Ideological Asymmetry between the Parties

Our overarching hypothesis is that Republicans take ideological considerations into account to a strikingly greater degree than Democrats. If this is true, what else must be true? Minimally, Republicans must be markedly more likely than Democrats to identify with the ideological outlook with which their party is identified. Republicans must also be markedly more likely than Democrats to understand the ideological logic of the party system; that is, to understand that the Republican Party is the conservative party, and the Democratic Party is the liberal party. Finally, Republicans must not merely be more cognizant of the ideological orientation of their party, but also be more likely to base their evaluation of their party on ideological or programmatic considerations. To the extent that Republican identifiers are decisively more likely than Democrats to meet these conditions, an assumption of the synonymy of the meaning of party identification should be rejected.

Figure 1 presents results addressing all three conditions – sorting, knowledge of the ideological reputations of the parties and active use of ideological/policy considerations in evaluating the parties – for Republican and Democratic identifiers, separately.Footnote ¹⁶ The solid lines track the proportion of Democratic partisans; the dotted lines those of Republicans. The data, which are smoothed over individuals and years, come from the 1972–2008 ANES cumulative data file.

Fig. 1

Indicators of ideologically grounded reasoning by party and over time

Although both Republicans and Democrats are markedly more likely to be sorted today than forty years ago, Republicans are markedly more likely to identify as conservatives than Democrats are to identify as liberals (Panel A). This gap has remained remarkably stable over time – a Republican identifier is roughly 30 per cent more likely than a Democratic one to be sorted.

Panel B of Figure 1 reports the proportion that knows the ideological reputations of the two parties.Footnote ¹⁷ The results in this panel are virtually identical to those in Panel A. The proportion of Republican identifiers that knows the parties’ ideological reputation is dramatically higher than that of their Democratic counterparts. If anything, this difference has increased over time, from a difference in 10 per cent or so in the 1970s, to more than twice that today.

Panel C of Figure 1 reports the proportions of partisans that give ideological/policy grounded reasons for liking their political party, again stratified by party identification.Footnote ¹⁸ The results are even more striking in this panel than in the upper and middle panels. Quite simply, the proportion of Republicans giving ideological reasons for liking their party outnumbers the proportion of Democrats giving similar reasons by a margin of two to one.

If only a minority of Democrats is bound to their party for ideological reasons, what is the majority thinking of when they think of themselves as ‘Democrats’? The answer is revealed in Panel D of Figure 1, which plots the proportion that likes their party because of its connections to various social groups.Footnote ¹⁹ Consistent with Green, Palmquist and Schickler,Footnote ²⁰ Democrats refer at least as often to the party’s constituent social groups as its ideological underpinnings. In contrast, the percentage of Republicans that likes their party for its relationship to social groups barely reaches double digits.

These results bring into sharp focus the depth of the ideological asymmetry in the American party system. Republicans are, and have been for at least forty years, in another league than Democrats. Politically sophisticated Republicans far outpoint their Democratic counterparts in terms of: (1) the likelihood that they are ideologically sorted; (2) their understanding of the ideological structure of the party system; and (3) their readiness to use ideological considerations to explain why they like their party.Footnote ²¹

Of course, it can be argued that the asymmetry between the two parties has little or nothing to do with the parties themselves. It may instead be a product of long-standing socio-economic differences between the parties. For instance, Democrats tend to draw support from more disadvantaged groups than do Republicans and, according the ANES data, over the past forty years, 17 per cent of Democrats hold a college degree compared to 28 per cent of Republicans. The Republican coalition has also been consistently smaller than the Democratic coalition. Since the New Deal, the number of citizens identifying as Democrats has outnumbered those identifying as Republicans, sometimes at a rate of almost two to one. The differences between parties could simply be due to this imbalance in coalition size: it is easier to hold together a smaller coalition than a larger and more diverse one.Footnote ²²

To explore these potential confounds, we document the asymmetry in two more ways. As the size of the Republican coalition has edged closer to that of the Democratic coalition over time, we focus on three years in which the party identification imbalance varies. In 1980, roughly 19 per cent more ANES respondents identified as Democrats than Republicans. By 1996, this gap narrowed to 11 per cent, and by 2004, only a 3 per cent gap remained. If the asymmetry is due simply to differences in coalition size, the relationship between party identity and ideologically grounded reasoning should be weaker in 2004 than 1996, and weaker still than in 1980. Additionally, if the asymmetry is due to the social diversity of party coalitions, then any effect of party identity on ideologically grounded reasoning should evaporate with the imposition of demographic controls.

