America’s Polarized Public: A Reply to Fiorina

In his essay, “America’s Missing Moderates: Hiding in Plain Sight”, Morris Fiorina reprises many of the claims that he and his co-authors have made in books such as Culture War? The Myth of a Polarized America, and Disconnect: The Breakdown of Representation in American Politics. The central argument of the essay as well as the earlier books is that polarization in the United States is limited to a “political class” made up of elected officials and a very small group of activists. It is this polarized political class, according to Fiorina, that is responsible for gridlock in Washington and for the failure to address what he views as the main threats to America’s future: debt and “the specter of rampant inflation.”

According to Fiorina, ordinary Americans today are no more polarized than they were thirty or forty years ago, although they are better “sorted.” That is, there is a stronger relationship now between party affiliation and ideology than there was then. But, according to Fiorina, even this relationship “weakens rapidly as one moves beyond the political class to the larger electorate.”

Not only are ordinary Americans no more polarized than in the past, according to Fiorina, they are turned off by the growing polarization of the political class and therefore less likely to support either party or even to vote. Yet, he says, leaders of both parties continue to ignore the views of the moderate majority and to cater to the extremists who dominate party primaries and caucuses. Meanwhile, those who still vote are limited to the polarized choices presented by the two parties even though most of them would prefer more moderate alternatives.

This elite theory of polarization has found considerable acceptance among pundits and political commentators in recent years. Its popularity may stem from the fact that it largely absolves the public of any responsibility for the polarized state of our nation’s politics. The American people, according to this theory, are innocent victims of the machinations of the polarized political elite. Fiorina has even compared them to civilians caught in the crossfire of a guerilla war.

Despite its popularity in some circles, however, the elite theory of polarization is deeply flawed. Political elites have certainly contributed to the rise of polarization over the past several decades. But one cannot understand the rise of polarization without taking into account important changes in American society such as increasing racial and ethnic diversity, the decline of the traditional family and growing economic inequality. Along with actions of political leaders, these societal trends have resulted in dramatic changes in the makeup of the electoral coalitions supporting the Democratic and Republican parties—changes that are driving the parties apart.

Perhaps the most important change affecting the American party system over the past several decades has been the growing racial and ethnic diversity of the American electorate. Between 1992 and 2012, the nonwhite share of the presidential electorate more than doubled, from 13 percent to 28 percent. This has resulted in a growing racial divide between the Democratic and Republican electoral coalitions. In 2012, according to data from the national exit poll, both Barack Obama and Democratic candidates for Congress received 45 percent of their votes from nonwhites while Mitt Romney and Republican candidates for Congress received only 11 percent of their votes from nonwhites.

The already deep racial divide between the parties is almost certain to increase in future elections as the nonwhite share of the electorate continues to grow. And this growing racial divide is contributing to a growing ideological divide between supporters of the two parties. Nonwhite voters hold much more liberal views on the role and size of government than white voters, and the growing dependence of the Democratic Party on nonwhite voters has contributed to the flight of racially and economically conservative white voters to the GOP. The growing importance of race and racial attitudes in shaping partisanship and voting behavior has been amply documented by scholars such as David Sears, Michael Tesler and Donald Kinder.[1]

Other changes in American society—including the decline of the traditional family, the growing economic independence of women, and the rise of the women’s and gay rights movements—have also contributed to polarization within the electorate. New issues such as abortion and gay marriage have emerged that divide Americans based on their religious and moral beliefs, and these divisions increasingly coincide with party affiliation, especially among white voters. As a result, in recent elections, religious beliefs and practices have come to clearly differentiate supporters of the two parties. White evangelical Christians now make up one of the largest and most loyal components of the Republican electoral coalition while secular voters overwhelmingly support Democrats.

The result of all of these trends has been the emergence of an electorate that is not only better sorted but also more polarized. In fact, sorting and polarization are best understood as two sides of the same phenomenon.

