There are many other aspects to social background besides religion, and there are additional aspects to religious background besides denominational attachment. Yet when the focus is Catholicism in American politics, the argument must sooner or later return to denominational attachment as the inescapable backdrop to that politics. There are other aspects to denominational behavior besides these mass attachments, even then – the place of institutional churches as organized interests, for example. But always in the background are the denominations, the great religious families. These denominations shape political values, especially economic and cultural values in this analysis. And those values shape political behavior, both indirectly, by way of the policy preferences that they foster, and directly, in a reflection of denominational membership that is unmediated by policy preferences.
Accordingly, this paper goes in search of partisan distinctions among the five major religious families in American society, here gathered as Catholics, Mainstream Protestants, Evangelical Protestants, Black Evangelicals, and Non-Christians. It asks how membership in these families shapes voting behavior and how the patterns to this behavior shape the strategic landscape of American politics. To that end, the paper takes advantage of new methodological opportunities, operationalized through a rich and ongoing but underutilized dataset. It begins by introducing methods and measures, in a highly condensed fashion but with signposts for those who want more. It moves to the links between denominational attachments and voting behavior in American politics a quarter-century ago. And it compares change in these to the situation in contemporary American society.
Methods, Measures, Definitions, and Data
The methodological approach used here is described in much greater detail in Shafer and Spady, The American Political Landscape.1 Methodological taxonomists might want to think of this method as a semi-parametric evolution in what is generally recognized as “item response theory.” Chapters 2 and 5 of that book should give such professional students what they need to know as an introduction to this approach. Those who prefer to cut immediately to the substantive chase should just know:
that in this approach, two great attitudinal domains, involving public preferences in the realms of economics and culture, are allowed to produce vote distributions stochastically;
that two simple conceptual distinctions in voting behavior, when applied to these vote distributions, produce three kinds of graphic displays that are accessible to nearly anyone who has ever read a topographical map;
and that these maps go a long way toward sketching out the strategic landscape of American politics as it is structured by the five major religious families in American society.
To begin at the beginning, then, the analysis uses a two-dimensional framework, deriving from economic values, involving the preferred distribution of material goods in society, and cultural values, deriving from the preferred distribution of behavioral norms in society. To pursue those two dimensions, we use the Pew Values Surveys, which have the great advantage of being large regular samples of national opinion offering the same 5-item scales for the opening and closing presidential elections of this paper, 1984 and 2008. These ten items are gathered in Table 1.2
Two attitudinal dimensions, analyzed in this way, allow three simple and straightforward graphical ways to present the result. One involves the relationships between economic or cultural values individually and the vote. This is the most common form of voting analysis in the scholarly literature, and it is often the most heuristically useful. The second involves the relationships between economic and cultural values jointly with the vote. This is the way that political values are actually linked to voting behavior in a world where the impact of economics or culture can vary both from person to person and from election to election, but where each always appears in the presence of the other. And the third type of graphic display moves away from relationships, individual or joint, and toward distributions. If relationships are beloved of social scientists, it is demographic and valuational distributions, and most especially the density maps which they permit, that are beloved of campaign practitioners.
Lastly, then, this article goes in search of individual relationships, joint relationships, and density maps for the five major religious families characterizing American society. For purposes of assigning specific individuals to these collective families, we distinguish the latter in the following way3:
“Catholics” combine Roman Catholic and Eastern Orthodox branches of attachment to the Catholic Church. This puts back together the split in the original Christian church and is the conventional way of proceeding in the literature on religion and politics.
“Protestants” are instead further divided into two main branches, sometimes defined as “Pietistic” vs. “Liturgical” but more commonly encountered as “Evangelical” and “Mainstream.” This division is accomplished through an item in the Pew surveys asking about a “born-again” experience: 1. “Evangelical Protestants” are born-again; 2. “Mainstream Protestants” are not.
The category “Non-Christians” gathers Jews, miscellaneous others, and, most especially, those who answer “none” when asked about denominational affiliation. While this is the most theologically heterogeneous category, its individual pieces co-vary not just in their values but also in their voting behavior, while they are otherwise too small individually to offer stable coefficients in the search for the relationship between social background and political values.
Lastly, in common with many other students of religion in American society, we acknowledge the distinctive theology characterizing historically Black churches through a category of “Black Evangelicals.” While this recognition does seem theologically accurate, it presents special problems here. Because there is so little variance in the partisan vote for this fifth denominational family, it cannot be subjected to an analysis of the link between political values and voting behavior,5 though Black Evangelicals will return at the end of the analysis as an important further element of strategic complexity on the American political landscape.
The Denominational Backdrop to an Old Order
Individual Voting Relationships
Figure 1 introduces the individual relationships to the vote for economic and then cultural values in the 1984 election, for the four religious families with sufficient internal variation in policy preferences and partisan choice to permit subsequent analysis. In passing, Figure 1 introduces a further, major, conceptual distinction. Voting in these two elections is always a tripartite, not a bipartite, variable: Republican Vote, Democratic Vote, or Non Vote.6 In this way, the Republican and Democratic vote are not forced to become the obverse of each other, which they rarely are in actual practice. Distinctions between the Republican and Democratic vote can still be strategically critical, yet this approach keeps these distinctions fully available without imposing a form on them which they do not naturally possess.
