1 - A House Divided against Itself?

Summary

Much of the modern study of mass political behavior in the United States often returns to three books released during the Eisenhower administration. Voting by Berelson, Lazarsfeld, and McPhee (1954) approached its subject from a sociological perspective. Anthony Downs’ (1957) An Economic Theory of Democracy is the foundational study of political decision-making from the rational-choice perspective. The American Voter by Campbell et al. (1960) pioneered the use of the mass survey for political research. These approaches to studying politics are ubiquitous now, but, at the time, these were pathbreaking methodological advances. The authors of these books were to the study of politics what Chuck Berry, Little Richard, and Elvis Presley were to popular music.

We need some people who are active in a certain respect, others in the middle, and still others passive …. How could a mass democracy work if all the people were deeply involved in politics?

Berelson, Lazarsfeld, and McPhee (1954)

Much of the modern study of mass political behavior in the United States often returns to three books released during the Eisenhower administration. Voting by Berelson, Lazarsfeld, and McPhee (1954) approached its subject from a sociological perspective. Anthony Downs’ (1957) An Economic Theory of Democracy is the foundational study of political decision-making from the rational-choice perspective. The American Voter by Campbell et al. (1960) pioneered the use of the mass survey for political research. These approaches to studying politics are ubiquitous now, but, at the time, these were pathbreaking methodological advances. The authors of these books were to the study of politics what Chuck Berry, Little Richard, and Elvis Presley were to popular music.

While these books are rightly praised for their insights, we want to briefly highlight their titles. The titles clearly state what the books are about and make it clear that these books are not shy in their ambitions. These books are about voting and democracy, and this is obvious to someone who can only see the spines of the books.

The title of this book is more of a mystery. What is the other divide? And if this book is about the other divide, this implies that another book could have been written about a different divide that is unstated but clearly important – after all, the divide at the center of this book is the other one and not the one that everybody is thinking about.

Since this is not a detective story, let us solve both of these mysteries at the start of the book. The other divide is the divide between those people who make politics a central part of their lives and those who do not. The unstated, more familiar divide is the partisan divide between Democrats and Republicans.

The partisan divide should be more familiar because there is no shortage of research articles offering evidence of its presence, most recently through the lens of affective polarization. Democrats and Republicans do not want to have dinner together (Chen and Rohla 2018); they appear to see the other party as less than human (Martherus et al. 2021); they would be upset if their child married someone of the opposing party (Iyengar, Sood, and Lelkes 2012); and they may even be happy if someone of the other party contracted a debilitating illness (Kalmoe and Mason 2019). Coverage of this partisan animosity has also become something of a news beat. Between the summer of 2018 and the summer of 2019, for example, The Washington Post published more than fifty articles invoking partisan polarization; The New York Times published nearly twice as many.

The other divide – the divide in people’s focus on politics that is at the center of this book – is actually not less documented. In fact, the books we mentioned at the start of this chapter all allude to this divide through studies on political attention. Both Voting and An Economic Theory of Democracy suggest that differences in levels of political attention are important to democracy. In a section titled “Involvement and Indifference,” Berelson, Lazarsfeld, and McPhee (1954) write that democracy functions better with “a distribution of voters than by a homogeneous collection of ‘ideal’ citizens” (315). Downs (1957) describes a division of labor in which masses of inattentive members of the public can free ride off the efforts of the smaller number of politically attentive citizens. On the other hand, the authors of The American Voter were less sanguine about the large proportion of the public who “pay much less attention to political events than is commonly realized” (Campbell et al. 1960, 182). They document the failures of the inattentive public, writing, “many people fail to appreciate an issue exists, others are insufficiently involved to pay attention to recognized issues, and still others fail to make connections between issue positions and party policy” (Campbell et al. 1960, 183).

Individually, some of the authors of The American Voter had still bigger concerns. Converse (1962), for example, worried that there were some people who were so “uninvolved” in politics that, even during elections, they received “no new relevant information” (587). He did not entirely blame the uninvolved for this outcome; media coverage of congressional candidates, he wrote, is “buried in such a remote section of the paper” that “it is no wonder that data that we have collected over the years show a large portion of citizens who fail to be aware of their congressional candidates as individuals at all” (Converse 1962, 586). He also wondered whether people who are so uninvolved in politics can engage in the type of self-governance that is required for the maintenance of American democracy (Converse 1964). For Converse, then, this divide in people’s attention to politics was not the “other” divide – it was the focal divide.

Six decades later, times have changed for both divides. The American Political Science Association (APSA) of the 1950s was concerned that the parties were not divided enough (APSA Report 1950). In 2020, the presidents of APSA wrote an op-ed noting that “doubts about whether the election will be fair are being raised from all directions” – an outcome, they suggested, fueled at least in part by deep-seated partisan divisions in America (Aldrich et al. 2020). The emergence of new media technologies means that people no longer have to seek out what Converse (1962) had termed the “remote section” of their local newspaper to learn about their congressional candidates. It is now easier for even the most casual, most “uninvolved” news consumer to come across “relevant” political information. Yet, although increased media options give people many more ways to learn about politics, the diversification of media also makes it easier to avoid politics altogether (Prior 2007) – potentially exacerbating the divides in political attention. These differences between our modern era and the post-World War II time period set the stage for the thesis of this book: The growing partisan divide in America can only be understood in the context of the growing gulf between people who spend their day following politics and those who do not (i.e., the other divide).