Over the past forty years, there has been a substantial increase in the relationship between party identification and ideological identification and in the relationship between party identification and positions on a wide range of specific policy issues. These trends are evident among the large majority of the American public, not just a small minority of activists.

Fiorina’s claim that the impact of party sorting weakens rapidly when one looks beyond the political class is simply not supported by the evidence. Between 1972, the first time a question about ideology was included in the American National Election Study, and 2008, the most recent election for which data are available, the correlation between the party identification scale and the ideology scale increased from .36 to .66 among all voters. This means that in terms of shared variance, the relationship between party identification and ideology was more than three times stronger in 2008 than in 1972. As a result, the distance between the average Democratic voter and the average Republican voter on the 7-point ideology scale more than doubled, from .9 units in 1972 to 2.2 units in 2008 with the average Republican voters going from a mean score of 4.7 in 1972 to a mean score of 5.4 in 2008 and the average Democratic voter going from a mean score of 3.7 in 1972 to a mean score of 3.2 in 2008. In other words, over these 36 years, Democratic voters moved from slightly left of center to well to the left of center while Republican voters, who were already well to the right of center, moved even further to the right of center.

But this was not the only significant change that occurred during these years. As Democratic and Republican voters were becoming better sorted, the overall distribution of voters on the ideology scale was also becoming more polarized. Between 1972 and 2008, the percentage of voters placing themselves in the center of the scale fell from 35 percent to 27 percent while the percentage placing themselves within one unit of the left or right pole of the scale increased from 29 percent to 46 percent. In other words, there were fewer moderates and far more voters with strong ideological views in 2008 than in 1972.

The relationship between ideology and policy preferences has also become considerably stronger over time and is now, in the case of most policy issues, quite strong. In the case of abortion, for example, the correlation between ideology and position on a four-point abortion policy scale has increased from .21 in 1980, when this abortion question was first included in the ANES survey, to .48 in 2008. In terms of shared variance, this means that the relationship was more than five times stronger in 2008 than in 1980.

Fiorina’s claim that a majority of self-identified conservatives have liberal views on issues such as abortion and gun control is not supported by the evidence from the 2008 American National Election Study. In this survey, only 23 percent of voters who identified themselves as conservative took the pro-choice position on abortion compared with 70 percent of voters who identified themselves as liberal and only 32 percent of voters who identified themselves as conservative favored stricter gun-control laws compared with 60 percent of voters who identified themselves as liberal.

Fiorina notes, correctly, that the distribution of opinion on some individual policy issues has not changed very much over the past twenty or thirty years. However, what has changed is the consistency of opinions across issues and the relationship between these issue positions and party affiliation. This is a crucial development because consistency of opinions on policy issues is the hallmark of ideological thinking, as Philip Converse observed in his seminal study “The Nature of Belief Systems in Mass Publics” in 1964.[2] By this standard, ideological thinking is far more prevalent in the American public today than it was during the era studied by Converse.

Fiorina is correct in asserting that politically active Americans are more polarized than those who are less active. In fact, political interest and knowledge as well as political activity are positively related to polarization. But it is not only those at the very upper end of the political engagement scale who are polarized. Even those who are only moderately attentive, informed and active are deeply divided by party.[3] And on some issues, such as the war in Iraq in 2004 or health care reform in 2008, the large majority of Americans are deeply divided.

Figure 1 shows the distribution of opinion of Democratic and Republican voters, including voters who leaned toward a party, on one of the major issues in the 2008 election: health care reform. The large majority of Democratic voters in 2008 strongly supported a proposal to have the Federal government pay for all of the costs of health care for Americans, in effect creating a single-payer health care system for the nation. The large majority of Republican voters strongly opposed this proposal. Very few voters took a centrist position on this issue.

