Political Values and Information Effects

Lupia, Levine, Menning, and Sin's primary empirical contribution is a regression analysis demonstrating the heterogeneity of information effects on support for the 2001 Bush tax cut. More-informed Republicans and conservatives were roughly similar to less-informed Republicans and conservatives in their views about the tax cut, while more-informed Democrats and liberals were vastly less favorable than their less-informed counterparts. These relationships appear weakly in Lupia, Levine, Menning, and Sin's figure 2 (due to substantial measurement error in the NES interviewers' information ratings); they appear much more clearly in the statistical alanlyses reported in Lupia, Levine, Menning, and Sin's table 1 (which rely, as my own analysis did, on instrumental variables estimation to correct for attenuation in the parameter estimates).¹

I certainly agree with Lupia, Levine, Menning, and Sin's key claim that political information may have different effects on different people. Indeed, I made the same point in print two decades ago, and again in critiquing Lupia's own previous work on information effects.²

Where we disagree is in assessing the implications of this heterogeneity for my conclusions about the bases of support for the Bush tax cut. According to Lupia, Levine, Menning, and Sin, the fact that Republicans and conservatives “supported the tax cut at extraordinary high levels, regardless of their information ratings” (777) renders my conclusions “erroneous” (774). My own view is that their reanalysis provides a useful but hardly surprising elaboration of the relevant finding in my article.

The fact that ideological and partisan disagreements about the tax cut only emerged clearly among relatively well-informed people will come as no surprise to anyone who has read John Zaller's influential study of The Nature and Origins of Mass Opinion. Zaller provided several examples of similar “polarization effects” in cases “in which political elites heatedly disagree.”³

However, the tax cut case is atypical in one politically crucial respect. In most instances I have seen of polarization effects, the meeting point of uninformed opinion is near the mushy middle of our opinion scales, with better-informed citizens gravitating more or less symmetrically toward their respective poles. In the case of the Bush tax cut, however, uninformed citizens regardless of their political values were very likely to support the policy—if they took any position at all.⁴

My point, in the portion of “Homer Gets a Tax Cut” that Lupia, Levine, Mening, and Sin focus upon, was that this marked asymmetry had a dramatic impact on the overall distribution of public support for the tax cut. Nothing in their reanalysis of the NES data provides any grounds for doubting that a better-informed public would have been markedly less enthusiastic about the tax cut. Rather, what they show is that the information effect I documented was concentrated among Democrats and liberals.⁵

They stress the fact that “higher information ratings have no significant effect on conservative or Republican support for the tax cut” (777), but fail to mention the fact that their estimated information effect is twice as large among Democrats as for the entire sample, and three times as large among liberals. Nor do they seem to notice that the implication of their reanalysis for the overall distribution of support for the tax cut is exceedingly modest. For example, the average information effect for Republicans and Democrats in the third and fourth columns of their table 1 (weighted by their numbers in the sample) is −.700, virtually identical to their estimate of −.721 for the entire sample. Thus, I take the results of Lupia, Levine, Menning, and Sin's reanalysis to be quite consistent with my characterization of the impact of information on public support for the tax cut. Since they clearly do not, I conclude that our disagreement on this point must be one of interpretation rather than of substance.

“Simple Ignorance” and Support for the Bush Tax Cut

Much of Lupia, Levine, Menning, and Sin's critique focuses on my claim that “the strong plurality support for Bush's tax cut … is entirely attributable to simple ignorance.” They insist that the partisan and ideological disparities in information effects documented in their table 1 invalidate this assertion, but never really explain why. Indeed, despite referring to the “simple ignorance” claim more than a dozen times, they never pause to consider what it means or how I arrived at it. The previous sentence of my article specified the basis for the claim: the predicted level of support for the tax cut implied by my analysis (on a scale ranging from +1 for strong support to −1 for strong opposition) ranged from +.84 (indicating near-unanimous support) at the bottom of the information scale to −.06 (indicating an equal mix of support and opposition) at the top of the information scale.⁶

Not −1 (“support for Bush's tax cut … is entirely attributable to simple ignorance”), but −.06 (“the strong plurality support for Bush's tax cut … is entirely attributable to simple ignorance”). This calculation seems to me to leave plenty of room for agreement with Lupia, Levine, Menning, and Sin's quite sensible insistence that “much of the support for the tax cut is attributable to something other than ignorance” (777).