With economic values, the dominant patterns are straightforward and overwhelming. The Republican vote, for Ronald Reagan in 1984, featured a strong negative relationship to economic values within all four major religious families (Figure 1A). Economic conservatives were strongly attracted, economic liberals were marginally attracted, and support was otherwise neatly aligned across the full ideological spectrum. The Democratic vote, for Walter Mondale in that year, featured a considerably weaker relationship to economic values in the opposite – the positive – direction (Figure 1B). Economic liberals were attracted and economic conservatives repelled, though less impressively than for Republican voters. Three of the four great religious families then featured a weak echo of the Democratic pattern among Non voters, with economic liberals least likely and economic conservatives most likely to turn out, while Non-Christians were characterized by a greater tendency not to vote that was largely invariant to political values (Figure 1C).
Within these dominant ideological relationships, the great religious families added further consistent but far less consequential twists. For the Republican vote, the Evangelicals contributed the largest further increment at every point on the ideological spectrum, followed by the Mainstreams, lagged by the Catholics, and lagged additionally by the Non-Christians, who in that sense contributed the greatest further decrement (Figure 1A). The denominational story of the Democratic vote was then the reverse: the largest overall increment coming from the Non-Christians, then the Catholics, then the Mainstreams, with the Evangelicals contributing the greatest decrement (Figure 1B). Though at the end of the day, when the vote was stratified by economic values, ideology strongly trumped denomination as a principle for aligning the Republican, the Democratic, and even the Non vote (Figure 1C).
With cultural values, voting relationships were likewise straightforward for all three voting categories among all four religious families. Only here, denomination actually trumped ideology for all but a small stretch (the far left) of the ideological spectrum. The Republican vote by cultural values now featured essentially no relationship across the vast bulk of the ideological spectrum, from strong cultural conservatives through moderate cultural liberals, before plunging among strong liberals (Figure 1D). The Democratic vote was likewise invariant to cultural values across most of the ideological spectrum, again from strong conservatives to moderate liberals, before surging among strong cultural liberals (Figure 1E). And the Non vote showed a very modest tendency for cultural conservatives to vote less and cultural liberals to vote more (Figure 1F).
Accordingly, when voting behavior was stratified by cultural values, it was denomination and not ideology that told most of the behavioral story, though this story was additionally different for Republican, Democratic, and Non voters. The Republican vote in effect showed the lingering influence of what would once have been widely recognized as “The Protestant Nation”7 (Figure 1D). The two great Protestant families, the Evangelicals and the Mainstreams, gave the Republican candidate a clear increment at almost every point on the ideological spectrum, while the two great non-Protestant families, Catholics and Non-Christians, were the reverse. From strong conservatives through moderate liberals, there was essentially zero difference in Republican support by cultural ideology but a full 20% voting gap by denominational attachment, Protestants vs. the rest.
The Democratic story was different, however. Just as economic and cultural values could tell different valuational stories, and just as the three voting categories could tell different behavioral stories, the four religious families could tell different denominational stories for different values and different voters – and the Democratic vote of 1984 as stratified by culture was an excellent example (Figure 1E). In the larger but less striking aspect of this example, the Evangelicals, the Mainstreams, and the Non-Christians who voted Democratic did essentially the same as their counterparts among Republican voters through the vast bulk of the ideological spectrum, before surging opposite to those Republican votes among strong cultural liberals. Yet in the other Democratic voting story, more individualized but more striking for its denominational distinction, cultural conservatives among Catholics gave the Democrats an additional increment of support across much of the cultural spectrum – and this increment actually rose as voters became more conservative.
The Non voting story, finally, was different from both the Republican and the Democratic picture. Denomination again trumped ideology at most points on the cultural spectrum, for all four great religious families. That much did not change. Stratified by their cultural preferences, Non-Christians were least likely to vote, Evangelicals were most likely to vote, and the Catholics and Mainstreams clustered roughly in-between. Yet this was a denominational patterning different from that of either Republican or Democratic voters. Only at the far ideological extremes did this pattern wobble, and once again it was really only the Catholics who wobbled, being characterized by a greater propensity to turn out and vote among both their extreme cultural conservatives and their extreme cultural liberals.
Joint Voting Relationships
In some years, it is especially important to remember that these individual relationships to the vote for economic and then cultural values, while very helpful in interpreting the social backdrop to political behavior, are not really the way that this vote occurs. Instead, economic relationships always occur in the presence of cultural values, just as cultural relationships always occur in the presence of economic values. On the other hand, in 1984 and with denominational attachment as the focus, combining the two main ideological dimensions changes the interpretation of the links between political values and voting behavior only modestly. One major relationship that is harder to intuit from individual relationships now stands revealed – or at least underlined – by comparing the four denominations in this joint analytic fashion. This pattern is then further embroidered by some distinctly Catholic idiosyncrasy (Figure 2).
First, consider Figures 2A and 2B, not for their immediate substantive content, but as a guide to reading these joint ideological landscapes. Economics is now the y-axis and culture is now the x-axis. The black lines superimposed on them are voting contours, that is, the share of those located at that point on the ideological landscape within the religious family in question – here the Evangelicals – who vote Republican (Figure 2A) or Democratic (Figure 2B). If this vote were purely dependent on economic values, these contour lines would be horizontal (parallel to the x-axis) and march neatly up and down the page. Conversely, if this vote were purely dependent on cultural values, these contours lines would be vertical (parallel to the y-axis) and march neatly back and forth.
