America is not Polarized, but the Parties Are, part 15...

Now that the battle over the Debt Ceiling is settled (at least until Congress comes back from recess) we are once again treated to the usual chorus of voices bemoaning how polarized America has become. As such, I feel compelled to once again dispell the myth that America is a polarized nation.

As I argued during the healthcare battle, after the Gabbie Giffords shooting, and prior to the 2010 midterms, Americans are not polarized (at least no more polarized than at any point in the last 40 years), but our two political parties are very polarized - dominated by ideological and issue activists who do not represent the beliefs of the larger electorate.

Since the 1960s and the end of the Democratic Party’s dominance of the New Deal Era, American politics has become far more competitive. Democrats lost their solid hold on the South and Republicans lost their hold on the North East. Only 7 states had divided government in 1954, in 2007 that number stood at 23 states. With few exceptions, either party has fair shot at winning statewide elections in most states. Control of Congress and the White House has bounced back and forth between the two parties. The parties have responded to this era of competition by becoming ever more partisan and polarized.

A recent study by political scientist Daniel Coffey determined that there is a direct and positive correlation between party competition and party ideology. As a state becomes more competitive between Republicans and Democrats the respective parties become ever more conservative and liberal. V.O. Key hinted at this in 1956 when he argued that competition would force parties to offer more distinct policies to voters in an effort to influence their choice. Additionally, as competition increases the parties come to rely more heavily, not on the mean, median, or moderate voter, but rather on the more committed and activist voter. This means liberal activists for Democrats and conservative activists for Republicans. This has the effect of pushing the parties ever farther to the extreme.

As the agendas of the two political parties became ever more divergent after the 1960s Americans began to "sort" more neatly into one party or the other, a phenomenon known as party sorting. As the Democratic Party became more liberal and the Republican Party more conservative, liberal Republicans left the party and became Democrats or Independents and conservative Democrats became Republicans or Independents. The parties polarized, the public did not.

In fact, there has been precious little change in the last 30 years with regard to the share of Americans who identify as Conservative, Moderate, or Liberal. Rather conservatives no longer feel comfortable in the Democratic Party and Liberals are no longer comfortable in the Republican Party - Moderates appear to be less comfortable with both parties and that explains why Independent voters have emerged as the fastest growing segment of the American electorate. As a result of this sorting, the Republican Party became more homogenously Conservative and the Democratic Party more homogenously Liberal. So the distribution of Liberals, Conservatives, and Moderates within the the two parties has changed significantly while the distribution of Liberals, Conservatives, and Moderates within  broader electorate has changed very little.

The electorate is no more polarized now than in 1970, 1980, or 1990 - but the parties are. So of course 90 percent of Republicans vote Republican and 90 percent of Democrats vote Democrat and Presidential approval correlates to party affiliation, and Democratic members of Congress have become more Liberal and Republican members of Congress more conservative. Meanwhile, Moderate voters tend to split their support between the parties, but also vote at lower levels - likely a result of feeling ill-represented by either party. All of this is a natural byproduct of party sorting, but party sorting is not the same as polarization.

Though some scholars (especially, but not limited to, Alan Abramowitz) have advanced the theory that this growing polarization among elected officials is in fact reflective of a polarized public, there is in fact little evidence that the mass public has become polarized. Rather the polarization has occurred among committed political activists and the interest groups they support – a relatively small share of the electorate.

This is evident in Figure One, panel A shows the self-identified ideological orientation of the general electorate at three points 1972, 1994, and 2008. Although the distribution has shifted over time, the distribution remains normal with most voters amassed in the center as Moderates. Panel B is limited to those members of the electorate who indicated that they had worked for a campaign and given money to a candidate – I define these voters as party activists. Among these activists, a clear bimodal distribution is evident with the peaks of the distribution to the left and the right of moderation.

According to political scientist Anthony Downs in a winner-take-all election system coupled with two dominant parties (as in the United States) the parties will adopt positions that are ideologically attractive to the greatest concentration of voters.

The distribution of the American electorate (in the broadest sense) is unimodal which, according to Downs, should promote more moderate parties as they compete for the median voter. But Downs’ theory further suggests that if the electorate was polarized into two distinct voting blocs, one left of center and one right of center, the parties would “diverge toward the extremes rather than converge on the center. Each gains more votes by moving toward a radical position than it loses in the center (p. 143).” Steven Hill writes "one of the defining characteristics of a winner-take-all system is that it promotes adversarial politics so that on a whole host of issues it is painfully obvious that the overriding agenda for both major parties is... to stake out positions vis-a-vis the other side."

In the 1960s Republicans exploited several emerging schisms in the ranks of the Democratic Party's coalition in order to become competitive - schisms revolving around national security, welfare spending, and policies with regard to race relations. The party defined itself by being what the Democratic Party was not. Over time both parties increasingly defined themselves by being the antithesis of the other party - this is why there can be no middle ground on isues ranging from abortion, entitlement spending, taxes, environmental regulation, healthcare reform - each party's identity is inextricably linked to being the mirror image of the other party.

