Fifty years after its underlying polling was done, “The Political Beliefs of Americans” by Lloyd Free and Hadley Cantril remains the starting point for anyone who wants to understand the big picture of American political opinion. But a new report on political polarization from Pew adds important new information, even as it leaves some questions tantalizingly unanswered and/or underanalyzed. Most notably, the twin issues of what’s driving polarization and how symmetric or asymmetric it is remain to be fully explored, while the greatest locus of continued agreement is overlooked.

In a section titled “Is Polarization Asymmetrical?” the study first suggests that Democrats are driving the process, before introducing two caveats. This finding is simply not credible in the America described by Norm Ornstein and Thomas Mann in “It’s Even Worse Than It Looks: How the American Constitutional System Collided With the New Politics of Extremism,” or, more basically, in the system that’s produced the sharp rightward trend in roll call voting among congressional Republicans since 1980 (as mapped by the DW-Nominate scores of Keith Poole and colleagues) compared to a relatively modest leftward drift among Democrats. [House mean, Senate mean.] Moreover, Pew’s report itself contains a variety of contradictory information — not least the fact that the most consistent liberals are the people who most want political leaders willing to compromise.

While others have written extensively about Pew’s new report, I’d like to focus more on questions raised, on what’s not resolved, and that requires an initial discussion of why Free and Cantril remain relevant to this day. Their most innovative approach was the use of three different frameworks — an ideological spectrum based on questions about government intrusion vs. individual self-reliance and the free market, an operational spectrum based on support for specific government spending programs, and self-identification as liberal, conservative or middle of the road. A key finding was that 50 percent of Americans were ideological conservatives, while 65 percent of Americans were operational liberals — an overwhelming supermajority. More particularly, this psychological split manifested within the ranks of ideological conservatives: 46 percent of them were operational liberals, compared to just 26 percent who were operational conservatives. Thus, when you hear Tea Party Republicans say, “Keep the government’s hands off my Medicare!” what you’re actually hearing is just the tip of an iceberg that Free and Cantril were the first to discover and begin exploring.

The ideological and operational spectrums paint starkly different pictures of America — with the self-identification scale falling in between, with a 38 percent plurality calling themselves middle of the road. One consequence of these two pictures was a suggested difference in how to identify core liberals and conservatives. Since so many people supported spending more money on the operational questions, the relatively small fraction who don’t are prime candidates for the conservative label. Similarly, since so many people opposed government intervention on principle, those who don’t are prime candidates for the liberal label. It makes sense to define these ideologies in different terms, because that’s what produces relatively compact groups with coherent policy views on both sides.

Compared to Free and Cantril, Pew’s handling of the ideological spectrum represents a dramatic improvement in two fundamental ways: First, they’ve created an ideological consistency scale that includes a broader range of issues, including questions on race, immigration, foreign policy, homosexuality and the environment — several of which weren’t even salient political issues in 1964. Second, their questions are framed by presenting two contrasting positions and ask people to choose between them, whereas Free and Cantril presented one statement, and asked people to agree or disagree — a practice now widely understood as biased toward agreement. What’s more, Pew uses its ideological consistency scale in a gradiated manner — consistent or mostly liberal or conservative, with a mixed category in the middle — which mirrors Free and Cantril’s scale of complete or predominant ideologues, with a middle-of-the-road category between them, and Pew uses this scale consistently and creatively throughout its study.

Despite Pew’s excellent use of its ideological consistency scale, it’s important to understand what’s not being captured in its analysis: the operational level, which is where much higher levels of agreement still reside. As an example of how ideological value statement questions differ from operational spending questions, consider what Pew has to say regarding spending on the poor, after noting that partisan divides have deepened on nearly every measure in the ideological consistency scale:

For instance, while Democrats have always been more supportive than Republicans of the social safety net, the partisan divide on these questions has increased substantially over the last 20 years. Two-thirds of Republicans (66%) believe that “poor people today have it easy because they can get government benefits without doing anything in return;” just 25% say “poor people have hard lives because government benefits don’t go far enough to help them live decently.” Among Democrats, just 28% believe the poor have it easy. The partisan gap on this measure is now 38 points, up from 19 points in 1994 and 26 points in 2004.