For the Evangelical Protestants, their Republican vote nearly is a pure embodiment of economic preferences. The lines do march neatly up and down the page, with the exception of extreme cultural liberals, where every contour plunges. For these Evangelicals, their Democratic vote in 1984, while likewise predominantly economic, shows more cultural impact as well. The neat economic progression characterizing the Republicans is invaded by much more cultural progression. Recall that this is inherent, if less explicit, in prior individual analyses. We already know that individual relationships to cultural values are roughly similar among Republican and Democratic voters in this year, but that individual relationships to economic values are stronger among Republicans than among Democrats.
Accordingly, economics should be more powerful among Republican voters, while Democratic voters should have comparatively more room for culture in their voting behavior, and this is indeed true among Evangelical Protestants. But the key points here are now two. One is just that we can now see this relationship in its entirety, jointly. The more important point, however, is that this joint relationship does indeed characterize every major denominational family. Which is to say, the Republican vote is more neatly aligned by economic values than the Democratic vote, not just among Evangelicals but also among Mainstreams (Figures 2C and 2D), Catholics, (2E and 2F), and Non-Christians (Figures 2G and 2H). By extension, the Democratic vote leaves more room for cultural alignment than the Republican vote in every great denominational family.
Other aspects of the previous individual voting relationships can still be derived from these joint presentations, as with the fact that the two Protestant families are more Republican than their two non-Protestant counterparts. To see this, merely compare the total territory below the line that offers, say, a 0.50 or better Republican vote for each of the great denominations. What also remains, more idiosyncratically but again more clearly, is those lesser Catholic exceptions to general patterns. Most strikingly, where cultural conservatives desert the Democratic candidate in every other denominational family (Figures 2B, 2D, and 2H), and increasingly so as the voter becomes more conservative, they do not do so among Catholics (Figure 2F). Instead, the Catholic Democratic vote rises not just among strong cultural liberals but among strong cultural conservatives as well, though to a lesser degree. On the other hand, the bulge of additional Democratic support among strong to moderate economic liberals that characterizes the other three religious families is absent among Catholics.
Voter Distributions and Density Maps
Those are the relationships to the vote for economic and cultural values, individually and jointly, when stratified by denominational attachment. They provide the answer to many questions about relationships per se. Was there still evidence of a political divide between Protestants and non-Protestants in 1984? Yes. If it was no longer large, its outlines could still be easily discerned. Did Catholics really offer a bonus to Democratic presidential candidates as late as 1984, quite apart from what their personal policy preferences would dictate? Yes, they did. Though it was restricted to Catholics whose cultural conservatism might otherwise have pulled them toward the Republican alternative.
Answers to those questions, in turn, should have – or at least, could have – offered strategic advice to both Republican and Democratic candidates, though even this was not of the simple, straightforward, “move toward the center” variant that electoral theorists might offer. Thus if a Democratic candidate wanted to please those strong cultural liberals who were so predisposed toward the Democrats, the policy indifference of the culturally conservative Catholics actually made a move away from the center much less costly. Conversely, if a Republican candidate wanted to benefit from the cleavage incipiently distinguishing the “Protestant nation,” that candidate was well advised not to move toward the non-Protestant minority, however “reasonably centrist” that advice might seem.
Yet these relationships, so beloved of social scientists, could also mask important facts about the stratifying influence of denominational attachments. In the process, they could gloss over major strategic challenges for electoral contenders, challenges that were integral to denominational identities. And the way to see this is to return to considering links between economic and cultural values jointly with the vote, but to change the underlying question that is being asked:
The question underlying joint relationships is effectively, “At any given point on the electoral landscape – the strategic world created by mapping economic and cultural values jointly – what share of the sample is choosing a Republican, a Democratic, or a Non vote?”
Yet that question can be reversed; strategic challenges can appear very different when it is; and denominational attachments are a great way to make the point. Now, the questions becomes, “At any given point on the electoral landscape, what share of the Republican (or Democratic, or Non) voters can be found there?”
What results from asking this question is a distribution, allowing the creation of density maps, the analytic tool beloved of campaign practitioners as opposed to social scientists. These often tell a distinctive story, as Figure 3 will indicate. That figure sets out where the four great religious families are located on the ideological landscape for political conflict. Beyond that, it shows where the three voting categories – Republican, Democratic, or Non – are located within each family. Accordingly, each panel in Figure 3 now provides two major pieces of potential strategic information:
The blue and the red lines, for the Republicans and the Democrats, respectively,8 show the areas that are at least 50% more Republican or Democratic than the nation as a whole. These are the overrepresented areas of policy preference within each party coalition. They are in some sense where the party is “headquartered” on the ideological landscape.
The underlying contours then reveal where the social group itself – in this case, the four great religious families as defined by denominational attachment – is over- or underrepresented on the ideological landscape. Both considerations can have major strategic importance, one that is often masked by a focus on statistical relationships between political attitudes and vote probability.
Seen this way, the key fact of voting behavior among the Evangelical Protestants in 1984 was that they were united by culture and divided by economics (Figure 3A). Two lesser facts also stand out immediately. In the less strategically consequential twist, the Republican vote among these Evangelicals was actually more culturally moderate than their Democratic vote. In the more strategically consequential twist, the overrepresented Republican terrain maps almost perfectly onto the high-ground of Evangelical Protestantism in general. As a result, this religious family should have been voting heavily Republican in 1984, and they were indeed the most Republican denominational group, going 56% Reagan, 20% Mondale, and 24% Non. Though keep in mind that the central strategic factor about the group is still that it was united by culture on the conservative end and divided by economics in its partisan choice.
