participants. We’re asking them to tell us what they have. And that also means that there’s going to be some people who see economic issues and social issues as being part of the same clump, but other people might see them as much more differentiated and other people see that nothing is connected ever. And I think that’s kind of a virtue of our method is that we can get at some of this diversity of opinions and approaches to politics that I personally think exist out in the world of people.

Matt Grossmann: So I guess it’s a virtue that everybody can describe themselves and their views. On the other hand though, it might kind of put everyone in this box that I guess Converse would think only something people should be in. So I guess if, let’s say, there’s someone whose politics really are defined by nature of the times. That is, it really is how are things going right now and who’s in charge and that’s how they think about things. It seems like this wouldn’t be the right approach and so at some point we’d need to say of well we might be able to characterize your belief system, but that’s not how you think about politics.

Mark Brandt: Yeah, I think what would be interesting there, and I think it’s an open question, is if we could get these kind of measures over a relatively longer period of time. So, in our own studies we have these measures over a one week period of time and over that period of time they’re pretty stable. But that’s not enough to capture the nature of time kind of argument. But if we have them available for a longer period of time, look at those people who change over that time, look at the people who stay pretty similar over that time and see if this… Takes a lot of forethought to do this. But have them do experiments where you try to change their attitudes and see if those experiments have different implications for those people who changed over time. And that would give us a sense of whether this is really how they’re thinking about things or if this is just reporting something that they’re observing out in the world without getting deep into their psychology of how things are connected.

Matt Grossmann: So you work at the intersection of psychology and political science, and obviously we’ve been talking about drawing from both literatures, which I know that you do. But I thought it was a good opportunity to tell political scientists the way that psychologists think that might be useful to them. So what do you think political scientists don’t pay enough attention to in the part of psychology that’s most related?

Mark Brandt: Yeah, it’s a good question actually. So, I think most scholars do fine. So I think people who are keeping up with the literature and so on are doing fine. This is not such a big problem overall, which I think is maybe different people’s to have different mileage may vary with that. But at least within this kind of ideology and belief system space, there’s a lot of kind of back and forth which I really like. I think one thing that I would like political scientists to think about more I think is how things might be more specific to an individual.

I don’t think that this is something that nobodies thinking about, just to be clear, because I can already hear the counter arguments in my head to this statement. But I think there’s something to be said about thinking about what it would mean to have more of a person focused psychology of political ideology or political belief systems. We’re often really focused on big picture stuff or at least on the political science side, especially institutions, political parties, these are obviously important. But when we think about a person and their kind of daily life, they don’t really experience parties or political institutions very much. They experience their friends and their family, they experience their workplace or just TV, whatever normal TV not Fox News, not MSNBC. They’re watching soap operas and things that are funny and so on, good TV. Thinking about how these things create a environment that not everybody’s going to have similar politics and it’s not going to just be purely a engaged, not engaged story, but all of these different experiences, their identities within these different opportunities to learn about politics but also about other parts of life might combine to have these really individualized experiences and thoughts about politics and worldviews. I think trying to think through what kind of the science of political ideology or political belief systems looks like from that point of view I think would be really interesting and I think would be a useful step forward.

Matt Grossmann: And what about from the political science side, is there anything you’re repeatedly telling psychologists that they should hear from political science?

Mark Brandt: Yeah, so I mean psychologists are really pay more attention to the institutions and the political parties and things like that. The thing that keeps me up at night, I guess as a researcher in this space and this might apply to political scientists to some degree too, but it’s just that everything we’re setting is just an accident of the times and that if we go back 10 years or we go forward 10 years, it’s going to be totally different and not because of any deep insight, it’s just kind of happenstance. For whatever reason, we’re all just studying happenstance.

I think this is maybe less of a concern for political scientists because they’re often the implicit and sometimes explicit goal is how is politics happening right now? For psychologists, our implicit and sometimes explicit goal is how do people work generally? And so if it changes radically over a couple of decades, well it suggests that we were wrong about how they operate generally. And so then I think we need to think about how do we develop general theories of politics that make good predictions about what happens, what happened two decades ago, what’s happening now and what’ll happen in a decade further? And maybe that’s not possible, but if we want to have a general kind of science of these things, that’s something we need to I think focus on and I think is challenging for anybody.