As we will suggest in this book, people’s focus on politics – which we will refer to as “involvement” – is best considered as a continuum. For the time being, however, it is easier to understand our argument if we can divide citizens into three groups. Some people are, to use Converse’s (1962) term, “uninvolved”; they are like the Nebraska respondent in his study, who explained that they “don’t just know what the parties have been up to lately” (587). Some people, a much larger group, are more likely to behave in the ways Hutchings (2005) suggests: They focus on politics when something happens that is important to them. Finally, there is a third group of people, a group whose focus on and attention to politics is outsized; they are, to foreshadow our core argument, deeply involved in politics (a term taken from Berelson, Lazarsfeld, and McPhee [1954]).

The divide between this third group, the deeply involved, and everyone else is key to understanding modern American politics. It would seem natural to think this third group has a lot in common with the second group – those who are sometimes involved in politics. Both of these groups know the basics, they likely know what is going on in the news, and they typically vote. But in this book, we will argue and show that the deeply involved group is unique in a variety of ways that are consequential to American politics.

It is this deeply involved group, we will argue, that has affected how many political observers evaluate the state of American politics. Many assume that the polarization that exists in modern America is experienced similarly by the vast majority of Americans. But this is not the case. Many Americans do dislike the political elites of both parties, but they do not necessarily direct this anger at ordinary voters. At times, these people may even perceive partisanship as unimportant and politics to be increasingly counterproductive. The loud, angry partisans who have come to define this modern political era of hyper-partisanship for so many, we show in this book, are largely concentrated in the group of the deeply involved.

In this way, the “other” divide is fundamental. On one side is a minority of Americans who are deeply involved in politics. On the other side is the majority of Americans who have much less investment in day-to-day political outcomes. These two groups have different social networks, different policy preferences, different ideas about family life and child-rearing, and, of course, different beliefs about political parties. The deeply involved minority does genuinely dislike rank-and-file members of the other party; this group may even wish ill on out-partisans. For many less involved Americans, political divisions are more complicated: They do not love the opposing party but are more likely to direct the bulk of this animosity at elites and party activists.

People who are deeply involved in politics are also more likely to express their opinions: They discuss politics with others, and they are more likely to raise their voices via social media. In turn, journalists have become drawn to exemplars of angry partisans, which means the information people get about American politics has become flooded with news about political hatred and partisan contempt. Though they form a minority, the amplified voices of the deeply involved are perceived as the voices of most – if not all – Americans. America appears profoundly divided by politics because when people visualize politics, the “pictures in our heads” (Lippmann 1922, 1) are of the deeply involved – and the deeply involved are, in fact, profoundly divided by politics.

1.1 America, Divided by Politics

In 2018, The New York Times ran a survey of 2,204 Americans. The main question in the survey – borrowed from the long-running General Social Survey (GSS) – began as follows:

We are all part of different groups. Some are more important than others when we think of ourselves.

In general, which in the following list are first, second and third most important to you in describing who you are?

(Badger and Bui 2018)

What followed was a list of possible identities that included such things as “my occupation,” “my race or ethnic background,” “my religion,” “my role in the family,” and “my political party or movement.” The Times was especially interested in that last category – politics. They were conducting this survey in what they described as an “era of acrid partisanship” and wanted to compare the results of their survey to the 2004 GSS result. In 2004, only 4 percent selected “my political party or movement” as one of their top three most important descriptors. “We suspected those numbers might be higher today,” wrote New York Times reporters, Emily Badger and Quoctrung Bui (2018).

The results seemed to surprise the reporters. In 2018, in the heat of a midterm election, 16 percent of survey respondents ranked politics in their top three most important identities. Of the ten possible identities given to people, politics came in second to last, followed only by social class. Only 3% of people ranked politics as their most important identity, compared to 39% who ranked family status first and 16% who placed religion first. Certainly, these patterns showed a considerable increase from 2004, but the importance of politics did not seem to increase “to a huge degree,” Badger and Bui (2018) wrote.

Using a slightly different question, the New York Times survey also asked respondents to rate the importance of these different identities. Now, the respondents did not have to select just three identities from the set; they could, in theory, report that all ten identities were equally very important to them. Again, however, politics came in next to last: Just over 20 percent of the respondents reported that their political and partisan identities were very important to them, compared to more than 50 percent who reported that their family identities were very important.

The New York Times survey is not an anomalous result. In a different survey, Druckman and Levendusky (2019) asked a different sample of Americans to engage in a similar task: rating the importance of six different identities on a scale of 1 to 5. Looking at the average ratings, Druckman and Levendusky (2019) found that partisanship tied for last. Political identities, they wrote, were rated as “significantly less important than all other identities apart from [social] class” (Druckman and Levendusky 2019, SI10).

Elsewhere, Karpowitz and Pope (2020) posed a similar question as part of the American Family Survey (AFS). Fielding their survey during a highly contentious presidential election, Karpowitz and Pope also asked respondents to rate the importance of a set of identities. Again, politics came in last – though 34 percent of people did report that their political party was either very or extremely important to them (Karpowitz and Pope 2020, 14). This is, notably, higher than the percentage who viewed politics and partisanship as very important in the New York Times survey. That being said, other comparable identities are also rated as more important in the AFS than the New York Times survey. In the AFS, 44 percent of respondents said that their religious identities were important, for example, relative to only about 35 percent in the Times. Still, while the actual percentage of Americans for whom political identities are important is an open question (likely, one highly dependent on measurement), a unifying pattern in these results is that political identities seem much less important to people than their other characteristics.