The parallel story in some regards – and the opposite story in others – is provided by the Non-Christians (Figure 3D). On the one hand, they are the other religious family that is most cleanly described as united by culture and divided by economics. On the other hand, both the overrepresented Republican and the overrepresented Democratic populations within this denominational group are found in the territory of strong cultural liberalism. There, economic conservatives and some economic moderates vote disproportionately Republican, while economic liberals and some economic moderates vote disproportionately Democratic. But this time and unlike the situation of the Evangelical Protestants, it is the overrepresented Democrats who reside in the territory of the greatest ideological density. They should thus be the most Democratic in terms of group contributions to national totals, as indeed they were. Even in a Republican landslide, Non-Christians were only 35% Reagan, 32% Mondale, and 33% Non.
Unlike the Evangelical Protestants or the Non-Christians, the Mainstream Protestants in 1984 were distinguished by having their overrepresented Republican and Democratic voters clustered on entirely different valuational dimensions (Figure 3C). Strong economic conservatism distinguished the overrepresented Mainstream Republicans. Strong cultural liberalism distinguished the overrepresented Mainstream Democrats. Those Republican economic conservatives were still moderate cultural liberals in their modal tendency. While those overrepresented Democrats were still true economic moderates, reaching comfortably into moderate economic conservatism. The overrepresented Republicans among these Mainstream Protestants nevertheless encompassed the densest part of the underlying ideological territory for this religious family, which should have made it clearly Republican, as indeed it did: 50% Reagan, 24% Mondale, 26% Non. Though the distinguishing factor for the Mainstream Protestants remained their clustering on entirely different valuational dimensions.
Last but not least, the Catholic electoral landscape was the most ideologically idiosyncratic among the four great religious families (Figure 3B). The overrepresented Republican terrain featured strong to moderate economic conservatives who were moderate cultural liberals. In that sense, Catholic Republicans were not unlike Mainstream Republicans. Yet the Catholic Democratic story was unique among the four great religious families in having two clear – and unconnected – patches of ideological high-ground. The larger of these was found among strong cultural liberals who varied from strong economic liberalism through true economic moderation. These overrepresented Democrats, too, were like their counterparts among the Mainstreams. But Catholic identifiers had a second overrepresented terrain among strong economic liberals who were strong to moderate cultural conservatives. The combination should have made Catholics more difficult to address in policy terms but more charitable toward the Democrats even in this landslide year, as indeed it did: 40% Reagan, 32% Mondale, 28% Non – considerably more Democratic than the Mainstream Protestants.
The Catholics thus stand out for both the internal and external demands that the family as a whole made on strategic thinking. Neither the Mainstream Protestants nor the Non-Christians had any counterpart populations to the lesser Catholic high-ground, although the Evangelical Protestants did. Moreover, this meant that the overrepresented Democrats among Catholics pitted strong cultural conservatives who were strong economic liberals against strong cultural liberals who contained many true economic moderates, a large ideological split. It also meant that the Catholic denominational population, while it featured a clear Republican vs. Democratic divide on economics – Democrats liberal to moderate, Republicans moderately to strongly conservative – featured an effective cultural unity between one faction of overrepresented Democrats and the overrepresented Republicans on culture against another Democratic faction that was more conservative than both on culture.
The result, when these four great religious families are put back together and examined simultaneously, was a much more complex strategic landscape than individual or even joint relationships to the vote would have suggested, an overall strategic landscape of substantial ideological complexity when stratified by denominational attachment. Overrepresented Republicans populations in all four religious families were united by economic conservatism. That was not a challenge. But they were remarkably diverse in their cultural preferences: strongly liberal among the Non-Christians, moderately liberal among Catholics and Mainstreams, moderately to strongly conservative among Evangelicals – where the latter were the religious family potentially most charitable to them.
Overrepresented Democrats were more united on culture, but the exceptions were still important. They were strong cultural liberals among the Non-Christians, Mainstreams, and Catholics, but strong cultural conservatives among the Evangelicals as well as that secondary cluster of Catholics – where both of the latter contributed a densely populated ideological terrain. These Democrats were then highly varied in their economic preferences. They ranged all the way into moderate conservative territory among their Non-Christians and Mainstreams. They ranged from strong liberals to moderates among their Catholics and Evangelicals. But they also had that non-negligible group of very strong economic liberals among their Democrats.
The Denominational Backdrop to a Newer World
Individual Voting Relationships a Quarter-Century Later
A quarter-century later, Figure 4 introduces the individual relationships to the vote for economic and then cultural values in the 2008 election, again stratified by the four great religious families. Seen this way, there are noteworthy changes uniform to all four families. Economic values actually increase their aligning power, but cultural values increase their power so much more that the two domains end up with roughly equal relationships between economic or cultural values and voting behavior. Yet there are also noteworthy changes in the comparative contributions of the four denominational groups, aligning them now in very different ways. Among the denominations, the Evangelical Protestants and the Non-Christians exaggerate their previous behavior, but the Mainstream Protestants and – most notably – the Catholics moderate theirs so extensively as to collapse into each other.