An analogy may be helpful here. Wand et al. provided very convincing statistical evidence that thousands of voters in Florida's Palm Beach County mistakenly marked their ballots in the 2000 presidential election for Patrick Buchanan rather than Al Gore. The authors concluded that Palm Beach County's confusing “butterfly ballot” was “pivotal in the 2000 presidential race.”⁷

Obviously, it was not the case that everyone who voted for Buchanan, even in Palm Beach County, did so mistakenly. Nor was it the case that mistakes were equally common among Democrats and Republicans or among liberals and conservatives. (Indeed, the authors' careful demonstration that excess Buchanan votes were more common in heavily Democratic precincts constituted one of their most important pieces of evidence that those votes were, in fact, cast mistakenly.) Nevertheless, Bush's slim plurality in Florida was entirely attributable to voters' mistakes in the following sense: no mistakes, no Bush plurality (and no Bush presidency).

The point I tried to make in the sentence on which so much of Lupia, Levine, Menning, and Sin's critique seems to hinge is that the strong plurality support for Bush's tax cut in the 2002 NES survey was entirely attributable to ignorance in just the same sense that Bush's narrow plurality in Florida was entirely attributable to the butterfly ballot: no ignorance, no strong plurality.⁸

Of course, that does not imply that everyone who supported the tax cut was ignorant or misguided. It does suggest that a fully-informed citizenry would have been lukewarm toward the tax cut, at best. Is that fact relevant to understanding “Inequality and Public Policy in the American Mind”? Lupia, Levine, Menning, and Sin don't say. I thought it was worth a paragraph.

Lupia, Levine, Menning, and Sin's apparent misunderstanding of my “simple ignorance” claim is facilitated by ignoring most of the rest of my article, which examined a variety of potential bases of public support for the Bush tax cut. Some of my findings seem quite consistent with their basic point that people “approach political problems from varying ideological perspectives and with different values in mind” (779). Indeed, that observation—which they offer as a basis for criticizing my analysis—is a virtual paraphrase of the first of my “four important conclusions,” that “opinions about tax policy are shaped by broad political values such as partisanship and ideology.”⁹

On the other hand, several of my findings struck me as very hard to square with any straightforward ideological or “values” interpretation. Why were people who thought their own taxes were too high so much more likely to support the tax cut, despite the fact that it would have only a modest impact on most people's own tax bills? Conversely, why did attitudes about the tax burden borne by the rich have no apparent effect on support for the tax cut, despite the fact that most of the benefits were targeted to wealthy taxpayers? Why were people with modest incomes, who got little or nothing in the way of direct benefits, more supportive than wealthy people, who got substantial tax reductions? Why were people who thought the government wasted a lot of tax money less supportive of tax cuts than those who thought the government was spending money efficiently? Lupia, Levine, Menning, and Sin have no light to shed on any of these findings—beyond insisting as a general matter that “much of the support for the tax cut is attributable to something other than ignorance” (777) and that “citizens have reasons for the opinions they have” (780).

Reasonable Disagreement

Lupia, Levine, Menning, and Sin attach considerable significance to the fact that some splendidly well-informed liberals and Democrats supported Bush's tax cuts. Their most striking example—an “endorsement” from prominent liberal economist (and my Princeton colleague) Alan Blinder—is highly strained, to say the least. The sum total of Blinder's comments in the radio news story they offer as a source for their characterization consisted of the following two sentences: “The timing turns out to be, more or less by accident, very good…. The right kind of bill to have passed, it's too late now, would have a lot of tax relief in the short term—even more than was legislated—but much less eight, nine, ten years from now.” Later the same year (November 28, 2001), Blinder testified before the Joint Economic Committee of the U.S. Congress, stressing the unfulfilled need for a bipartisan short-term stimulus plan that “cuts taxes only where the tax cuts do the economy some good” (emphasis original). A few years later, he wrote that “the most profound, and profoundly disturbing, innovation in budget policy during the administration of George W. Bush has been to discard the old-fashioned notion that presidents who propose a tax cut or new spending should also propose some way to pay for it.”¹⁰

If Blinder is the poster boy for support of Bush's tax cut among liberal elites, the “variance on the left” (778) is much less impressive than it sounds in Lupia, Levine, Menning, and Sin's account.