Given such results, it is hardly surprising that the debate over health care reform during 2009 and 2010 proved deeply divisive. Democrats and Republicans in Congress were accurately reflecting the deep divisions that existed and continue to exist among ordinary Americans on this issue. Even after the Affordable Care Act was signed into law in 2010, numerous polls showed that the vast majority of Republicans wanted the law repealed while the vast majority of Democrats wanted the law preserved or expanded. In fact, many Democrats who disapproved of the ACA did so because they felt that it didn’t go far enough in expanding access to health care.

Over the past several decades, as Republican elites have moved to the right, so have Republican voters and as Democratic elites have moved somewhat less sharply to the left, so have Democratic voters. There is no evidence to support Fiorina’s claim that the large majority of Americans are turned off by the growing divide between the parties. Democratic voters, on average, place themselves very close to the position of the Democratic Party while Republican voters, on average, place themselves very close to the position of the Republican Party. However, both sets of partisans place themselves very far from the opposing party.[4]

As the parties have grown more polarized in recent years, party ties within the electorate have actually grown stronger. The vast majority of American voters either identifies with or leans toward one of the two major parties. In 2008, for example, according the American National Election Study data, pure independents made up only seven percent of the electorate. And despite Fiorina’s statement to the contrary, there is a substantial body of evidence showing that independent leaners think and act much more like partisans than like true independents.[5] In 2008, for example, as in most recent elections, almost 90 percent of independent leaners voted for the presidential candidate of the party they leaned toward and their opinions on major issues such as health care reform were very similar to those of voters who identified with the same party.

Party loyalty and straight-ticket voting have been increasing since the 1970s and reached record levels in 2012.[6] According to the 2012 national exit poll, 92 percent of Democrats voted for Barack Obama and 93 percent of Republicans voted for Mitt Romney. This was the highest level of party loyalty since the beginning of exit polling in 1972. Moreover, only six percent of Obama and Romney voters split their tickets by choosing a House candidate from the opposite party, a record low.

The powerful impact of partisanship was also evident in the remarkable consistency between the results of the 2008 and 2012 presidential elections at the state level. Only two states, Indiana and North Carolina, switched sides, and across all 50 states and the District of Columbia, the correlation between Obama’s share of the vote in 2008 and his share of the vote in 2012 was a near perfect .98. And despite the closeness of the national popular vote, only four states were decided by a margin of less than five points while 27 were decided by a margin of at least 15 points included several of the most populous states in the nation: Barack Obama carried New York by 27 points, California by 23 points and Illinois by 17 points while Mitt Romney carried Texas by 16 points.

In comparison with the 1950s, 60s and 70s, the United States today is much more deeply divided along geographic lines. There were far more deep red and deep blue states in 2008 and 2012 than in earlier closely contested presidential elections such as 1960 and 1976. And the same trend is evident when it comes to other geographic units including congressional districts and counties. This trend is clearly not a result of partisan gerrymandering, as some scholars and pundits have claimed. State and county boundaries don’t change after every Census the way congressional district boundaries do. The growing geographic divide is a result of deeper forces in American society, including immigration, migration patterns, and ideological realignment as relatively conservative areas like the South trend Republican and relatively liberal areas like the Northeast trend Democratic.[7]

Polarization is directly responsible for both the growing partisan divide among voters and the growing geographic divide in the nation. The growing consistency of party affiliation and ideology means that these two sets of attitudes increasingly reinforce one another, pushing voters in the same direction. Just as there are almost no liberal Republicans or conservative Democrats left in Congress, there are very few left in the electorate.

Party elites are much more polarized today than they were forty or fifty years ago, but there is no evidence that the growing divide between the parties in Washington and in many of our state capitols is turning off large numbers of voters. In fact, voter turnout reached its highest level in decades in 2004 and rose even further in 2008. And while turnout apparently fell slightly in 2012, it was still substantially higher than the average turnout in presidential elections since 18–20 year olds were added to the electorate in 1972.