That politics seemed so much less important to people relative to their other identities surprised the New York Times reporters (the political scientists who found similar patterns seem less surprised). Indeed, much of the article about these results – “Americans Say Their Politics Don’t Define Them. But It’s Complicated” – offers possible explanations about why the data patterns are actually hiding just how important politics is to the American public. Identities are inherently contextual, and perhaps, the article posits, more people would have reported that their political identities were important to them had the survey begun with a political prime. Or perhaps, “other identities on this list – religion, race, gender, even occupation – have increasingly become intertwined with politics.” People do not need to “explicitly prioritize their politics,” Badger and Bui (2018) wrote, because “these other identities now offer a clearer window into their politics.”

Badger and Bui (2018) are, without a doubt, correct. Indeed, they should be – their reporting on this topic relies not only on the survey but also on interviews with five different political scientists studying American political partisanship. Political parties have become better sorted (Levendusky 2009), and the result is a clearer division of the American public (Fiorina 2016). People are increasingly receiving social and political cues about the way others who are like them are supposed to behave in various political contexts (Barber and Pope 2019; Connors 2020; Druckman et al. 2021b). People are bringing politics to, ostensibly, nonpolitical contexts more than they have in the past (Iyengar et al. 2019). Politics is obviously divisive.

But there are two ways to consider the divisions that politics creates, and both are present in the New York Times article. Badger and Bui choose to focus on the one that they believe lurks beneath the surface of their survey: America is so divided that partisan divisions are inherent even in people’s nonpartisan characteristics. Yet, the data also suggest the possibility of another political divide: There is a minority of people for whom politics is of clear, explicit importance.

Even if politics is inextricably linked to our other identities, there is likely a difference between people who select politics and partisanship when asked to pick just three most important identities and those who do not.

Spry’s (2018) multidimensional approach to identities offers a useful way to think through this distinction. There is a difference, Spry argues, between belonging to a group (what she terms “membership”), identifying with a group (“identity”), and believing that what happens to other members of the group also affects you (“consciousness”). In Spry’s framework, many people are group members, but only some people are what she calls “strong identifiers” – people for whom a personal identity is heavily connected with a particular group membership. What makes someone a strong identifier, Spry argues, is that “the self and the group are inextricably tied” (Spry 2018, 60). Extrapolating this idea to the New York Times survey, what Spry’s theory first suggests is that we cannot conflate the idea of having a partisan team with the importance of that team for one’s sense of self. Second, however, Spry’s argument underscores the importance of self-categorization: There is something unique about a group of people who, when given a set of other identities, chose politics.

The New York Times acknowledges that the data suggest that “most Americans don’t live and breathe politics the way Washington news fiends do (or, to be honest, the way we do).” Yet, The New York Times misses an important nuance. When asked to describe themselves, there are relatively few people for whom politics is primary. The authors of the article assume that this outcome is somehow a function of their measure being imprecise and failing to capture the fundamental place political and partisan divides hold for many Americans. But, to be a cliché of the terrible anonymous reviewer every academic has encountered, we suggest the data point to a different question: If so few people believe that politics is important to them, why does America seem so divided? The answer, as we will suggest throughout this book, is that there is a critical divide between those who believe politics holds a primary place in their lives and those who do not.

1.2 Capturing the Relationship with Politics

Imagine that there are two people, whom, for the sake of this example, we will call Chip and Dale. Imagine that Chip does not want to read any news about politics, nor does he want to hear his friends discuss political campaigns. Chip may know that an election is coming but has little interest in stories about the candidates; he knows next to nothing about politics. Dale, on the other hand, checks political news on an hourly basis, and he will specifically search out information about an ongoing campaign; he feels an odd sense of anxiety when he cannot follow political news. Dale is knowledgeable about politics, but, more than that, he seeks out social interactions that focus on politics and these social interactions often take the form of being vocal – he regularly posts news stories and shares his political opinions via social media. It makes Dale frustrated and angry when he sees people posting things about politics that he finds incorrect or contrary to his own position. It also makes him frustrated and angry to know how little attention Chip pays to politics. Were Chip and Dale to be asked about interest in and attention to politics in a survey, Chip would likely select the category that reflects the least interest and attention, while Dale would likely place himself in the top category of both measures. The survey measure, then, reflects the very clear distinction in how Chip and Dale relate to politics.

Now let’s say we have a third person: Pete. Pete checks in with political news every day – though he never feels as anxious about it as Dale. Pete will discuss politics with some coworkers or friends and may even post “I voted!” via social media on Election Day. Pete feels some frustration when he sees others share opinions that he does not agree with, but he usually ignores those types of posts on social media and has never shared a post with his own political opinion. Pete believes he has enough political knowledge to feel comfortable with politics. In a survey, Pete would likely select response options that reflect that he is very interested in and pays a good deal of attention to politics – the same response options as Dale.