Figure 2 documents the relationship between ideologically grounded reasoning and party identity in each of the three years. Here we operationalize the dependent variable as the sum of the individual indicators of ideologically grounded reasoning (sorting, awareness of reputation, ideological attraction and group attraction (reverse scored)).Footnote ²³ In each year, we utilize coarsened exact matchingFootnote ²⁴ to ensure that Democrats and Republicans are balanced across a set of covariates that distinguishes them, including socio-economic status, race and indicators of political attentiveness. In each of the three years, Republicans scored higher on average – roughly three-quarters of a point on a four-point scale – than Democrats on the Ideological Awareness Index.Footnote ²⁵ The apparent secular trend toward a closing gap is an artifact of the years we chose in this analysis – in 2002 the difference between Republicans and Democrats was the same size as in 1980; in 1972 the difference was the same as in 2004.Footnote ²⁶

Fig. 2

Ideological sophistication between matched partisans

The asymmetry in ideological awareness between Republicans and Democrats is a matter of degree: it is not as though one party is ideologically committed and sophisticated and the other is not. But to appreciate the potential political import of this asymmetry, it is important to have at least an approximate sense of its magnitude. Accordingly, we again utilize the Ideological Awareness Index. Then, following Zaller and Bartels,Footnote ²⁷ we employ interviewers’ assessments of respondents’ knowledge of politics as a measure of ideological sophistication to illustrate how striking the difference in ideological awareness is between the parties over the last four decades.

These results, which can be seen in Figure 3, provide perhaps the clearest evidence of an ideological asymmetry. The level of ideological sophistication is, and has been for at least forty years, higher for less politically aware Republicans than it is for highly aware Democrats. In 2004, the mean ideological sophistication score for Republican respondents rated as more informed was 3.5; for more-informed Democrats the ideological sophistication score was only 2.77, and less-informed Democrats were far less ideologically sophisticated (2.07) than less-informed Republicans (3.26). The evidence is clear: ideological sophistication is tied to being a Republican to a markedly greater degree than to being a Democrat.

Fig. 3

Ideological sophistication by party, political knowledge (PK) and over time

Uphill Issues for Republicans and Ideological Awareness

The liberal position on most issues that make up the social welfare agenda, Stimson and his colleagues have shown, is more popular than the conservative position. All in all, the social welfare agenda thus favors the Democratic Party.Footnote ²⁸ To index ideological awareness, we rely on the 2012 ANES and form an additive measure combining knowledge of the ideological logic of the party system, classifying oneself into the ideological category matching one’s party, knowledge of political figures and institutions (α=0.71).

Hypothesis 2: predicts that Republicans who are ideologically aware will distinctively resist the strong appeal of increasing support for core social welfare policies – including social security and childcare. Table 1 summarizes the mean scores of Republicans and Democrats, divided into high and low scorers on the index of ideological awareness (based on a median split), on a selection from the large number of easyFootnote ²⁹ issues for Democrats, garnered from the 2012 ANES.Footnote ³⁰

Table 1

Spending Preferences on Republicans’ Uphill Issues

For Democrats, a wide-gauge ideological orientation is hardly required to see the appeal of increasing government support for social security, public schools, child care and the like. On two issues, high-awareness and low-awareness Democrats are more than two digits apart, but on average, there is only a 0.04-point difference between the two Democratic groups, and both groups take the liberal position 100 per cent of the time.

Ideological awareness, by contrast, counts for Republicans. Republicans who display low levels of ideological awareness are much farther from their co-partisans than are high-awareness Democrats from their co-partisans. In fact, in many instances, low-awareness Republicans are closer to Democrats than to high-awareness Republicans (for example, on social security and public schools). By contrast, even for the most popular social policies (social security and public schools), ideologically aware Republicans resist increasing spending, and for other nearly as popular policies (unemployment, health care, child care and poverty programs) they decisively support cutting government spending. For Republicans, the mean difference in policy positions between the groups was 0.17 – four times larger than for their Democratic counterparts.