The increase in voter turnout in recent elections occurred even in states that were not closely contested and where there was little campaigning, so the explanation clearly has little or nothing to do with increased voter mobilization by the parties. Rather, it appears that when voters perceive important differences between the presidential candidates, they are more likely to care which one wins and therefore to vote. Indeed, according to the American National Election Study data, the percentage of Americans who cared a great deal about the outcome of the presidential election was higher in 2004 and 2008 than in any previous presidential election since the beginning of the ANES surveys in 1952. And Americans were not only voting in large numbers, they were also engaging in other election-related activities in record numbers—2004 and 2008 saw the largest percentages of Americans trying to influence the vote of a relative, friend or neighbor in the history of the ANES surveys.

While there is no evidence that polarization has resulted in lower public interest or turnout in elections, it is clear that Americans have become increasingly frustrated with the policy-making process in Washington in recent years. But that frustration is itself a direct result of the deep partisan divide within the American public. Democrats were very unhappy with the policy-making process in Washington during most of the George W. Bush years because they opposed almost all of the Bush Administration’s policies. Likewise, Republicans were very unhappy with the policy-making process in Washington during the first two years of Obama’s presidency because they opposed almost all of the Obama Administration’s policies. Since the 2010 midterm elections, however, supporters of both parties have felt unhappy and frustrated because divided government has meant that neither side has been able to advance its policy agenda.

The problem that we face as a nation today is not polarization per se. It is polarization combined with divided government and the abuse of the filibuster by the minority party in the Senate. In 2012, Americans voted for a Democratic House of Representatives, a Democratic Senate and a Democratic president. They didn’t get what they voted for, however, because the concentration of Democratic voters in urban districts along with gerrymandering by Republican-controlled state legislatures enabled Republicans to hold onto a majority of House districts despite losing the national popular vote. And Senate Republicans have continued to use the filibuster to an unprecedented degree to block legislation supported by Democrats and even presidential appointments.

As Thomas Mann and Norman Ornstein have persuasively argued, the major responsibility for partisan gridlock in Washington today rests with a Republican Party increasingly dominated by a strongly conservative base and billionaire funders.[8] In resisting any compromise with congressional Democrats or the White House over issues ranging from taxes and spending to the minimum wage and gun control, congressional Republicans are accurately reflecting the views of their base and wealthy donors, but they are not reflecting the views of a majority of Americans or the results of the 2012 election. That is why the Federal government is failing to address the real challenges facing the nation in 2013—not debt, which is falling, or inflation, which is non-existent, but economic stagnation, unemployment and rising economic inequality.

Morris P. Fiorina replies:

Sorting Doesn’t Mean Polarization

To the best of my recollection this is the sixth exchange I have had with Professor Abramowitz on this topic since 2006.[1] In my essay, I tried to make clear the distinction between polarization and party sorting. Polarization is the erosion of the center as citizens move to the extremes. That has not happened in the United States; the center is still there. Rather, the parties have sorted, so that compared to 1976, say, the Democrats are a more homogeneously liberal party and the Republicans a more homogeneously conservative party. This is a commonplace observation that is fully supported by detailed research, including that of Abramowitz. But as the figure in my essay shows, the subcategories of Democrats/Independents/Republicans and liberals/moderates/conservatives are pretty much unchanged from a generation ago.

Abramowitz throws out a potpourri of findings and assertions he regards as evidence of polarization. But on inspection, his claims are about sorting, not polarization. He writes, “Over the past forty years, there has been a substantial increase in the relationship between party identification and ideological identification and in the relationship between party identification and positions on a wide range of specific policy issues.” Exactly—that’s basically the definition of party sorting. And after noting social trends that Samuel Abrams and I discuss in our 2009 book, Abramowitz writes, “Along with actions of political leaders, these societal trends have resulted in dramatic changes in the makeup of the electoral coalitions supporting the Democratic and Republican parties—changes that are driving the parties apart.” Exactly—that’s party sorting. All the differences he cites between Democrats and Republicans are real, but they do not show a decline in the middle and an increase on the extremes; rather, they signify a sorting of liberal and conservatives into the “correct” parties as compared to a generation ago.[2]