This led to party sorting, internal partisan homogeneity, and the bimodal distribution of the engaged voters evident in Figure One. The distribution of the most engaged and activist elements of the electorate, those most likely to participate and vote in elections, is bimodal which has resulted in two polarized parties and the system now feeds off of that bimodal distribution. So long as the more Moderate members of the general electorate remain less engaged there is little incentive for either party to moderate and pursue Moderate voters; rather to do so would risk alienating more committed activist party members. So long as neither party moderates and pursues more Moderate members of the electorate there is little incentive for Moderate voters to become more engaged – a self-reinforcing cycle exists.

The increased electoral competition between the Democratic and Republican parties, in the current era, has had a significant impact on public opinion, partisanship, polarization, and the institutions of governance. As described by Congressional scholar Norm Orstein the present era is “an extended era of close partisan margins, which gave rise to high-stakes legislative politics and sharply reduced incentives for lawmakers to work across party lines to solve problems.”

The only hope for restoring moderation, compromise, and good government - for breaking the hold on the current system of partisan activists - is for the vast and unengaged middle to re-enter the game and wrest control from the ideologues. Simply chosing between the tweedle-dee Democrats and the tweedle-dum Republicans will not suffice. Rather the current party system must be brought down.

Independent Voters are no Myth

Over at Larry Sabato's Crystal Ball page, senior columnist Alan Abramowitz takes on what he calls "myths" about Independent voters. Several of his points are quite correct and I will not quibble with them, but Abramowitz is far too dismissive of what it means to be an Independent. Abramowitz writes "the large majority of independents are independents in name only. Research by political scientists on the American electorate has consistently found that the large majority of self-identified independents are “closet partisans” who think and vote much like other partisans."

What do the numbers tell us about Independents? Though some scholars (Abramowitz among them) contend that there has been a marked rise in partisanship among the mass electorate in recent years and a solidification of partisan support, there is equally compelling evidence that the trek away from party allegiance that began in the 1960s continues through today.

According to data from the American National Election Study, in 1964 approximately 76 percent of the country identified as a “Partisan” (either weak or strong) in contrast only 24 percent identified as Independent or Leaning Independent (Leaners). By 1984 it was 65 percent to 35 percent and in 2008 it was 60 percent to 40 percent. Across that 42 year span the share identifying as Strong Partisans decreased from 38 percent in 1964 to 32 percent in 2008 and Weak Partisans declined from 38 percent to 28 percent. The share identifying as Leaning Independent, so-called Leaners, doubled from 15 percent to 29 percent. Though research has shown that many self identified Independents will express a partisan preference if pressed, there has been a clear trend toward an initial preference of Independent.

Through 1964 fewer than a quarter of the electorate self identified as a Pure Independent or an Independent leaning Democrats or Republicans. By 1968 that share had risen to 30 percent, and as of 2008 was at 39 percent of the electorate. There is no indication of decline in the Independent preference among voters. Though they may lean Democrat or Republican Americans are clearly less willing to express allegiance to one of the two major parties than they were in the 1950s and 1960s.

Some scholars use the same data to make the opposite argument concluding “if voters are becoming more partisan, we would expect declines in pure independents to result in increases in independent partisans. This is evident for both Democrats and Republicans.” That conclusion is debatable. If indeed we are seeing an increase in partisanship then perhaps we would see a decline in pure independents, but it is also likely that we would see a rise in partisans of all stripes – Strong, Weak and Independent. For Democrats, this has not happened and the rise in Independent Democrats appears not to have come from declines in pure Independents, but rather a decline in Weak Democrats – this does not suggest growing partisanship, but rather weakening partisanship.

The chart above tracks Democratic Party identification since 1952 and shows a clear decline in Democratic Party affiliation. From 1952 through 1968 Democratic Party affiliation averaged 54.9 percent of the electorate, from 1970 through 1980 the average was 52.3 percent and since 1982 the average is 49.8 percent. If one were to narrow the window to the 1990-2008 period the average is 49.7 percent. Though the share of Strong Democrats has rebounded from its lows in the 1970s it remains well below its average from 1952 through 1968. There also has been a marked decline in the share of Weak Democrats since the 1960s and it appears that the increase in Independent Democrats has been driven by the decline in Weak Democrats – not by a decline in Pure Independents. In short, fewer people are identifying as Democrats than did so prior to 1968 and those who do identify as Democrats are increasingly identifying as Independent Democrats.

The ANES data does show an increase in Republican partisanship, but this in entirely consistent with the collapse of the New Deal era and with it the dominance of the Democratic Party.  Even among Republicans, however, there has been considerable growth among Independent partisans and evidence of a rebound among Strong Republicans following the lows of the 1970s – likely an artifact of the Nixon Administration and Watergate. The Republican share of the electorate averaged 33.9 percent from 1952 through 1968, then fell to 32.1 percent during the 1970s, and climbed to 38.2 percent in the decades since 1980. But the share of Strong Republicans from 1982 through 2008 was 12.2 percent, essentially the same as the average of 12.4 percent prior to 1968 – the “rise” in Strong Republicans is again likely an artifact of a collapse in Republican Party identification during the 1970s.

