Abstract

Though predominantly Christian since the time of its founding, the United States has become more religiously diverse in recent decades. Yet since the mid-1990s, the proportion of Americans who see their country as a “Christian nation” has reportedly increased. Though initially paradoxical, these trends are less mysterious if the idea of a “Christian America” (CA) is understood, not as a description of religious demography, but as a discursive practice that seeks to align the symbolic boundaries of national belonging with the boundaries of the dominant faith community. Using data from the 1996 and 2004 General Social Survey, it is shown that the growing prevalence of CA was restricted to Americans of Christian faith, thereby widening an existing religious divide over the meaning of American identity.

INTRODUCTION

In the first decade of the twenty-first century, Christians remain the largest religious group in the United States, as they have been since the country's founding. Hence it is not surprising that most Americans tend to think of their country as a “Christian nation.” According to a 2006 survey by the Pew Research Center, two-thirds of American adults say they “consider the United States a Christian nation” (Pew Research Center 2006:5), while nearly a third of respondents in another study strongly agree that, “In the twenty-first century, the United States is still basically a Christian society” (Wuthnow 2005:200).

In demographic and cultural respects, of course, America is no longer as “Christian” as it used to be. Over the last few decades, the proportion of Americans professing Christian faith, especially of the Protestant confession, has experienced a steady decline. Although Christians still account for about 78 percent of U.S. adults as a whole, the percentage of Protestants has fallen from over 60 percent in the early 1990s to around 50 percent as of 2006 (see also Kosmin and Keysar 2009; Pew Forum 2008; Smith and Kim 2005). Meanwhile, the ranks of the religiously unaffiliated have grown from fewer than 10 percent to between 14 and 16 percent over the same period (Hout and Fischer 2002; Pew Forum 2008).

Yet paradoxically, the growth of religious diversity has not produced a concomitant decline in the view of America as a Christian country. On the contrary, such views have generally grown more prevalent among U.S. adults since the turn of the century. As of the mid-1990s, only 60 percent of Pew respondents said they regarded their country as a “Christian nation.” By 2002, agreement with this view had risen to 67 percent and reached a peak of 71 percent in 2005 before settling to 67 percent the following year (Pew Research Center 2006:5).

Why is America increasingly perceived as a Christian nation, even though Christians have not been growing as a proportion of the American population? If the notion of a “Christian America” is not just a reflection of demographic realities, what could account for its varying appeal over time? In this article, we suggest that such questions become more tractable when Christian America discourse is viewed through the lens of recent work on symbolic boundary construction (Edgell et al. 2006; Kunovich 2006; Lamont and Molnár 2002; Wimmer 2008; Zimmer 2003). Rather than merely describing the demographic status quo, statements like “America is a Christian nation” represent a discursive practice that seeks to align the boundaries of authentic national belonging with adherence to the dominant religious faith.

As we go on to argue, practices of symbolic boundary alignment are always a double-edged sword, simultaneously designating certain groups as prototypical of the larger community while relegating others to the symbolic margins. For example, by conditioning recognition as an authentic American on adherence to Christian faith, the idea of a Christian America tacitly reinforces the moral prestige of the religious majority, even as it presents Americans of other faiths, or with no formal religion, with invisible barriers to symbolic inclusion. Consequently, we anticipate that such ideas will appeal primarily to members of the religious ingroup and are likely to be contested by outgroups at risk of marginalization. In other words, we expect Christian America discourse to be divisive along religious lines, rather than a source of consensus and solidarity transcending religious boundaries.

To the extent that Christian America discourse appeals mainly to U.S. Christians, it follows that most of its aggregate variation over time should be due to fluctuations within the Christian community. Moreover, because Christians are no more numerous today, relative to non-Christians, than in previous decades, the growing prevalence of Christian America beliefs most likely reflects an increase in the importance of Christianity to American identity among Christian Americans themselves. Precisely why this should have occurred when it did is a question we take up in the course of our theoretical and empirical analysis. Whatever its causes, however, the recent surge in these beliefs likely signals a growing divergence in attitudes between American Christians and their non-Christian compatriots. To the extent that such polarization can present an obstacle to political consensus on major issues (DiMaggio et al. 1996), a widening religious divide over the meaning of American identity could become a source of future social conflict.