The other alternative here though is to take more of the political science type approach, which is to treat kind of social psychology more as history in a sense. We’re documenting what psychology looks like now or what psychology belief systems look like now and in a decade we’ll document it again and it might be the same or it might be different and we’ll find out. I think that’s seems to be less satisfying to psychologists overall, but I mean it might be what we’re doing just by accident. And that’s okay, but we maybe should be aware.

Matt Grossmann: So you say that we assume a lot more temporal validity maybe than we can. There’s also geographic differences that I know that are getting a lot more attention. So the conventional split is that in political science we have people call it Americanists, just like history, who think they’re studying the US only. Psychology, but we have 80% of the data coming from the US. So, talk a little bit about that. Are we studying belief systems or are we studying the American public?

Mark Brandt: I think it depends of course on the study for example. Well, so I’m a political psychologist, that’s my sub-fields or whatever, but there is a branch of psychology called cultural psychology and a closely related branch called cross-cultural psychology. Please don’t ask me the difference between those two, but people in them know there’s a difference between the two. They are studying other cultures and it’s kind of similar to the comparative just in political science. I guess as a social psychologist who mostly studies Americans, I should maybe call myself an American social psychologist to speak consistent with what’s actually happening.

But I do think the cross country differences are really important and are often understudied. There are some people who have been doing a really nice job taking advantage of things like the World Value Survey, European Social Survey and these sorts of things to try to get a sense of where there’s similarity and differences, which I think is really nice. And some of the studies I talked about today, or at least in the background of some of that we’re we are using data from other countries. So I mentioned a 10 year study that was data from the Netherlands for example. There actually what’s kind of interesting is in that study we have data from the US that’s longitudinal and we have data from the Netherlands that’s longitudinal. But people’s positions about parties can’t be the same. The Netherlands has, I think around 20 parties in the parliament at the moment. The US has our normal two plus a few weirdos essentially. So you can’t ask what those things the same.

But the policy stuff is not radically different in how it’s correlated and a lot of the properties and the data were otherwise the same attitudes were similarly stable. The kind of size and of correlations were similar and the difference between person societal level stuff and within person overtime stuff is very similar across the studies. So I think there are going to be stuff that’s similar, it’s just a matter of kind of identifying what’s going to be similar beforehand and what’s like due to just the different parts because of it’s different country and different system versus what’s about the psychology of the person. And I think there’s people working towards that, but it is a big issue and I don’t think we’ll have that solved anytime soon.

Matt Grossmann: So your chance to tell what’s next, what are the biggest open ended questions here and what are you working on?

Mark Brandt: Yeah, so I think there’s a few things that’s just to tap into what we talked about. So one is that right now we’re going through and planning some new studies using this conceptual similarity measure, looking to see how this works out over time. Again, that’s super long time so we’re not going to get your nature of times arguments in here. But looking at if people with differently structured belief systems maybe have more stable attitudes, so people who are connecting their attitudes more to partisanship, do they then have more stable attitudes over time than people who are connecting their attitudes to other things. In the same study, because it’s going to be a pain or might as well hit multiple things at the same time, we’re also looking at and seeing if we can run some more of these persuasion studies and looking to see if we can targeting different attitudes if this has different sorts of dynamics depending on the structure of people’s belief systems. And so those are some things that we’re working on now.

The other thing that I’m kind of interested in that we have kind of started to look at in different realms is this idea that people’s political attitudes and kind of how they’re related is kind of in some ways a very person specific thing. We’ve been looking at longitudinal data, looking at how even things like the strength of partisan identity and its relationship with effective polarization, how that actually what we find is that’s really different across people.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host Matt Grossman. If you like this episode, I think you’ll like our prior discussions of partisanship, ideology, and public opinion. Check out the American public’s growing ideological sophistication, how primary elections enable polarized amateurs, racial protest violence and backlash, when liberals and conservatives use genetics to explain human differences, and how political values and social influence drive polarization. Thanks to Mark Brandt for joining me. Please check out measuring the belief system of a person and belief system networks can be used to predict where to expect dynamic constraint, and then listen in next time.

Photo Credit: iStock