That said, the true growth area in partisan politics has been among the group of Americans Identifying as Independent, either Pure, or Leaning Democrat or Republican. Though there are fewer Pure Independents as a share of the electorate than during the decade immediately following the collapse of the New Deal Era in the late 1960s and during the height of the Watergate era there are more Pure Independents, Independent Democrats, and Independent Republicans today than during any measured era dating to the 1950s.

So when Gallup release a graph, like the one below, Democrats and Republicans should be concerned. Though Democratic Party affiliation matched a 20 year peak of 36 percent in 2008 it has since fallen to its lowest levels (according to Gallup) in 22 years. The Gallup data clearly shows that neither party can claim the allegiance of a majority of the electorate and since 1990 has rarely captured a plurality. At best, the data on party affiliation, whether from ANES or Gallup, suggest that there has been some solidification of the Strong Partisan base of each party, though collectively these Strong Partisans account for less than one-third of the electorate.

Are Independents independent in name only? No, not really. In The Myth of the Independent Voter, Keith et al., (1992) argued there has been little change in partisan attachment since the 1960s. Rather, the authors contend people identifying as Independents often reveal a preference for one of the two major parties when pressed to make a choice. These Leaners behave much the same as their more partisan counterparts with regard to issue positions and vote choice - this is the Abramowitx argument. In a 2009 article in Electoral Studies, political scientist John Petrocick argued, “Leaners are partisans. Characterizing them as independents underestimates the partisanship of Americans…”

But studies of partisanship often consider the views of Leaners at a given point in time or their votes in a specific election or examine the stability of partisan identification by merging all partisans – Strong, Weak, or Leaning – together and measure macro-level party identification. These measure do not take into consideration the temporal nature of partisan attachment and the propensity to change party affiliation over time. Additionally, in an era of candidate-centered politics it is possible that Idependents express a preference for the party of the candidate that they have chosen to support. So it would not be surprising that an Independent that leans Democratic votes the same as a Strong Partisan Democrat - in a specific election. But the Strong Partisan Democrat voted Democratic because of partisanship, the Independent, however, voted Democrat and expressed a Democratic preference because of the candidate.

But the lerger question really pertains to the size and stability of a governing coalition over time. For a President or a political party to succeed they must have a stable electoral coalition. If Independent voters are the fastest growing segment of the electorate and if they are truly independent, then Democrats and Republicans need to worry. But if folks like Alan Abramowitz are correct, then the parties can ignore the threat of the Independent voter.

A review of ANES data from a panel survey that included 2000, 2002, and 2004 however, shows that Independent Democrats and Independent Republicans (roughly a third of each party) are much less attached to their party than either Weak or Strong partisans over time.

Of those respondents who self-identified as an Independent Democrat in 2000, 31.4 percent no longer identified with the Democratic Party in 2002, nearly as many, 29.8 percent, no longer identified with the party in 2004. For Republican Leaners the results were similar, 27.2 percent no longer identified with the Republican Party in 2002, 26.1 percent in 2004.

Strong as well as Weak partisans left their respective parties at far smaller rates over time. Equally worthy of not, Independent Democrats and Republicans who left their 2000 party were just as likely to identify with the opposition party in 2002 and 2004 as they were to simply identify as pure Independents.

A review of 2004 partisan identification shows that fully one-third of Independent Democrats and Independent Republicans in 2004 identified with another party or no party in 2000. Simply stated, partisan identification is much less stable among Independent partisans and Eric McGhee and Daniel Krimm (in a 2009 issue of Polity) found that Independent partisans are more moderate than Strong partisans.

Abramowitz and others are correct - Independent Partisans do vote much like their Strong Partisan counterparts in a given election, but the ANES panel data suggests that a significant share of Independent Partisans (between a quarter and a third) may well have a different partisan stripe by the next election cycle. Far from the “unmoved mover” described in The American Voter (Campbell et al., 1960), partisanship among Independent Partisans moves and they are the fasted growing segment of the electorate.

Fully 11% of the electorate are Pure Independents, another 30% are Independent Partisans (about 18% Democrats and 12% Republicans) and between one-quarter and one-third of these Independent Partisans are far less attached to party from election to election. At the very least, this suggests a 20% voting bloc that is quite volatile - in a country where our presidential elections have been decided by margins of 7 percentage points or less since 2000 and the difference between the national two party vote share in House elections has averaged about 5 percentage points since 1990.

Independent voters are no myth, they matter, and (when you include among them Independent Partisans) they absolutely sway elections. Indeed, the present and highly competitive political era in which we are living is a direct result of a decrease in partisan attachment coupled with an increasing level of partisanship among each party's relatively equal (and small) in size base.