Pete and Dale end up in the same interest and attention categories, though their relationships with politics are markedly different. Politics is more important to Dale; he is much more involved in politics. Dale is more likely to be politically vocal. If one of these two people is going to end up in a protracted political argument, it is more likely to be Dale. When journalists turn to social media to consider the shape of political opinions on a topic (McGregor 2019), they are going to be much more likely to encounter Dale’s opinion than Pete’s. Just as Dale and Chip have different relationships with politics, so too do Dale and Pete. The difference between Chip and Dale is reflected in how they respond to survey questions about interest and attention; the difference between Pete and Dale, however, is less clear-cut.¹

Chip, Dale, and Pete reflect two types of variation in measures that categorize people’s levels of interest in and attention to politics. The first is the expected variation across the categories – for example, between Chip and Dale/Pete; the second, however, is the variation within categories – for example, between Dale and Pete. This second form of variation is certainly to be expected; there is no ordinal survey measure that can avoid within-category variation. Indeed, measures of attention and interest are likely better than many other ordinal measures in capturing relevant individual distinctions (see Prior 2019 for a discussion). Our argument is not a critique of these measures (in fact, we use these measures at various points in the book). Rather, our argument is that variation within the top categories of interest and attention hints at a meaningful but heretofore unexplored political divide between people like Pete and people like Dale.

Of course, our example is just hypothetical. As a next step, then, we turn to data from two national surveys. Our goal in the next sections is very simple: Given that there are different ways in which someone may engage with politics, can we observe variation within the response categories of interest and attention measures? In other words, do surveys offer any patterns that suggest that Petes and Dales end up in the same attention and interest categories?

1.2.1 Over-Time Patterns

If, as we suggested previously in this chapter, people are imagining an America where people are extraordinarily politically vocal, the implication is that people are also imagining an America where people are highly interested in and attentive to politics. There are glimmers of this possibility in some survey data. In a 2017 Pew survey, for example, 52 percent of Americans reported that they started paying more attention to politics after the 2016 election. Of course, paying more attention does not necessarily mean paying a high level of attention – after all, if one begins at no attention, even a slight shift is an increase. Also, an increase in attention may not necessarily reflect patterns in political interest.

When we track over-time patterns in interest using the American National Election Study (ANES) from 1956 to 2016, we do see some evidence of an increase in interest (Figure 1.1a). It is not an entirely clear increasing pattern – there is a dip in interest during the 1996 election and then an increase again in 2004 – but it is a line that generally trends upward. In 2016, about 50 percent of ANES respondents categorized themselves as “very much interested,” compared to 29.6 percent in 1956.

Figure 1.1 Changes in campaign interest and activity from 1952 to 2016.

Source: Data from the American National Election Study cumulative file

To put this increase into a broader context, we also plot other variables that may reflect greater over-time engagement in politics. In Figure 1.1b, alongside the interest measure, we also plot the percentage of ANES respondents who engaged in any campaign activities over the course of the campaign. In 1952, 23.9 percent of ANES respondents reported undertaking some campaign activity, and in 2016, 23.4 percent reported doing so. Across the entire time period, campaign activities have always lagged behind levels of interest. What is more, in the three most recent campaigns with the highest interest levels – 2008, 2012, and 2016 – campaign activity lagged about 20 percentage points behind interest.

In Figure 1.1b, then, we see that people’s attempts at influencing others have increased considerably over the time period. In 1952, 28.1 percent of people reported trying to influence someone’s vote, compared to 48.9 percent in 2016. On the other hand, the patterns in postelection conversation (a measure that is only included on the ANES starting in 1992) are less clear. Generally, few people discuss politics after an election ends, though more than 40 percent continued postelection political discussions in 2000 and 2016, likely reflecting the postelection challenges (in 2000) and a surprise outcome (in 2016).

The patterns in Figure 1.1, then, suggest that increases in interest do not always co-occur with increases in other forms of political engagement. We do not see similar shifts in campaign activities, for example, and shifts in postelection discussion behavior seem more reflective of the election context than of some intrinsic interests. In 2012, while 41.5 percent of ANES respondents reported being “very much interested,” only 11.4 percent were still talking about politics after the campaign was over.

Our goal is not to explain these over-time patterns in levels of interest and other measures of engagement.² Rather, our goal is to suggest that the divergences in Figure 1.1 hint at the possibility that the “very much interested” category includes people who vary in their relationship with politics. In the next section, we examine the possibility of variation within this top category of interest more directly.

1.2.2 Variation in Top Categories

Focusing on over-time patterns, as we did in the previous sections, limits the measures that we can track. Therefore, in this section we rely on more recent data and look more directly at variation within interest and attention categories. Our goal, again, is not to critique these measures but merely to explore the possibility that the highest interest and attention categories include different types of people. We again rely on the ANES but, in this section, also include data from the 2018 Cooperative Congressional Election Study (CCES). Before we turn to the variation, we present the distributions of the different interest and attention measures we are using (Figure 1.2 a-c). As Prior (2019) demonstrates, interest is unidimensional, suggesting that the “Interest and Following Campaigns” measure (Figure 1.2a) and the “Interest in Public Affairs” measure (Figure 1.2c) are likely capturing similar ideas. In Figure 1.2b, however, it is possible that the attention measure is capturing a different aspect of people’s approach to their political surroundings. The distributions in Figures 1.2 reflect the final data point in Figure 1.1: 50 percent of respondents, in both the ANES and the CCES, select the highest interest categories. The attention measure in Figure 1.2b looks somewhat different: Only 20.2 percent select the highest category, though 55.3 percent report that they pay attention either most of the time or all of the time.

Figure 1.2 Distributions of various measures of political interest and attention.