On the other hand, there clearly were important Democratic elected officials who supported Bush's plan. Lupia, Levine, Menning, and Sin claim that seven Democratic senators voted for the 2001 tax cut; in fact there were twelve. However, there was less to this bipartisan support than meets the eye. Four of the twelve Democratic supporters—Max Baucus, John Breaux, Blanche Lincoln, and Robert Torricelli—sat on the Senate Finance Committee, which had “crammed in several more provisions” to make the president's original proposal more palatable to Lincoln and Torricelli, among others.¹¹

Six of the twelve—Baucus, Toricelli, Jean Carnahan, Max Cleland, Tim Johnson, and Mary Landrie—faced extremely tough reelection fights in 2002.¹² Three others—Herbert Kohl, Diane Feinstein, and Ben Nelson—were themselves multi-millionaires, reporting net worth in excess of $5 million in 2002, according to a CNN tally.¹³

Of course, neither personal wealth nor reelection concerns nor special provisions for committee members figured in the reasons these senators offered for voting the way they did. If our goal as social scientists is, as Lupia, Levine, Menning, and Sin (779) have it, to “better fit our analyses into their rationales,” then these factors are simply irrelevant to understanding the politics of the tax cut. However, that would seem to me to be a counterproductive narrowing of our explanatory repertoire. People's “rationales” are sometimes unreliable bases for understanding and explaining their behavior.

Beyond Extenuation

I noted in “Homer Gets a Tax Cut” that 40 percent of the respondents in the 2002 NES survey said they hadn't thought about whether they favored or opposed the 2001 Bush tax cut. Lupia, Levine, Menning, and Sin counter that the vast majority of these people did claim to have thought about the imminent war in Iraq or corporate scandals or the 2002 campaign. “The fact that many do not pay attention to an issue such as tax cuts,” they argue (778), “may say little about their political competence more generally.” Since I did not assert anything about people's “political competence more generally,” I am happy to leave to others the difficult question of how many multi-trillion-dollar issues a competent citizen can afford to ignore. I agree that citizens are busy, that they pick and choose what to pay attention to, and even that they “lack the opportunity to act on individual policy proposals” (778). However, those are (potential) justifications for ignorance and superficial thinking about tax policy, not evidence against my characterization of public opinion on this issue.

Lupia, Levine, Menning, and Sin's unsubtle effort to, quite literally, change the subject is in service of a broader goal: rationalizing everyday political thinking and behavior. As Luskin aptly put it, after decades of scholarly research “the controversy over how sophisticated most people are has not really died so much as transformed itself. The Panglossian line has merely shifted from denial to extenuation.”¹⁴

Lupia, Levine, Menning, and Sin seem to me to be shifting the Panglossian line another significant step further—from extenuation as a predilection to extenuation as a philosophical axiom and methodological precept. The heart of their argument is that “citizens have reasons for the opinions they have,” and that we as social scientists should “conduct scholarship that attempts to better fit our analyses into their rationales.” By thus “rethinking the relationship between the people who study social phenomena and the people who are being studied,” they suggest, we can “learn a great deal about modern politics” while also providing “a stronger and more credible foundation for progress” in meeting “critical social needs (such as those of the poor)” (780).

I agree that understanding people's rationales for their own behavior is one crucial element of successful explanation in the social sciences. However, decades of psychological research have amply demonstrated that people's subjective understandings of their own behavior are often incomplete, predictably biased, and sometimes highly misleading.¹⁵

Thus, a methodology of social inquiry limited to understanding people's rationales seems decidedly incomplete on purely scientific grounds, whatever merits it may have from the standpoint of “social marketing.”