With economic values, the dominant patterns for a modern era are both straightforward and familiar. The Republican vote, for John McCain in 2008, still features a strong negative relationship to economic values within every major religious family (Figure 4A). Strong economic conservatives are strongly attracted, strong economic liberals are minimally attracted, and support is neatly aligned in-between. The Democratic vote, for Barack Obama this time, still features a slightly less pronounced relationship to economic values in the opposite direction, again for every major family (Figure 4B). Economic liberals are clearly attracted, economic conservatives just as clearly not. But this relationship is now much stronger among Democratic voters than it was in 1984, and this change is why economics as a whole became more influential in 2008 than it was in that earlier year. The Non vote, finally, remains essentially unchanged, featuring a very weak echo of the Democratic pattern, except for the Non-Christians, whose non-voting is largely unresponsive to economic preferences of any sort (Figure 4C).
Yet there is a secondary pattern to these relationships that is likewise straightforward but essentially new. Evangelical Protestants still give the Republican candidate an additional increment of support at every point along the ideological continuum, and this increment is much-expanded. Just as Non-Christians still award him an additional decrement at every point, and this decrement is much-expanded, too (Figure 4A). But the Mainstream Protestants and the Catholics have now clumped together, more or less perfectly in-between the Evangelicals and the Non-Christians. And the exact opposite can now be said of Democratic voters: an augmented voting increment at every ideological point among Non-Christians; an augmented voting decrement at every point among Evangelicals; with Mainstreams and Catholics together in the middle and roughly equidistant from both other groups (Figure 4B). Only among the Non voters do these denominational differences essentially vanish.
With cultural values, the ideological patterns for a modern era are likewise straightforward, though almost completely unfamiliar by comparison with an earlier time. The Republican vote, again for John McCain, features a clear but modest negative relationship to cultural values for every denomination (Figure 4D). Conservatives – cultural conservatives here – are more likely to vote Republican than cultural liberals, now at every gradient across the ideological spectrum. The Democratic vote, for Barack Obama this time, is opposite to the Republican vote, with cultural liberals attracted and cultural conservatives repelled, likewise at every gradient along the ideological continuum (Figure 4E). In that sense, the two partisan electorates have achieved the same general form of alignment on culture as they had (and still have) on economics. And the Non vote remains a weak version of the Republican pattern, except for the Evangelicals this time, whose non-voting is largely unresponsive to cultural preferences (Figure 4F).
Yet there is a secondary pattern to these cultural relationships, too, and while this denominational contribution could be summarized crudely in the same manner as with economic relationships, it also differs between Republican and Democratic voters in important ways with cultural values. An element of parallelism shows the Evangelicals again pulling additionally Republican, the Non-Christians again pulling additionally Democratic, and the Mainstreams and Catholics once more uniting in the middle. This is the sense in which the denominations interact in parallel fashion for both economic and cultural values. In the case of the Republicans, however, denominational distinctions on culture are very consequential, such that the distance between Evangelicals and Non-Christians is actually larger than the distance between strong conservatives and strong liberals within those two denominations. Yet in the case of the Democrats, the distance between strong conservatives and strong liberals on culture is much greater than the distance between even the Evangelicals and the Non-Christians. As a result, culture is actually more strongly aligned with the vote among Democrats than among Republicans.
For economic and cultural values, then, two further things can be said when these individual relationships to the vote are examined side by side. First, economic and cultural values have acquired not just a roughly similar potential for aligning the great religious families, but a clearly similar patterning when they do so. Economics did not have to decline in its potential for ideological alignment in order for this to happen, but culture had to increase substantially, which in fact it did, in all four great families. But second, these denominational families both shuffled the way they gave shape to these ideological alignments by way of economic values and extended that same shuffling to cultural values this time. In the process, they appear to have annihilated the old (indeed historic) alignment appropriate to a “Protestant nation” and replaced it with one in which Catholics and Mainstream Protestants were in some important ways indistinguishable.
Joint Voting Relationships a Quarter-Century Later
By now, the next main theoretical point should be more or less automatic. Individual relationships to the vote, for economics and culture by denominational family, set clear limits on the possibility for joint relationships involving both economics and culture simultaneously, but they do not guarantee a simple proportionate transfer. On the one hand, Figure 5, portraying these joint relationships graphically, does suggest that both sets of individual relationships – for liberals vs. conservatives among Republican and Democratic voters in both main issue domains – continue to shine through their joint presentation, that is, with economics in the presence of culture and culture in the presence of economics. Yet what Figure 5 does that individual relationships can never do is to convey their joint – interactive – power.
First and overall, every denominational group, and indeed both partisan factions within every denominational group, still shows the effect of both economic and cultural values in the manner that these appeared when examined one by one. Conservatives are voting disproportionately Republican, liberals are voting disproportionately Democratic, and ideological scores are arrayed neatly between the extremes – now for economic and cultural values jointly and not just individually. Partisan factions of Republican and Democratic voters within each denominational group do add some further distinctions, and denominational groups do differ across their partisan factions in noticeable ways. Yet the first and most basic fact here is that joint consideration of the relationship between economic plus cultural values and the vote does not change the impressions generated by considering these relationships individually.
That said, there remain important residual effects from partisan factions across denominational groups, and slightly larger distinctions from denominational groups across partisan factions. It is these two effects that complete the picture of joint relationships between political values and voting behavior. Though in examining these further denominational contributions, it is important to begin with the major effect that does not any longer distinguish Mainstream Protestants from Catholics. Now, for both these Mainstreams (Figures 5C and 5D) and these Catholics (Figures 5E and 5F), there is a neatly symmetric relationship between economic plus cultural values and the vote, rising for Republicans in the jointly conservative corner of the landscape and falling away in both directions from it, while rising for the Democrats in the opposite, jointly liberal corner of the landscape and falling away in both directions from it as well.