Source: Data from 2016 American National Election Study and the 2018 Cooperative Congressional Election Study. Error bars represent 95 percent confidence intervals

Within these categories, however, we see considerable variation in other forms of engagement with politics (Figures 1.3–1.6). It is certainly clear that people in the highest interest and attention categories are much more knowledgeable and engaged than those in the lowest categories. Yet, within the highest categories of both interest and attention are people who seem to have very different relationships with politics.

Figure 1.3 Political knowledge levels by different levels of interest and attention.

Source: Data from 2016 American National Election Study. Horizontal lines represent the mean of political knowledge for that level of interest (panel a) or attention (panel b). Random jitter is added to the scatterplot to make it easier to see the various respondents

Some people who pay a lot of attention to and have a lot of interest in politics, for example, are very knowledgeable – others are much less so (Figure 1.3). Although being more interested in politics makes one more likely to take a variety of political actions, the proportions of people who take those actions is still on the lower end (Figure 1.4). Indeed, among those who report the highest levels of interest and attention are people who are heavily engaged and spend a good deal of time talking about politics, but there are also people who appear much quieter (Figure 1.4 ). These differences in expression also translate to social media patterns (Figure 1.5). Attention to politics increases people’s political engagement on social media, but, nonetheless, among respondents who report that they pay attention to politics most of the time, about half report that they never post on social media about politics.

Figure 1.4 Campaign activities and political discussion by levels of interest and attention.

Source: Data from 2016 American National Election Study. In panels a and c, horizontal lines represent the mean number of activities for that level of interest (a) or attention (c). In panels b and d, horizontal lines represent the mean number of days a respondent discusses politics for that level of interest (b) or attention (d) Random jitter is added to the scatterplot to make it easier to see the various respondents.

Figure 1.5 Social media use by attention levels.

Source: Data from 2018 Cooperative Congressional Election Study. Error bars represent 95 percent confidence intervals

In short, the patterns in the figures suggest that it is relatively easy to distinguish between Chip, who pays no attention, and people like Dale and Pete, who are attentive to politics. Variance patterns clearly show that at the lowest levels of interest and attention are people who have less knowledge, take fewer political actions, and are much less likely to engage in any political expression. It is more difficult, however, to distinguish between Dale and Pete. Whin the top categories are people for whom, like for Dale, interest and attention co-occur not only with other political behaviors but also with high levels of political expression. On the other hand, these top categories also include people like Pete: They do pay attention to politics, but they are not necessarily going to spend a part of each day engaging with political content or expressing their political opinions.

It is, without a doubt, important to distinguish between those who have a good deal of interest and attention and those who have none. This distinction offers a means by which we can determine who is likely to be entirely disengaged from politics. But we suggest that there is another divide that is hiding within the interest and attention measures: the divide between those who, like Pete, are merely interested and attentive and those who, like Dale, are deeply focused on politics.

We want to be clear that this is not merely a question of measurement. Survey measures of interest and attention reflect real, valid, and predictive differences among the electorate. Indeed, scholars have shown that interest and attention affect a variety of political outcomes and reflect a dispositional stability (Prior 2019). In this book, however, we will argue and show that there are more gradations to people’s relationships with politics, and these gradations form an especially meaningful political divide in the highest categories of interest and attention. The dividing line in American politics may seem to be between Chip and Dale/Pete, but it is actually between Dale and the other two voters.

1.3 Beyond Political Interest and Attention

Let’s return to our three hypothetical exemplars: Chip, Dale, and Pete. Distinguishing between Chip and the other two is simple: Chip has no connection to politics at all, whereas Dale and Pete do. The more complicated part is, as we already suggested, distinguishing between Dale and Pete. Both Dale and Pete pay attention to political news, but only Dale feels anxious when he is in a context that does not allow him to follow political news (e.g., a meeting at work). Both Dale and Pete feel some frustration when they encounter people whose political opinions differ from theirs, but Dale is much more likely to voice his disagreement. Both Dale and Pete may read about politics on social media, but Dale is much more likely to post about his opinions and share political content.

Distinguishing between Dale and Pete is challenging. Both of them are clearly very interested in politics. Yet, Dale’s relationship to politics seems very different from Pete’s – politics seems much more of a focus for Dale. This idea of importance offers a starting point for differentiating these two people. What separates Dale and Pete, we will argue in this book, is their level of involvement in politics. Pete is interested in politics, but Dale is deeply involved.

Before we go any further, it is important to make clear that we do not claim any credit for introducing the term “involvement.” The idea of involvement in politics has been part of political science for decades. In fact, in the previous sections of this chapter, we quoted several scholars who used the terms involved and uninvolved to distinguish between groups of Americans (Berelson, Lazarsfeld, and McPhee, 1954; Campbell et al. 1960; Converse 1962). In this early research on political behavior, involvement included a broad set of political predispositions – one of which was, notably, interest in politics (Campbell et al. 1960). The authors of The American Voter, however, point out that involvement was an idea beyond political interest. Rather, they suggested that it was a psychological predisposition, an idea that reflected an individual’s “commitment” to politics (Campbell et al. 1960, 104). The authors of The American Voter also foreshadowed one of the results in this book: “[A] really intense commitment to politics is probably limited in American society to a small fraction” (104).

Although this book does not introduce the term “involvement” to political science, in Chapter 3 we do draw on research from political science, psychology, sociology, and marketing – all disciplines that have worked to capture what it means to be involved in and committed to something – to outline the psychology of political involvement. Our particular interest is in the psychology of the “deeply involved” (a term that has also been previously used in political science by Berelson, Lazarsfeld, and McPhee (1954)) or people who have an especially intense focus on politics as we explain in Chapter 3.