Uphill Issues for Democrats and the Status Quo as a Focal Point

Notwithstanding the Democratic Party’s broad advantage on the social welfare agenda, the Republican Party’s position on some issues is more popular. While it is a smaller number of issues, they are arguably more electorally consequential than most of those we have canvassed – for example welfare, food stamps and government aid to minorities. On these issues, Republicans, whatever their level of ideological awareness, will be of one mind: government spending on each should be cut. But how do Democrats react, and why?

Redistributive policies turn on the status quo: more should be done than is currently being done, or less than is being done or about the same as is being done. Championing more government spending on welfare is a bridge too far for most Democrats, while supporting cuts to welfare equates to defecting. In these circumstances, standing by the program as it is counts as defending it. Thus the status quo serves as a focal point, even for ideologically aware Democrats.

Table 2 reports mean support scores for Democratic and Republican identifiers, again centered on 0.5 and separated by two levels of ideological awareness, on three issues that are easy for Republicans and hard for Democrats.Footnote ³¹ Just so far as Democrats are of one mind whatever their level of ideological awareness on issues that are easy for them, so, too, are Republicans on issues that are easy for them. As the two columns on the right make plain, whatever their level of awareness, Republicans are overwhelmingly in favor of cutting spending for welfare, aid to blacks and food stamps.

Table 2

Spending Preferences on Democrats’ Uphill Issues

Just as the status quo serves as a focal point for social welfare issues that are hard for Democrats, they should coordinate with keeping spending about the same. Table 2 shows that this is, unsurprisingly, true for Democrats who score low on ideological awareness. For welfare, aid to blacks and food stamps, their mean support level, on a 0 to 1 scale, is 0.44. The status quo, perhaps more surprisingly, is also a coordination point for Democrats who score high on ideological awareness (0.47, 0.47 and 0.51 for welfare, aid to blacks and food stamps, respectively).

Uphill Issues for Democrats: Partisan vs. Ideological/Policy Appeals

We have seen that Democrats will support keeping spending about the same for policies like welfare. But if their party wants to rally them in support of a more liberal stand, even if only briefly, how can it do so? Previous research suggests that party-centered – rather than ideology-centered – appeals are one mechanism.Footnote ³² The Ideological vs. Partisan Signals experiment tests this hypothesisFootnote ³³ focusing on whether ‘the federal government should make every effort to ensure that everybody has a good standard of living’ or ‘each individual has a responsibility to get ahead on their own’.Footnote ³⁴ These alternatives, bringing out the tension between the values of policy liberalism and the core value of individual responsibility, present Democrats with a harder policy choice than Republicans. The experiment focuses on the comparative efficacy of ideological and partisan cues. In one condition the question begins: ‘Democrats feel that […] Republicans feel that’; in the other condition, the question begins: ‘Liberals feel that […] Conservatives feel that’.

Table 3 presents mean levels of responsiveness to partisan and ideological signals by party identification and their level of ideological sophistication, this time indexed by whether or not they were sorted. The second and fourth rows of Table 3 illustrate that partisan and ideological signals are synonymous for both non-sorted and sorted Republicans, or close enough to make no substantive difference. Non-sorted Republicans are strongly opposed to the policy regardless of the label, and the arithmetic difference of 0.07 points between the two conditions is statistically insignificant (t(55)=0.79, ns). For sorted Republicans, the label makes no difference whatsoever, with a mean level of support of about 0.09 in both conditions (t(131)=0.12, ns).

Table 3

Ideological vs. Partisan Signals Experiment

Ideological and partisan cues are not synonymous for non-sorted Democrats, though. When policy positions have party labels, their mean score is 0.76; when they have ideological labels, it is only 0.54 (t(110)=2.87, p<0.01). Non-sorted Democrats are markedly more likely to rally to their party’s side in response to party labels than to ideological signals. By contrast, sorted Democrats strongly favor their party’s policy when it is party branded (0.94), and nearly as strongly when it is ideologically branded (0.84; t(130)=1.87, p=0.06). Thus ideologically aware Democrats – indexed by whether they are sortedFootnote ³⁵ – resemble ideologically aware Republicans. The difference between the two parties is a function of the asymmetry between them in ideological awareness.