Moreover, even when the reasons people offer for their own behavior are accurate reflections of why they thought or did what they did, it does not necessarily follow that those reasons are cogent or politically sensible. Psychologists who study human decision-making focus much of their attention on the implications of pervasive biases, cognitive shortcuts, and inferential errors. In assessing judgments or choices, they attempt to “avoid overly strict interpretations, which treat reasonable answers as errors, as well as overly charitable interpretations, which attempt to rationalize every response.”¹⁶

Usually, the psychologists' evaluative task is greatly simplified by focusing on simple experimental situations in which an “error in judgment” can be “demonstrated by comparing peoples' responses either with an established fact (e.g., that the two lines are equal in length) or with an accepted rule of arithmetic, logic, or statistics.”¹⁷

For better or worse, politics in natural settings is generally much more complicated. As Lupia, Levine, Menning, and Sin (780) tell us, “Members of politically relevant groups often disagree.” As a result, one side's “established fact” may be disputed or distorted by political opponents, and one observer's “reasonable answers” may be another's “errors.”

The complexity and contestedness of the political world make evaluation of political decision-making a difficult enterprise, fraught with opportunities for misjudgment and misunderstanding. But they also make it more important in both scientific and political terms. Ignorance, confusion, misinformation, and wishful thinking are likely to loom large in any realistic account of ordinary citizens' political beliefs and judgments—that is, unless we are willing to stipulate, as a matter of faith, that the same people who consistently misjudge the lengths of lines and the ramifications of simple probability statements somehow invariably succeed in mastering political issues of staggering difficulty.

Lupia, Levine, Menning, and Sin (779) worry that “to reach the conclusion that voters are ‘unenlightened,’ ‘simple-minded,’ or ‘simply ignorant’ requires a great deal of speculation—speculation for which direct evidence is scant.”¹⁸

I agree that that is a daunting concern, and well worth worrying about. However, I do not think that it must or should be crippling. Rather, we should use whatever evidence we have to shed whatever empirical light we can on the consistency and cogency of people's beliefs, values, attitudes, and behavior—and we should recognize that better data or more insightful analysis may later prove us wrong. If that sounds a lot like what social scientists always do when tackling difficult issues with imperfect evidence, it is.

In an essay on “New Challenges to the Rationality Assumption,” Daniel Kahneman noted that “The criteria for using the terms ‘rational’ or ‘irrational’ in non-technical discourse are substantive: one asks whether beliefs are grossly out of kilter with available evidence, and whether decisions serve or damage the agent's interests.”¹⁹

In my view, several of my findings in “Homer Gets a Tax Cut”—including the finding that politically uninformed people were, on average, much more supportive of the Bush tax cut than better-informed people were—bear on the “rationality” of public opinion in this non-technical sense.

That is not to say that the beliefs and behavior of better-informed people should always or automatically be considered substantively rational. Sometimes, as Lupia, Levine, Menning and Sin (782) surmise, “even when people are well informed they still do not meet [my] preferred standard of decision making.” Well-informed people are sometimes quite wrong about things—even when it comes to straightforward factual matters. For example, well-informed conservatives in the 2002 and 2004 NES surveys were significantly more likely than less-informed conservatives to deny that differences in income between rich people and poor people in the United States had increased over the past 20 years—a denial “grossly out of kilter with available evidence.”²⁰

Here, as in many other instances, better-informed people seem mostly to have grasped the biased world-view of “their” political elites rather than an accurate perception of real social conditions.²¹ Results like these have cured me, at least, of the notion that the views of well-informed people might by themselves form a reliable basis for normative assertions about political interests or “enlightened preferences.”²²

Criticizing the behavior of citizens or policy-makers is a messy and contentious business. It may even be counterproductive from the standpoint of “social marketing”; I don't know. In any case, our primary task as social scientists is not to tell people what they want to hear, or to “better fit our analyses into their rationales,” or even to “match critical social needs (such as those of the poor) with citizens' desires.” Rather, it is to understand how people think and act, and why. If they think and act like Superman, so much the better for them, and for us. Still, we should leave some room in our understanding of social science—and in our understanding of human nature—for the possibility that they (we all) sometimes think and act more like Homer Simpson.