The deeply involved spend a good deal of time following politics. More than merely checking in with the news every morning or evening, those who are deeply involved pay attention to political news consistently throughout the day. The deeply involved feel anxious when they cannot follow politics. They can see the political consequence of even seemingly mundane, small political events. They are also likely to express and discuss their political opinions – often publicly.

Of course, political interest and attention are linked to involvement. Deeply involved people are certainly interested in and attentive to politics. Indeed, it would be difficult to imagine someone being deeply involved but completely uninterested in a topic (see e.g., Mittal 1995). Deep involvement, however, is a characteristic that describes a relationship with politics that is beyond that which is captured with higher levels of political interest. This is not a critique of either research on or measurement of political interest or attention. Rather, it is to underscore that deep involvement is a psychological characteristic that is different from interest; some people are very interested in politics, but others are deeply involved. It is the difference between those who are deeply involved and everyone else that forms the “other divide” in American politics.

We draw on research on the sociology and psychology of deep involvement to argue that people who are deeply involved have strong perceptions about the appropriate and inappropriate ways of being involved. Being engaged – even deeply so – in a topic is one thing, but it is important to engage with the topic in ways that are correct. Translating this idea to politics suggests the possibility of a relationship between deep involvement and people’s political positions: For those who are deeply involved, it is equally important that people who are engaged belong to a particular party or even more specifically support a particular set of positions within that party. The implication, then, is that deep involvement in politics is likely to co-occur with animosity toward people who are politically different and greater affinity for your own political side – that is, higher levels of affective polarization.

Along with this stronger psychological commitment to politics comes a desire for social interaction about politics. The deeply involved are more likely to begin political conversations and engage in political conversations. This tendency is likely to be most visible via social media, as a key affordance of these platforms is broad, public expression of a person’s views or thoughts. Social media platforms, then, allow the deeply involved an opportunity to share information, share their opinions, and correct people whose political positions or ideas they view as incorrect.

It is this intersection between deep involvement, animosity, and expression that not only forms the foundation of the “other divide” but also reinforces the image of the American public as one that is deeply divided. The people who are most engaged in politics – the deeply involved – know the most about day-to-day politics. They are also most likely to feel strongly about what is politically right and politically wrong, where “right” and “wrong” are political outcomes, rather than procedures and rules. Indeed, it is possible that a person who is deeply involved so strongly believes in the potential benefit of a given political outcome that they are willing to support processes that break with political rules and norms. Because the deeply involved are drawn to social interactions about politics, they are more publicly expressive, and their political views and positions are more likely to be amplified – first by their own voices and then by journalists and pundits as “exemplars” of ordinary voters (see e.g., Levendusky and Malhotra 2016). To restate our previous argument: America appears heavily divided along partisan lines because partisanship is, likely, the main division among the deeply involved. What we suggest in this book, however, is that once we broaden our scope to the entire electorate, the emerging division is often between those who are and those who are not deeply involved in politics.

1.4 Plan of the Book

In what follows, our goal is to consider the various ways in which deep involvement in politics divides people. Before we can delve into this “other divide” in American politics, however, it is important to consider the political divide that has gotten the most attention in both contemporary politics and political science research: the partisan divide. After all, there is no “other” divide if there is no “main” divide. Therefore, in Chapter 2, we begin not with deep involvement but with affective polarization. In this chapter, we first track the evolution of affective polarization in political science research, the translation of this concept to media coverage, and its emergence as a critical political divide in America. Our goal in this chapter is to draw out the nuance inherent in political science research on affective polarization that is often missing from its translation to media coverage. Weaving together published research with new experiments, we suggest that research points to a complicated, conditional portrait of affective polarization in America. Some people are clearly polarized, but there are limitations to the power of overt polarization cues in politics (see e.g., Costa 2021). Levels of affective polarization have increased in the American electorate over time, but just how polarized America is also depends on how one measures people’s feelings about the other party (Druckman et al. 2021b; Klar, Krupnikov, and Ryan 2018; Druckman and Levendusky 2019).

What emerges from Chapter 2 is that, without a doubt, there are people in America who harbor an unconditional animosity for the opposing party; there are likely more of these types of people at the time of this writing than there were two decades ago (Iyengar et al. 2019). For many people, however, most of their political animosity is directed toward partisan elites (Druckman and Levendusky 2019; Kingzette 2020) and members of the opposing party who are very vocal about their political positions (Druckman et al. 2021b). Some people seem to dislike the political opposition, but many other people seem to also dislike politics. Jointly, these two ideas set the stage for the remainder of the book. What if a dividing point in American politics is broader than just people’s attachment to a party but is actually their attachment to politics itself?

We begin exploring this “other divide” in Chapter 3 through a theory of political involvement. Although we have already introduced the idea of “deep involvement” previously in this chapter, we define involvement with greater precision in Chapter 3. Relying on an interdisciplinary approach, we outline the characteristics of deep involvement: (1) spending a tremendous amount of time on a topic, (2) interpreting even mundane events as being highly important, and (3) seeking out social interactions related to politics. Although we discuss the relationships between these characteristics, we do not make a causal argument about their emergence – there is evidence suggesting that all three emerge simultaneously (Thorne and Bruner 2006), but there is also evidence suggesting that seeking out social interactions follows the emergence of the first two characteristics (Kozinets 2001). The key idea is that when someone is deeply involved, all three are present.