But it can be argued that these results are not dispositive. Rather, they show reactions to labels. The ‘liberal’ label, these (and many other) results indicate, does not have the same appeal as the ‘conservative’ label. But a label is only a label. Perhaps Democrats appear to rely more on partisan than programmatic appeals, not because they are less responsive than Republicans to the substance of the matter, but because they are not as excited by the wrapping it came in. The label ‘liberal’, after all, is less popular among Democrats as well as Republicans.Footnote ³⁶ What is more, a number of studies have demonstrated that both ideological and party labels can move supporters every which way.Footnote ³⁷ It is perfectly possible, then, that Democrats may be as likely as Republicans to respond to the motivating idea behind social welfare policies.

To put this possibility to the test, the Policy Argument vs. Party Label experiment was carried out on a nationally representative online sample.Footnote ³⁸ The experiment is built around four conditions: (1) policy arguments (but not party labels) are supplied; (2) party labels (but not policy arguments) are provided; (3) both party labels and policy arguments are supplied; and (4) in the control condition, neither party labels nor policy arguments are provided (see the Appendix for item wording). Note that the social welfare issue, food stamps, is one on which Democrats face headwinds.

Table 4 shows the levels of support for increased spending on food stamps in the four experimental conditions: the higher the score, the higher the level of support for coupons that can be exchanged for food; the lower the score, the lower the level of support for food stamps. Again, policy preferences are calculated conditional on whether or not the partisan also shares the party’s ideology. Without partisan labels, non-sorted Democrats tend to be on the wrong side of the issue. When non-sorted Democrats receive the party label, they are slightly in favor of food stamps (0.60 and 0.58 in the parties and arguments condition and parties without arguments condition, respectively). Without the label, they tend to be against them (0.42 and 0.38 in the arguments without parties condition and no parties, no arguments condition, respectively; the difference between the first two conditions and the second two conditions was significant, t(108)=2.37, p<0.05). However, sorted Democrats need either a label or an argument to be in favor of food stamps; otherwise, they are opposed to the policy (the difference between the first three conditions and the last was significant, t(124)=3.08, p<0.01).

Table 4

The Policy Argument vs. Party Label Experiment

The reaction of Republicans could not be more different from that of Democrats. The non-sorted Republican takes a conservative position when he or she receives the argument (0.20 and 0.07 in the parties and arguments and arguments without parties conditions, respectively, t(20)=1.23, ns), but prefers the status quo without an argument (0.50 and 0.45 in the parties without arguments and no parties, no arguments condition, respectively, t(12)=0.22, ns; the difference between the no arguments and arguments conditions was significant, t(34)=2.94, p<0.01). For the sorted Republican, preferences are approximately the same across conditions, ranging from 0.12 in the arguments without parties condition to 0.22 in the parties without arguments condition (the difference between these two conditions was not significant, t(116)=1.64, ns). In a word, for sorted Republicans, party labels are superfluous. So, too, are policy arguments. A tough issue for Democrats is an easy issue for ideologically aware Republicans. They have no need of signals: they have internalized what goes with what, and why.

Competing Hypotheses

Skocpol and others argue that the political constituency willing to pay for policies that benefit particular groups rather than society as a whole is limited in size,Footnote ³⁹ and is almost certain to shrink as programs inevitably fail to fully achieve their ambitiously advertised goals.

The principle of benefits being awarded because they have been earned (rather than because they happen to be needed or wanted) goes to the core of the conservative temperament. One may hypothesize that this principle gives Republicans who identify with their party’s ideological program a heuristic principle with which to oppose redistributive policies. There is a plausible competing hypothesis, however. Conservative positions on social welfare issues characteristically consist of opposition to increased government spending and, often, support for reduced government spending. Conservative positions can thus be fiscally self-serving. Republicans may maintain their bloc opposition to increases to scope-of-government services not for ideological reasons, but out of a concern for their pocketbooks. Self-interest is thus an eminently credible mechanism underpinning opposition to increased government spending for Republicans in particular, since they tend to be better off than Democrats.Footnote ⁴⁰

To arbitrate the merits of these two lines of reasoning – self-interest and policy particularism – the Taxes/Beneficiaries experiment was carried out.Footnote ⁴¹ In September 2008, Knowledge Networks administered an online survey to a nationally representative sample of adults. In all, 60 per cent of the sample completed the interview (n=1,501), yielding an American Association for Public Opinion Research R3 response rate of 7.3 per cent.