Things look dramatically different, however, if we look at specific spending questions. I prefer those on the General Social Survey (GSS — online archive here), which is considered the gold standard of public opinion polling in the U.S. GSS spending questions ask if we are spending “too much,” “too little” or “about right.” In 2012 — the last date GSS data is available — 52.3 percent of Republicans thought we were spending “too little” on the poor, compared to 23.1 percent who thought we were spending “too much.” That’s almost exactly the reverse of the picture Pew painted above — a perfect illustration of the ideological/operational divide that Free and Cantril first observed in 1964 survey data.

If we ask about “welfare,” views are decidedly more negative because the term itself is so pejorative. Even liberals tend to think we’re spending “too much” rather than “too little” in most years. But even support for spending on welfare skyrockets if we shift our focus even closer to the ground. In 1986, GSS included a special module of questions presenting people with different scenarios of families in need, asking how much support they should receive. The results were presented in detail in “The Deserving Poor” by Jeffry A. Will. Broken down into nine regions nationally, one of Will’s charts showed that the average levels of support people thought appropriate exceeded the actual levels of support by 2-to-1 or better in all but one region, where it was just slightly less than 2-to-1. In one region — the East South Central — the average level of support people thought appropriate was $265/month, more than 10 times the lowest level of actual support for one of the states in the region.

This study clearly indicates that the ideological/operational divide that Free and Cantril discovered only gets more extreme, as the operational questions become more concrete, more rooted in questions of specific need. At the “hyper-operational” level that Will explored, virtually everyone is a liberal, compared to existing policy.

As for the kind of trend data Pew cited in the passage above, we can compare GSS’s spending question trend using the so-called liberalism index, to get a single number to measure support — the number saying we’re spending “too little” by the number saying “too little” or “too much.” In contrast to Pew’s ideological findings, between 1994 and 2012, both liberals and conservatives became more supportive of spending on the poor and on welfare, though the gap between them did widen, because liberal support increased more. The gap increased 9.8 percent for spending on welfare, from 17.1 percent in 1994 to 26.9 percent in 2012. So the gap widened much less — barely over half the change in Pew’s measure — and did so entirely because liberals became more supportive more rapidly than conservatives did, not because conservatives grew less supportive (they didn’t). The gap increased even less, 5.1 percent for spending on the poor, from 19.7 percent in 1994 to 24.9 percent in 2012.

More broadly, on 25 GSS spending questions, the “liberalism index” moved in the same direction for both liberals and conservatives on 17 items — more than two-thirds of them. It only moved in the opposite direction on eight of them. On those eight, the conservative movement averaged 7.7 percent more conservative, compared to 6.8 percent more liberal on the other side. So, on the minority of items where liberals and conservatives moved in the opposite direction, conservatives moved more on average than liberals did. This markedly different and more nuanced array of shifting opinions shows the continued importance of the ideological vs. operational view of American politics, as well as the greater degree of agreement on the operational level. Pew’s considerable advances in developing a more sophisticated understanding of the ideological dimension cry out for a comparable level of work on the operational dimension — and, most important, on the interactions and inter-relations between the two.

But what about the questions I raised earlier, about the twin issues of what’s driving polarization and how symmetric or asymmetric it is? Here Pew hews toward a “balance narrative” approach, in which “both sides do it” more or less equally. But its own reported findings undercut that narrative, as does other available evidence.

First, let’s take a closer look at what Pew reported in a section titled “Is Polarization Asymmetrical?” Here, Pew singles out the long-term leftward shift among Democrats as “particularly noteworthy”:

The share of Democrats who are liberal on all or most value dimensions has nearly doubled from just 30% in 1994 to 56% today. The share who are consistently liberal has quadrupled from just 5% to 23% over the past 20 years.

In absolute terms, the ideological shift among Republicans has been more modest, in 1994, 45% of Republicans were right-of-center, with 13% consistently conservative. Those figures are up to 53% and 20% today.

Pew then goes on to say, “there are two key considerations to keep in mind before concluding that the liberals are driving ideological polarization.” First is that 1994 was a relative high point in Republican conservatism, as Republicans moved “substantially toward the center ideologically” from 1994 to 2004, “as concern about the deficit, government waste and abuses of social safety net that characterized the ‘Contract With America’ era faded in the first term of the Bush administration,” after which “Republicans have veered sharply back to the right on all of these dimensions,” so the GOP’s rightward shift “over the past decade has matched, if not exceeded, the rate at which Democrats have become more liberal.”