In the same chapter, we also discuss the implications of deep involvement for political behavior and political outcomes. We consider what deep involvement means for news habits and political discourse, and we also explore why people who are deeply involved in politics may be especially likely to harbor animosity toward the opposing party. Then, we turn to those who are not deeply involved in politics. Here, we use the psychology of self-categorization to consider what the amplified voices of the deeply involved mean for the way people perceive their own relationships with politics.

The deeply involved, we theorize in Chapter 3, are different from those who have lower levels of involvement. The four chapters that follow are devoted to analyzing just how different this group of people is from the majority of Americans. Specifically, we analyze a variety of political outcomes to track involvement divides in politics – differences between those who have a strong focus on politics and those who do not. These analyses, however, depend on a measure that allows us to determine gradations of political involvement, distinguishing those who are deeply involved from everyone else. Therefore, we begin Chapter 4 with a measure of involvement, developed based on the theoretic arguments in Chapter 3. In Chapter 4, we walk through the development and validation of our main measure, as well as the distribution of the measure across different types of samples. As we will stress in Chapter 4 (and throughout the book), the measure we develop does not include any items that are associated with measures of partisan strength, partisan identity, or affective polarization. Rather, it is a measure that is focused on behaviors toward politics rather than partisanship.

After the introduction of this measure, Chapter 4 focuses on a series of descriptive, exploratory analyses that track the relationship between involvement and traditional correlates of political behavior. First, we consider whether there are certain demographic characteristics that are more heavily associated with higher levels of involvement. Second, we turn to the relationship between involvement and affective polarization. Next, we analyze whether people who are deeply involved differ in which political issues they find most important. We conclude the chapter with a study that investigates whether there are involvement gaps in one of the most critical determinants of political behavior: political efficacy. Across these analyses, we find evidence that involvement divides the American public. Higher levels of involvement are, as anyone who has had a passing introduction to research in American politics may already suspect, associated with stronger partisanship, more ideological extremity, and greater affective polarization. In particular, while many people show animosity toward the opposing side, those who are deeply involved are unique in their affinity for their own side. Moreover, involvement also divides co-partisans – people who are deeply involved prioritize different political issues than those who are less involved. Finally, we find that those who are deeply involved have a much greater sense of political efficacy across a variety of indicators. People who are deeply involved are more likely to believe that they know what is best for American politics.

Building on the patterns in Chapter 4, in Chapter 5 we turn to a more challenging question: What might lead people to become deeply involved in politics? We begin Chapter 5 with a caveat, and it is a caveat we will note here as well – tracking the development of political predispositions is a difficult task. We cannot, for example, randomly assign life experiences and track their causal outcomes. Still, we rely on a foundation of research in political science and psychology to focus on one factor that scholars have long suggested is pivotal to people’s political development – their socialization. Therefore, in Chapter 5, we consider how a variety of social experiences – family interactions, educational contexts, and friendship networks – relate to involvement. We find deep involvement is associated with more early-in-life political socialization; it is also more heavily associated with a very specific college experience. Turning to people’s social networks, we find “involvement bubbles”: People tend to associate with others who are equally as (un)involved as they are.

Jointly, Chapters 4 and 5 document involvement divides across a variety of fundamental political factors. People who are deeply involved have different relationships to partisanship and are more likely to be affectively polarized. Moreover, involvement divides people’s feelings about issues within parties and bisects efficacy. We also see involvement gaps in people’s networks and perceptions of social distance to others. People who are deeply involved, these chapters hint, are living different lives than those who are less politically involved. Chapter 6, then, turns to a more unusual context: parenting.

We consider the relationship between involvement and parenting because people’s views on parenting are revealed preferences of their most fundamental values – people’s thoughts about how children should spend their time speak to their beliefs about what is most important in a society (Darling and Steinberg 1993). From this perspective, parenting is an ideal test of people’s beliefs about the role politics should play in American society. Focusing on overt political socialization, Chapter 6 relies on a series of experimental studies that manipulate parenting decisions. We find that, generally, people prefer parents who avoid politics – the participants in our studies seem uncomfortable with an explicit political socialization. The group that seems most comfortable with political parenting, however, is the deeply involved. Indeed, it is the deeply involved who are most likely to report that they would make the same decisions as the parents who introduce politics to their young children’s lives. If people’s beliefs about raising children are a form of revealed value preferences, then people who are deeply involved value politics much more than those who have lower levels of involvement.

The deeply involved are different. They are more confident in their political positions (Chapter 4); they have different social networks (Chapter 5); and they place a greater value on politics (Chapter 6). These differences become more important if they are exacerbated by unequal political voices. In other words, it is less important that involvement affects which issues co-partisans find important if people at all levels of involvement are equally likely to be heard by policy-makers. This, however, may not be the case. Indeed, in Chapter 3, we theorized that seeking out social interactions is a fundamental component of deep involvement; this need for social interaction, we argued, translates to greater expressiveness. This greater tendency toward expressiveness is something that, to this point in the book, we have not tested directly; we focus on expression in Chapter 7.