In one condition, the introduction took the standard form: ‘Here is a list of some federal programs. For each one, please indicate whether you would like to see spending increased or decreased.’ In the other condition, however, the introduction took this form:

After respondents expressed their positions on the different policies, they were asked who benefits most from each – particular groups, society at large or people like themselves.Footnote ⁴²

Consider first the Self-Interest Hypothesis. The intuition is that support for increased government spending on social welfare policies will decrease when citizens face the prospect of having to pay higher taxes. Table 5 presents levels of support for federal spending, by party, as a function of whether the likelihood of increases in spending entailed higher taxes or not. Positions of federal spending are recorded so that decrease is coded 0, keep the same as 0.5 and increase as 1.

Table 5

The ‘Tax Increase Experiment’

Scrutinizing Table 5, one can see evidence of a retreat from liberal policies when having more means paying more. But the retreat is on the political left, not the right. Democrats give less support, for example, to increased spending for poverty programs and welfare when it means higher taxes (0.09 points, t(444)=2.87, p<0.01, and 0.10 points, t(444)=2.98, p<0.01, less support in the Tax Increase condition than the Control condition, respectively). For these policies, concern for their pocketbooks may not be surprising. What should be surprising, certainly it was surprising to us, is that Democrats also back off their support for popular policies like social security and public schools when spending more means paying more (0.12 points, t(444)=4.29; p<0.01, and 0.11 points, t(444)=4.09; p<0.01, less support in the Tax Increase condition than in the Control condition, respectively). By contrast, Republicans only balk a bit in the face of higher taxes for one program: public schools (0.07 points less support in the Tax Increase condition than in the Control condition, t(426)=1.87, p<0.10).

These details noted, the larger lesson to draw is that pocketbook considerations make a difference only at the margins, even for Democrats. Yes, they are consistently a little less enthusiastic about more spending if it means higher taxes. But the politically consequential point is that, even so, they overwhelmingly throw their weight behind more spending across the whole array of social welfare policies (with the exception only of welfare, which is hardly surprising). Self-interest is not irrelevant. But neither is it politically dispositive.

How, then, does our second hypothesis, Particularism-Universalism, fare? Figure 4 shows levels of support for social welfare policies, for Republicans and Democrats, depending on whether they see a policy as benefiting particular groups or society at large. For each issue, we show the percentage of Republicans and Democrats that sees the issue as particularistic (Panel A) as well as the mean level of support (Panel B). The policies are arrayed from least- to most-often judged to benefit a particular group. Roughly, both Republicans and Democrats are likely to judge a policy as particularistic in inverse proportion to the numbers that benefit from it. For our purposes, though, the pivotal point is that the two parties differ sharply on the likelihood of judging a policy to be particularistic. With the exception only of social security, Republicans consistently see a policy as benefiting a particular group to a larger degree than Democrats. Thus Republicans are roughly twice as likely as Democrats to see public schools as particularistic rather than universalistic (45 per cent vs. 25 per cent). Moreover, with the exception of archetypical universalistic policies (social security and public schools), Republicans are consensually agreed that policies on the social welfare agenda do not benefit society at large.

Fig. 4

A test of the Particularism-Universalism Hypothesis

Consider, then, the policy that poses the highest hurdle for the Particularism hypothesis to clear: social security. More government spending is clearly the modal position. Not surprisingly, the overwhelming majority of Democrats who see it as universalistic strongly support increased spending (0.76). But the comparable majority of Republicans who agree it is universalistic are nearly as supportive (0.69); as, interestingly, are Democrats who see it as particularistic (0.71). Against this background, the position of Republicans who do not see social security as universalistic stands out starkly. They alone resist more spending by maintaining the status quo, if that (0.45).

Social security is the policy for which it is hardest for the Republican Party to garner allegiance – only a minority of its supporters views it as particularistic. For every other policy, a majority (or very close to it) of Republicans views it as particularistic and favors less spending, with the exception of public schools. Child care is an apt example. Three in every four Republicans see it as benefiting a particular group, and they come down clearly in favor of decreased spending (0.32).

All in all, then, the results very mildly support the Self-Interest Hypothesis, but strongly support the Particularism-Universalism conjecture. These results thus identify a particularly significant mechanism that helps Republicans maintain a remarkably unified front against widely appealing policies in the electorate as a whole.