To say the same thing differently: if you imagine the contour line, for example, of the median Catholic Democrat at 0.375, you will see that it must fall almost perfectly on the off-diagonal, the one running from strong economic liberals who are strong cultural conservatives to strong cultural liberals who are strong economic conservatives. And the same can be said for Mainstream Protestant Republicans, Mainstream Protestant Democrats, and Catholic Republicans. This should mean that Mainstreams and Catholics vote nearly identically, and in 2008 they certainly did: 35% Republican and 36% Democratic for Mainstreams, 34% Republican and 38% Democratic for Catholics. That net difference of three percentage points is in some sense a measure of all that is left of the alignment that once embodied “the Protestant nation.”
The Evangelical Protestants are then not quite as symmetric, as between the aligning power of economic and cultural values, because they retain a bit more power for economics among both their Republican and their Democratic voters (Figures 5A and 5B). Especially among Evangelical Republicans, the contour lines do still march perceptibly up and down the y-axis. There is then a further echo of the old pattern in which economics is comparatively more powerful among Republicans and culture comparatively more powerful among Democrats. Yet with all that said, the dominant fact about both partisan factions within this most-conservative religious family – 56% McCain, 20% Obama in 2008 – is just that the impact of both economic and cultural values runs neatly from the same off-diagonal corners of the ideological landscape, as it did for the Mainstreams and the Catholics.
Perhaps not surprisingly, the Non-Christians are slightly less symmetric in their joint voting contours than these Mainstreams and Catholics but in the opposite manner to the Evangelicals (Figures 5G and 5H). In this, the Non-Christians show a bit more aligning power for culture in their voting relationships than any other religious family. Their Democratic voters are the true embodiment of this, being massively dominant within the group: 57% Obama, 13% McCain in 2008. Thus their voting contours are the closest to a true cultural model as any partisan faction within any religious group, marching visibly across the ideological landscape on the x-axis. Though even they are dominated by economic and cultural lines still running essentially from the same off-diagonal corners of the ideological landscape. That this picture looks curiously truncated in the case of Non-Christian Republicans is mainly due to their shortage: this denominational faction is essentially absent above the main off-diagonal.
Density Maps a Quarter-Century Later
Once again, two superficially contradictory facts accompany the shift from a focus on joint relationships between economic plus cultural values and the vote to a focus on the distributions – and density maps – underlying these relationships. From one side, both forms of graphical display are derived not just from the same data, but essentially from the same methodological techniques for analyzing same. But from the other side, the strategic advice that follows from a focus on relationships between political attitudes and vote probability, even joint relationships, may differ sharply – it does not have to differ, but it often does – from a focus on density maps instead. And the density maps for denominational families as stratifying influences on political values and voting behavior remain a great case in point. Once again, these density maps reveal a) where the four great religious families are located on the ideological landscape and b) where the three voting categories are located within each family. For 2008 as for 1984, the result intensifies the strategic choices – and dilemmas – for both political parties.
Viewed this way, the great religious family that is most stable – least changed – across this quarter-century is in fact the Mainstream Protestants (Figure 6C). In 1984, they were distinguished from the three other great families by the fact that their overrepresented Republicans were organized by the economic values of this partisan faction, being diagnostically conservative on economics, while their overrepresented Democrats were organized by the cultural values of this partisan faction, being diagnostically liberal on culture. In 2008, precisely the same thing could still be said. Indeed, the two density maps were almost a perfect overlay (Figures 3C and 6C).
Such a summary does, however, ignore one subtle but important difference, and understanding this difference requires going on to examine the underlying distribution – the colored underlay – for the Mainstream Protestants. In 1984, the overrepresented Republican part of the landscape included the most densely populated areas in Mainstream Protestantism. That should have made them solidly Republican, and at 50% for Reagan and 24% for Mondale, they were. By 2008, however, the overrepresented Republican and Democratic territory among these Mainstreams was now effectively equal. That should have made them closely divided as between the parties, and at 35% McCain and 36% Obama, they now were.
The other religious group that was only marginally changed across this same quarter-century is the other great Protestant family, the Evangelical Protestants (Figure 6A). In 1984, they were distinguished by being united by culture, on the conservative end of the ideological spectrum, and divided by economics, with Republicans conservative and Democrats liberal. Yet now, there was no longer any overrepresented high-ground of economic liberalism among Evangelical Republicans or of economic conservatism among Evangelical Democrats. What remained overwhelmingly the same was that Republicans were located in the densest ideological terrain within Evangelical Protestantism, the Democrats on much more thinly populated terrain. This should have made Evangelicalism the most charitable denomination for the Republican candidate, and with a vote of 56% McCain and 20% Obama, it did.