I have argued that the extent of party sorting in the general population as compared to the political class is greatly exaggerated. Abramowitz disagrees. I believe that the evidence supports my conclusion. Yes, the correlations between party and ideology or issues have increased. But Matthew Levendusky’s extensive analyses show that while sorting has increased, it is still quite imperfect in the mass public.[3] D. Sunshine Hillygus and Todd G. Shields report that even strong partisans disagree with their parties on important issues.[4] Jeremy C. Pope reports similar findings.[5] And after a massive statistical analysis of American public opinion, Delia Baldassarri and Andrew Gelman conclude:

Thus, increased issue partisanship is not due to higher ideological coherence; rather, as suggested by Fiorina, et. al. (2005) it mostly arose from parties being more polarized and therefore doing a better job at sorting individuals along individual lines. Individuals themselves have not moved; simply, they now perceive parties as being more radical and they split accordingly.[6]

I discussed the touchstone issue of abortion in my essay as one of the most striking instances of the limits of party sorting. Abramowitz writes—incorrectly—that I claimed that “a majority of self-identified conservatives have liberal views on issues such as abortion and gun control…” No, I wrote that “more than a third of strong Democrats are closer to Mitt Romney’s position on abortion” and “more than a third of strong Republicans were closer to the Democratic position.” Here is the table using the 2008 ANES data that Abramowitz claims contradict my conclusion:

Table 1. When Should Abortion Be Permitted?

Abramowitz contends that downward trends in straight ticket voting and increased temporal consistency in voting are indicators of polarization. On the contrary, elsewhere I have explained how they are indicators of party sorting.[7] Briefly, if all the Democratic candidates on the ticket are liberals, and all the Republican candidates are conservatives, there is much less reason to split your ticket or vote differently from election to election than if each party’s candidates hold a variety of positions.

Similarly, Abramowitz contends that party loyalties are stronger today than in the past. There is no conclusive evidence of that. While it is true that Democrats (especially) and Republicans vote for their parties more consistently than in the past, that is what party sorting would lead us to expect. For example, very few Democrats today are racial conservatives, as was the case in 1964 when many defected to Republican Barry Goldwater. And very few Republicans today are racial liberals, as was the case in 1964 when many defected to Democrat Lyndon Johnson. When issues and ideology reinforce party, one cannot attribute all the causal influence to party. It may be issues and ideology producing consistent party voting, not party loyalty.

Abramowitz also confuses cause and effect in his discussion of independents. Despite the ritual citation of Bruce Keith and his co-authors, there is no conclusive evidence for the claim that independent leaners are closet partisans.[8] Rather, the tendency of leaning independents to vote for the party toward which they lean may indicate that they use their voting intention to answer the directional probe. That is, “I’m going to vote for Obama, so I guess I lean to the Democrats.”

On a more specific note, while consistent voting across states is consistent with party sorting, Abramowitz’s claim that geographic polarization is increasing is incorrect. Elsewhere, we have debunked the Bishop “Big Sort” analysis that captivated the political class some years ago.[9] And taking a long-term view Stephen Ansolabehere, Jonathan Rodden and James M. Snyder report, “The past century has produced, not a growing political divide, but political convergence among the American states and counties.”[10] Consider that about half the states have divided government after recent elections, and the fact that we see states that are solid blue at the presidential level electing Republican governors and vice versa. If partisans really were as set in their positions as Abramowitz believes, and if independents as rare as Abramowitz believes, Republican Chris Christie would not be cruising to re-election in a state that gave Democrat Obama 58 percent of its vote last November. Given different choices, many citizens will make different decisions.

Finally, as for Abramowitz’s concluding partisan broadside, I will leave it to the reader to attribute blame. There is plenty to go around.