The rest of this article is structured as follows. In the next section, we outline the contours of a boundary-oriented approach for explicating the link between religious identity and conceptions of national belonging. In the section thereafter, we develop a series of hypotheses which link the intensity of Christian America beliefs to characteristics and attitudes of individuals, as well as identifying a number of mechanisms to account for their changing patterns over time. In the analytic sections of the paper, finally, we put our hypotheses to the test using data from the 1996 and 2004 General Social Surveys (GSS).

THEORIZING RELIGIOUS BOUNDARIES OF AMERICAN IDENTITY

Our theoretical approach is informed by recent research on the role of symbolic boundaries in the construction of social identities (Edgell et al. 2006; Kunovich 2006; Smith et al. 1998; Wimmer 2008; Zimmer 2003; for a general discussion see Lamont and Molnár 2002). As Lamont and Molnár (2002) define them, “Symbolic boundaries are conceptual distinctions made by social actors to categorize objects, people, practices, and even time and space. They are tools by which individuals and groups struggle over and come to agree upon [or contest] definitions of reality” (Lamont and Molnár 2002:168). A key function of symbolic boundaries is to “separate people into groups and generate feelings of similarity and group membership” (Lamont and Molnár 2002:168). As Edgell et al. (2006:231) have shown, “Symbolic boundaries are effective … in promoting a sense of solidarity and identity by virtue of imagining an ‘other’ who does not share the core characteristics imagined to be held by those who are legitimate participants in the moral order; the imagined community must have outsiders as well as insiders.”

In constructing symbolic boundaries, social actors select from a variety of culturally available symbolic resources (Lamont and Molnár 2002; Zimmer 2003), or boundary criteria. Thus, “religious identity often overlaps with national identity because [religion] provides one means of distinguishing ingroups from outgroups” (Kunovich 2006:438). Yet defining one type of social identity (national belonging) in terms of another (religious affiliation) is not simply a matter of boundary overlap. Rather, it involves a practice we call symbolic boundary alignment, or the “nesting” of one symbolic boundary within another. For example, the statement that “America is a Christian nation” not only posits an intersection between religious and national boundaries; it also implies that the boundary between Christians and non-Christians helps regulate the threshold between more and less “prototypical” Americans (on prototypicality, see Mummendey and Wenzel 1999).

Because symbolic boundaries are always moral boundaries (Edgell et al. 2006; Lamont 2000), aligning national with religious boundaries is likely to have very different consequences for religious ingroups and outgroups within the national community. Among members of the ingroup, the notion of a Christian America may, at times, serve as a rhetorical weapon in the arsenal of religious activists seeking to mobilize supporters around social or political issues of the day (Regnerus and Smith 1998; Schwadel 2005). At other times, what sounds like the battle cry of cultural warriors may only be intended to strengthen religious identification and solidarity among coreligionists. For example, Smith et al. (2000) argue that much evangelical discourse about “reclaiming” America “is simply talk to construct and maintain collective identity. This evangelical rhetoric is functioning not so much to actually get the troops ready to re-Christianize America as to express and reinforce a distinctive identity for its adherents” (Smith et al. 2000:56). Thus, we do not assume that the appeal of Christian America beliefs will always rise and fall in tandem with religious politicization (Schwadel 2005; Servin-Gonzales and Torres-Reyna 1999) and deprivatization (Regnerus and Smith 1998; Wilcox and Goldberg 2002). The idea of a Christian America does not, by itself, imply that Christianity should be imposed on non-Christians, or that Christians should enjoy special rights and privileges as a matter of public policy (see also Wuthnow 2005:200). Nor does it necessarily entail the belief that religion should play a more prominent role in politics. In principle, a Christian who sees her/his faith as central to American identity could still regard her/his religious beliefs and practices as a personal matter.

Yet, whatever its symbolic function for self-described insiders, any talk of a Christian America is likely to have a dispiriting impact on those it implicitly marginalizes. Asked to comment on a conference dedicated to “Reclaiming America for Christ,” one Jewish American responded that, “When the language becomes exclusively Christian, Jewish groups become at best ambivalent, at worst hostile” (Smith et al. 2000:22). Such fears are not always misplaced. Even if their immediate effects are largely “imagined,” symbolic boundaries can also have material consequences, serving as “an essential medium through which people acquire status and monopolize resources” (Lamont and Molnár 2002:168). By attributing contrasting degrees of social prestige to insiders and outsiders, symbolic boundaries can confer differential access to material benefits and other advantages. “As a result of these benefits, group members limit membership to protect their rewards” (Kunovich 2006:440). Indeed, if taken to extremes, symbolic boundaries can have even direr implications, as when restrictive definitions of nationhood serve as a pretext for depriving marginalized citizens of their civil rights or denying citizenship to outsiders on the basis of race, religion, or national origin (Somers 2006; Zolberg 2006).