Change was much more noticeable among Non-Christians, and it was a change confounding the expectations of many who worked with either voting relationships or density maps (Figure 6D). The overrepresented Democratic population actually changed very little. It had been headquartered in 1984 in terrain that featured strong cultural liberalism and economic preferences that ranged from strong liberalism through moderate conservatism. In 2008, it was still headquartered in precisely that territory. Yet the overrepresented Republican population looked different. In 1984, it had been headquartered in terrain that was moderately to strongly liberal on culture and that ranged from truly moderate to strongly conservative on economics. But by 2008, it was instead located on terrain that was moderate on culture and moderately to strongly conservative on economics. In the process, it had left its place within the densest ideological territory among these Non-Christians. That should have made it heavily Democratic in this latter year, and with a vote of Obama 57% and McCain 13%, it was massively so.9
The biggest change, however, came among Catholics, and it appeared to drive the main change in the denominational contribution to the ideological landscape (Figure 6B). The story of partisan factions among the great religious families in 1984 had in fact been most complex among the Catholics. There were overrepresented ideological factions among Catholic Democrats on terrain that was strongly liberal on economics and strongly conservative on cultures; there were overrepresented ideological factions among Catholic Democrats on terrain that was strongly liberal on culture but ranged into true moderation on economics; and there were overrepresented ideological factions among Catholic Republicans on terrain that was moderately to strongly conservative on economics but moderately liberal on culture.
By 2008, that whole story was strikingly different. More than with any other great religious family, partisan Catholics were now headquartered – solidly overrepresented – in the two ideologically consistent corners of the ideological landscape. Overrepresented Democrats were headquartered in territory that was strongly liberal on both economics and culture, and then fanned out in both directions: staying liberal on economics but becoming more conservative on culture, and staying liberal on culture but becoming more conservative on economics. Conversely, overrepresented Republicans were headquartered in territory that was strongly conservative on both economics and culture while fanning out in the same two alternative directions, though this Republican vote remained more concentrated than its Democratic counterpart.
Yet once again, in featuring a different location on the political terrain, Democrats had solidified their hold on the denser part of the overall Catholic landscape, while Republicans had now inhabited a less-dense part of the same landscape. Two larger observations appear to follow. Among denominations, it was change within the Catholic population that was actually most diagnostic of change within the nation as a whole, when the ideological positioning of the two major parties was the focus. In other words, it was the Catholic population that was divided in partisan terms most like the national divide, when examined by way of density maps. Beyond that and because this was also the largest change within denominational families over the preceding quarter-century, it appeared to play a much larger role in the convergence of Catholicism with Mainstream Protestantism than did shifts among those Mainstreams.
Beyond even that, the strategic challenge of the American political landscape, when stratified by denominational attachment, had become even more complex:
The Republicans remained united on economics. The overrepresented Republicans within every denomination family were moderate to strong economic conservatives. But the party was now even more diverse on culture, with every denominational family on different cultural terrain. The modal Evangelical was a strong cultural conservative; the modal Catholic was a moderate cultural conservative; the modal Non-Christian was a true cultural moderate; and the modal Mainstream was a moderate cultural liberal.
The Democrats were far more divided on economics, managing to feature over-represented ideological terrain that ranged from strong economic liberalism to moderate economic conservatism in every great denominational family. They were somewhat less divided on culture, but cultural divisions were still substantial. The Mainstream Protestants, the Non-Christians, and a major part of the Catholics were headquartered on the terrain of strong cultural liberalism. But the Evangelicals were headquartered instead on the territory of strong cultural conservatism, where they were joined by another major piece of the Democratic Catholic population.
And at this point, it is worth noting that reintroduction of the Black Evangelicals would make the strategic landscape even more complex for both parties, though in very different ways. Recall that Black Evangelicals have not appeared in the analysis of political values and voting behavior because their aggregate partisan choice, alone among the five major religious families in American society, has been invariant with respect to their values:
As a result, for the Republicans, adding Black Evangelicals back into the mix does not increase their internal strategic complexity, since the Republican presence among theses Black Evangelicals is effectively exiguous, at roughly 97% Democratic and 3% Republican in 2008. Though what this addition does that is externally consequential is to add a further, reliable, non-negligible increment to Democratic vote totals.
For the Democrats, on the other hand, this increment comes at the price of increased internal complexity. Black Evangelicals hardly lack political values; they are tightly concentrated on the terrain of strong economic liberalism and strong cultural conservatism. They thus add substantially to policy tensions within the Democratic Party, by holding cultural values that are anathema to the overrepresented Democratic factions among Mainstream Protestants and Non-Christians while preferring economic policies that are alien to large parts of these same groups.
Denominational Attachments and American Politics
That is a portrait of the American political landscape as stratified by its major denominational families, and it can be summarized quickly and succinctly. There was an old world of American politics, accessible by the first couple of Pew Values Surveys, in which much of the ideological landscape for political conflict was organized by economic values. Economic conservatives voted Republican, economic liberals voted Democratic, and gradients of conservatism and liberalism in-between were proportionately distributed between the parties. In this world, economic preferences were additionally consequential in structuring the Republican rather than the Democratic vote, although these preferences mattered to both. To the extent that cultural values came into play, their effect was isolated to strong cultural liberals, who deserted the Republicans and flocked to the Democrats. Yet for something like 80% of American society, these cultural preferences were simply unrelated to the vote.
The four major religious families in American society all reflected these national trends within their memberships. None operated evidently opposite to national trends. Yet each of these denominational families not only embodied these national trends in distinctive fashions. Denominational attachments also added shape to the alignment of both economic and cultural values with voting behavior. In this additional shaping influence, it was cultural more than economic values that actually showed the power of denominational attachments to shape the relationship between political values and the vote. Here, the two Protestant families were simply more Republican at every point on the ideological spectrum, and the two non-Protestant families correspondingly more Democratic. A weaker version of the same phenomenon could be seen with economic values, too, though it might not have seemed worth mentioning if it were not stronger – religiously stereotypical in its time – with cultural values.