To address questions of political voice and expression, in Chapter 7, we modify our measure of involvement to ensure that there are no indicators within the scale that could, on their own, measure expression. Using this modified involvement measure, we rely on a series of tests to track whether people who are more involved in politics are more likely to discuss politics with others and more likely to post their positions on social media. Moving beyond relationships between involvement and expression, we next investigate what this tendency of the deeply involved to express their political positions means for people who are less involved. We find a notable tension. The deeply involved express their political positions because they believe that they are informing others about important political outcomes and events. Those who are less involved, however, see this form of political expression less charitably: They do not believe that political expression on social media is designed to inform but rather that it is designed to somehow bolster the status of the person doing the posting. Moreover, those who are less involved perceive people who post about politics on social media as more extreme but no more knowledgeable about politics. There are involvement gaps not only in people’s willingness to express their political opinions on social media but also in how people perceive the political opinions they see shared.

Chapter 7 suggests a consequential outcome: The voices people are most likely to hear discussing politics are those of the deeply involved. These voices are not always perceived in the most positive manner by those who have lower levels of involvement. Chapter 8 tracks the broader implications of the involvement gaps documented in the previous chapters by exploring the possibility that the voices of the deeply involved are especially likely to be amplified by the media. Unlike previous chapters, which focus on the American public, Chapter 8 turns to a particular group of elites: political journalists. Weaving together research on journalistic practice, surveys of journalists, a content analysis, and qualitative interviews with journalists, we argue that journalists end up amplifying the deeply involved because they often focus on covering political polarization among the public. Because journalists often seek to create narratives that will be accessible and interesting to their readers, these stories are also likely to include exemplars of polarization – regular people who personify political divisions. Exacerbating these patterns is an increasing reliance on social media among journalists (see e.g., McGregor 2019) as social media over-represents the voices of the deeply involved.

We note in this introductory chapter and in Chapters 2 and 3 that the deeply involved have come to exemplify what it means to engage in politics and to be a partisan. We return to this idea in the final section of Chapter 8 through a discussion of co-authored research demonstrating that people overestimate both the extremity and the expressiveness of other Americans. In other words, people assume that most people are like the people amplified by media coverage of partisanship and polarization: strongly partisan and highly vocal about their political positions. Notably, people’s perceptions of partisans differ starkly from the actual modal partisan – a moderate who rarely discusses politics. That modal partisan, however, is probably unlikely to make the news or appear in someone’s social media feed sharing a (moderate) political opinion (Bode 2016).

We conclude in Chapter 9 by turning the spotlight on people who are not deeply involved. Not being involved, we argue, should not be confused with being apathetic and not caring about political outcomes – indeed, most of the people who voted in the 2020 election were likely not deeply involved. One should not assume deep involvement is synonymous with caring; rather, it reflects a very particular engagement with politics. We also return to the quote from Berelson, Lazarsfled, and McPhee (1954) that opened this chapter to consider what it would mean to have a public that is fully deeply involved and why the political uncertainty of the less involved may sometimes be valuable. We conclude, however, where we began: the relationship between the partisan divide and the “other” divide that is at the center of this book. Understanding growing affective polarization, we suggest, means seeing politics through the lens of involvement differences.

1.5 A Word about the Data and Statistical Analyses

Before we begin, we want to set the stage for the data analyses that follow in the six of the remaining eight chapters. The data used in this book come in three forms. The bulk of the evidence comes from eight original studies (N = 8,026) conducted in 2019 and early 2020. In addition to these original studies, we also rely (as we did in this chapter) on archival data from studies conducted by other researchers (total N = 115,855). In addition, we also use data from four studies (N = 10,105) that we collected along with co-authors for previous projects.

Of the original studies included in this book, we want to highlight three surveys conducted on national samples, since they serve as the main studies for this project. The first survey (N = 1,564) was conducted in two waves by the survey company Dynata in April 2019. Throughout the text, we will refer to any study that was part of this large survey as being part of study “D19.” The second survey (N = 1,586) was conducted in two waves by the survey company Qualtrics in the first two months of 2020. Throughout the text, we will refer to any study included as part of this survey as being part of study “Q20.” The third survey (N = 1,500) was conducted in a single wave by YouGov in July 2020. Throughout the text, we will refer to any study included as part of this survey as being part of study “YG20.”

In an appendix that follows at the end of the book, we have outlined which studies are part of which surveys. In that appendix, we also include information about the demographics of the respondents in these three main surveys, which includes information about how the respondents measure basic political variables (e.g., their partisanship). For greater detail on the studies, we direct you to the website we have created for this book: www.otherdividebook.com. The website has the full question wording and response options for each of the three main surveys.

The materials on the website also include the full results for the various models and statistical tests performed in this book. Throughout the book, we will primarily present the results from regression models in figures – typically, plotting the marginal effects of key variables or the predicted values of the dependent variables based on a particular set of values for the independent variables in the model. For this reason, we often do not discuss the coefficients for control variables. Therefore, we present full coefficient tables in the online appendix, organized by chapter and result.

We also want to note that many of the results in this book are descriptive; indeed, in many cases, we deliberately refer to outcomes as co-occurrences rather than using language that implies causality. Although the research we rely on to develop our theory makes a very clear causal link between involvement and animosity toward other people – we deliberately do not do so. Demonstrating causal relationships would mean manipulating levels of involvement. However, if our theoretic arguments are correct, and deep involvement is a type of intense focus on a topic that develops over time, then by definition, it would be likely impossible to manipulate involvement in an experiment. There are experiments that help demonstrate causality, but in those cases, involvement is moderating a treatment effect and is not the cause of the effect. Still, we believe that the results we demonstrate here do speak to the broader point of this book: There is a clear political divide between people who are deeply involved and those who are not.