HYPOTHESES

Drawing on our theory of symbolic boundary alignment, our next goal is to develop specific hypotheses for predicting the salience of Christianity to national identity in the American context. Two sets of hypotheses are proposed. First, we attempt to specify the groups in society for whom the appeal of “Christian America” beliefs (or “CA”) should be strongest, as well as some of the factors that should further enhance its appeal among potential supporters. Second, with these hypotheses in mind we consider some of the mechanisms that could help account for changes in CA over time.

Predictors of CA

Most Americans are strongly attached to their country, exhibiting high levels of pride in citizenship and national symbols (Smith and Kim 2006). However, any restrictive definition of national identity is likely to be controversial, even divisive. Because the alignment of national identity with internal symbolic boundaries will have different consequences for ingroups and outgroups, we believe that affinities with CA will differ significantly between dominant and non-dominant religious groups. In general, we concur with Kunovich (2006) that “religious majorities are more likely to state that the dominant religion is very important for national identity,” while “religious minorities are likely to downplay the dominant religion as a relevant factor for national identity because their own religious background excludes them from membership in the nation” (Kunovich 2006:440). Hence, we expect that the salience of Christianity to American identity will be more pronounced within the Christian majority than the non-Christian minority (viz. believers of other faiths and the religiously nonaffiliated).

Hypothesis 1: CA will be higher among Christian Americans than among non-Christians

Among U.S. Christians, we expect to find significant variation between different denominations and confessions (Steensland et al. 2000:292). One way that denominational heritage could motivate religious understandings of national identity is by encouraging adherents to think of themselves as prototypical representatives of their faith tradition. For example, Smith and colleagues (1998) argue that evangelical Protestants are especially prone to regard themselves as prototypical Christians. Thus, “when evangelicals use the word ‘Christian,’ most likely they are instinctively meaning evangelical Christian” (Smith et al. 1998:124, original italics; cf., Smith et al. 2000:200; Wuthnow 2005:200). CA could also have disproportionate appeal for Christians in traditionally marginalized social categories, perhaps offering a means of compensating for their historical exclusion. Such a dynamic might be at work in black Protestant denominations, in which racial and religious identities are tightly interwoven (Pattillo-McCoy 1998; Steensland et al. 2000). We therefore expect that CA beliefs will have greater appeal to (white) evangelicals and black Protestants than to mainline Protestants, Roman Catholics, and other Christian groups.

Hypothesis 2: Among U.S. Christians, CA will be higher among evangelicals and black Protestants than other Christians

Independently of denomination or confession, we further expect that the salience of religion to national identity will increase with the importance of either religious or national identity. As Hoge and De Zulueta (1985) have argued, “the importance of any one component [of identity] relative to others” should predict “whether the component in question is likely to influence others” (Hoge and De Zulueta 1985:22). In previous research, religiosity has been measured by the frequency of attendance at religious services (Kunovich 2006), while closeness to one's country is a good indicator of national identification (Huddy and Khatib 2007). We therefore predict that:

Hypothesis 3: Among U.S. Christians, CA should increase with (a) frequency of religious attendance and (b) feelings of closeness to America

According to recent studies, religious identities often become more salient in a religiously diverse environment (Kunovich 2006; Smith et al. 1998; Wilcox and Goldbert 2002). For example, Christian Smith and colleagues argue that solidarity among evangelical Christians grows stronger with feelings of embattlement in the face of religious pluralism (Smith, et al 1998, 2000), while Wilcox and Goldberg (2002:371) suggest that highly committed Christians “may see growing diversity as threatening their religious hegemony and as undermining the nation's status as a ‘Christian nation.’ ” Since immigration is a major factor in the growth of religious diversity (Smith and Kim 2005), feelings of embattlement and outgroup hostility should be manifested in preferences for reducing the current rate of immigration (Bail 2008; Esses et al. 2001).1 Hence:

Hypothesis 4: Among U.S. Christians, CA will increase with preferences for restricting the rate of immigration to the United States

Accounting for change over time

From these propositions, we can derive a number of mechanisms that could affect changes in the overall prevalence of CA beliefs over time. Our main focus is on two types of mechanisms: (1) gradual shifts in the religious composition of society (compositional effects) and (2) major events that affect the salience of religious and national boundaries (period effects). As we will see, there are good reasons to doubt that demographic changes since the 1990s could have directly produced a concomitant increase in CA in the population. We therefore expect that any aggregate rise in CA will have occurred in response to major events—a reaction, moreover, that should be especially pronounced among members of the Christian majority.