Yet that still suggests more denominational similarity than the full picture of these great religious families would affirm. For once density maps were added to the analysis – maps showing the differing ideological locations of the four great families, along with the additionally different behavior of their overrepresented Republican and Democratic voters – the picture became much more denominationally differentiated:
Protestant Evangelicals were united by (conservative) culture and divided by economics, where the values dominant within the group were in clearly Republican territory.
Non-Christians were united by (liberal) culture and divided by economics, where the values dominant within the group were in clearly Democratic territory instead.
Mainstream Protestants had their partisan concentrations in entirely different domains, with Republican Mainstreams united by conservative economics and Democratic Mainstreams united by liberal culture, where Republicans were modestly advantaged by the dominant values of the group.
And Catholics were distinguished by pitting three main internal factions against each other, where Democrats were split off from Republicans by economics but two main clusters of Democratic voters were additionally split by culture, and where Democrats still managed to capture more of the dominant values of the denomination.
Flash forward a quarter-century, and there was a new order to American politics. In this, the political landscape was organized simultaneously by both economic and cultural values, across the entire spectrum. Economics had actually gained some aligning power, but culture had gained even more, to the point where the landscape could now be organized symmetrically for both domains across that full ideological range. Indeed, cultural values were now even stronger in organizing the Democratic rather than the Republican electorate, and it was this Democratic gain that had most completely energized the newly consequential cultural divisions, now present everywhere and not just among strong cultural liberals.
Once again, the four major religious families all reflected these – now-different – national trends within their memberships. Despite overall change, none operated evidently opposite to them and none even lagged them in a major way. Yet the direct effect of these denominational families on the link between political values and voting behavior was now strikingly different from its effect a quarter-century before. The Evangelical Protestants had expanded their contribution to the Republican vote at every point on the ideological spectrum for both economics and culture. The Non-Christians had expanded their contribution to the Democratic vote at every point for both issue domains. And the Mainstream Protestants and Catholics had come together at what was pretty much the mid-point between Evangelicals and Non-Christians. In the process, denominational patterns which had once embodied a “Protestant nation” had effectively expired.
Once again, there was even more ideological and partisan heterogeneity among the four great religious families once density maps were added to the analysis:
Mainstream Protestants had changed the least between 1984 and 2008 – they could thus hardly be the drivers for the coming together of Mainstreams and Catholics – though the dominant political values characterizing the denomination had shifted slightly so as to neutralize their previous Republican advantage.
Non-Christians looked more like these Mainstream Protestants than they once had, being organized by conservative economics among their Republicans and liberal culture among their Democrats. Though they had shifted additionally in no longer being united by (strongly liberal) cultural values, a shift which actually saw Non-Christian Republicans move away from the dominant values of the group.
Protestant Evangelicals, by contrast, remained united by (strongly conservative) cultural values, while becoming increasingly divided by economics. The old territory of substantial overlap between overrepresented Republicans and Democrats in their midst had simultaneously evaporated, and this time the new distributions were much more advantageous to the Republicans.
And Catholics had moved the farthest of all in the location – and conformation – of their overrepresented partisans. They were now almost the stereotypical embodiment of national partisan conflict, with a Republican high-ground in the territory of strong economic and cultural conservatism and a Democratic high-ground in the territory of strong economic and cultural liberalism instead. In the process, what had once been the beacon for the dissident side of a Protestant nation had become the leading battleground for that nation as a whole.
Byron E. Shafer and Richard H. Spady, The American Political Landscape (Cambridge: Harvard University Press, forthcoming February 3, 2014).
Pew Research Center for the People and the Press, “American Values Survey,” multiple years. The survey containing votes for presidential 2012 became available too late to be developed for use in this article.
These denominational distinctions are derived most directly from, and can be explored in detail within, Geoffrey Layman, The Great Divide: Religious and Cultural Conflict in American Party Politics (New York: Columbia University Press, 2001).
This item referred to drug dealers prior to 9/11, to terrorists post 9/11.
These Black Evangelicals constituted 3.5% of the 1984 sample and were 68% Dem, 3% Rep, and 29% Non-voting. They were then 5% of the 2008 sample and were 78% Dem, 2% Rep, and 20% Non-voting.
In elections with a serious third-party or independent vote for president, it is of course a quadripartite variable.
See, among many, Sanford Levinson, Constitutional Faith (Princeton: Princeton University Press, 1988) and Mark Noll, The Protestant Origins of American Political Thought (Princeton: Princeton University Press, 1994).
In this, we use the traditional color codings for the right and the left, blue for parties of the right, red for parties of the left.
Note that a partisan faction can change its location on the political landscape over time either by changing its political values or by changing its partisan vote, or of course through some combination thereof.
About the article
Byron E. Shafer
Byron E. Shafer is Hawkins Chair of Political Science at the University of Wisconsin, Madison.
Richard H. Spady
Richard H. Spady is Research Professor of Economics at the Johns Hopkins University and Senior Research Fellow of Nuffield College, Oxford. They are joint authors of The American Political Landscape, forthcoming from Harvard University Press in early 2014.
Published Online: 2014-02-08
Published in Print: 2013-12-01