Compositional effects

In light of Hypothesis 1, it is indeed paradoxical that belief in America as a “Christian nation” would have grown more prevalent during a period when the proportion of Christians in the population was on the decline. If CA is typically higher among U.S. Christians than among non-Christians, then a decrease in the percentage of Christians should cause CA to become less prevalent in the population, since its adherents have grown proportionately less numerous. By the same token, Hypothesis 2 implies that the overall prevalence of CA should vary with the ratio of Christians in certain denominations (especially evangelicals or black Protestants), other things being equal. According to the most recent wave of the American Religious Identification Survey, there was some growth in the share of Christians belonging to evangelical denominations or non-denominational congregations between 1990 and 2008, mostly at the expense of mainline Protestant churches (Kosmin and Keysar 2009:5). However, the rate of growth in these groups does not appear to have outpaced that of non-Christian religions or of Americans who do not identify with any religious tradition. Whereas the U.S. population is 30 percent larger today than in 1990, there are 50 percent more non-Christians and 138 percent more “nones” (Kosmin and Keysar 2009:4; on the latter category, see also Hout and Fischer 2002).

On balance, then, we do not believe that changes in the religious makeup of American society would be sufficient to account for the growing prevalence of CA beliefs since the 1990s. Indeed, it seems more likely the net effect of compositional change has been to diminish the rate of growth in CA in recent years.

Hypothesis 5: Changes in the religious composition of American society have favored the attenuation of CA in the population

Period effects

Independently of demographic trends, religious conceptions of American identity could have experienced a net surge if major events of the day have brought religious and national issues to the forefront of public attention. For example, Huntington (2004:340) asserted that, in the wake of religiously motivated attacks on their homeland, Americans increasingly define themselves in religious and cultural terms: “September 11 dramatically symbolized … the beginning of a new era in which people define themselves primarily in terms of culture and religion … . For Americans the religious component of their identity takes on new relevance in this environment.”

Recent studies confirm that beliefs about the role of religion in public life often fluctuate in response to historic events. For example, Regnerus and Smith (1998) argue that religious deprivatization tends to occur in cycles, becoming especially pronounced in times of significant crisis. Throughout U.S. history, “disturbing shifts in the moral order … [have] prompted a revisiting of traditional or commonsense strategies of action by the organizational sources of political and religious meaning systems ….” During these “unsettled” times, religious convictions play an increasingly prominent role in the public sphere, before receding again into “taken-for-granted ways of life” during more “settled” periods (Regnerus and Smith 1998:1351; cf., Swidler 1986).

Such phenomena are commonly known as period effects, or societal reactions to sudden changes of political climate during a particular interval (Alwin and McCammon 2003; Weil 1987). In the case of “pure” period effects, major events will produce a more or less uniform increase (or decrease) in the prevalence of particular beliefs in society as a whole. Unfortunately, it is often difficult to determine whether any particular set of events is directly responsible for pure period effects, since we cannot rule out other developments that may have occurred during the same interval. Certain kinds of events, however, have more distinctive consequences, affecting different segments of the population in contrasting ways. An instructive example is what Regnerus and Smith (1998) call selective deprivatization (see also Casanova 1994). In the wake of the 1960s, they argue, conservative evangelicals came to supplant socially liberal Protestants as the most vocal advocates for the public role of religion, thereby reversing a longer trend of privatization among religious conservatives (Regnerus and Smith 1998; cf., Smith et al. 2000). By extension, we will speak of selective intensification whenever historical developments cause particular attitudes to rise only in certain groups, while declining or remaining stable in others.

In the case of CA, we doubt that the impact of recent history will have been uniform. Although events like the attacks of 9 of 11 or the invasion of Iraq could have enhanced the salience of religious and national issues for Americans in general, Hypothesis 1 leads us to expect that the relevance of Christianity to American identity should intensify primarily within the Christian majority. Hypotheses 3 and 4 provide additional support for this expectation. If recent events have caused CA to intensify among U.S. Christians, they may have done so, in part, by elevating the level of religiosity, patriotism, and xenophobia in the population. Religious minorities, on the other hand, are unlikely to respond by recognizing the dominant religion as a hallmark of the true American, since this would reinforce their symbolic marginality within the imagined community (cf., Kunovich 2006).

Hypothesis 6: “Unsettling” events since the mid-1990s have enhanced CA among U.S. Christians

Finally, we expect that period events which trigger selective intensification on the part of Christian Americans will also have heterogeneous effects on different categories of Christians. In general, we expect CA beliefs to intensify the most among those who were most inclined to embrace them in the past. Specifically, we predict that any increase in CA should have occurred disproportionately among evangelicals and black Protestants (H2), Christians with frequent religious attendance (H3a), Christians with high levels of patriotism (H3b), Christians with more negative attitudes toward national outgroups (H4). We refer to such patterns as (selective) intensification “at the top.”

Hypothesis 6a: Selective intensification will occur faster among those categories of Christians where CA was previously higher

DATASET AND VARIABLES

Sample

Our analysis draws on two waves of GSS data collected by the National Opinion Research Center and obtained electronically from the Inter-University Consortium for Political and Social Research (ICPSR 04295). The GSS has conducted face-to-face interviews with a representative sample of American adults every year or two since 1972 (Davis et al. 2007). In 1996 and 2004, items from the 1995 and 2003 national identity modules of the International Social Survey Project were included in the GSS questionnaire for about 2,500 cases.2

Dependent Variable

In 1996 and 2004, the GSS questionnaires included a battery of items prefaced by the following question: “Some people say the following things are important for being truly American. Others say they are not important. How important do you think each of the following is?” Responses for each boundary criterion were coded as “very important” (1), “fairly important” (2), “not very important” (3), or “not important at all” (4). We measure the salience of Christianity to American identity (CA) using the variable AMCHRSTN (“To be a Christian”), which we reverse coded to rise with level of importance. AMCHRSTN closely resembles items commonly used to measure what we call boundary alignment, or the salience of particular boundary criteria for symbolic belonging in a specified ingroup (Bail 2008; Citrin et al. 2001; Edgell et al. 2006; Jones and Smith 2001; Kunovich 2006). We generally expect this variable to behave in similar ways to other measures of CA beliefs (e.g., Pew Research Center 2006; Smith et al. 2000; Wuthnow 2005). An advantage of the AMCHRSTN item over alternative items (e.g., “Is America a ‘Christian nation’ ”) is the specification that being a Christian makes one more “truly American,” thereby highlighting the normative implications of symbolic boundary construction.

Independent Variables

Religious affiliation

In order to capture distinctions among clusters of Protestant denominations, as well as between these and other religious categories, our analysis employs an aggregation method developed by Steensland et al. (2000), which combines information in the RELIG, DENOM, and OTHER variables into a single polytomous measure of religious tradition. The resulting variable (RELTRAD) codes self-reported religious affiliation as evangelical Protestant, mainline Protestant, black Protestant, Roman Catholic, Jewish, other, or nonaffiliated. However, about 40 percent of respondents in the “other” category are self-identified Christians (36 percent Protestant, 4 percent Orthodox). Because we are often interested in contrasting Christians and non-Christians, we have moved respondents who are coded as “other” on RELTRAD, but as Protestant or Orthodox on the RELIG variable, into a separate category of “Other Christian.” The remaining “others” (comprising adherents of non-Christian traditions) are combined with Jewish respondents and labeled as “Other Faith.” Our modified RELTRAD variable thus codes religious affiliation as: evangelical Protestant (1), mainline Protestant (2), black Protestant (3), Roman Catholic (4), Other Christian (5), Jewish/Other Faith (6), or nonaffiliated (7).

Other independent variables

Additional predictors of CA for Christians include religious commitment, national attachment, and immigration preferences. In our multivariate models, we also include controls for birth cohort, education, sex, region of residence, and race.3 To adjust for non-response bias, household sampling design, and sampling differences between survey years, all of our results were generated using a weighting variable (WT7204) provided by the National Opinion Research Center.

WHO THINKS BEING A CHRISTIAN MAKES ONE MORE TRULY AMERICAN?

Descriptive Analysis

We begin our empirical investigation by examining the distribution of CA beliefs at the outset of our period of study. As a provisional test of our first two hypotheses, we cross-tabulate the bivariate relationship between CA and religious preference using our 1996 GSS sample (table 1). In the sample as a whole, we find that 53 percent of respondents regarded being a Christian as either fairly (15 percent) or very important (38 percent) for being truly American. As expected, however, this majority consensus masks a sharp divergence of opinion between Christians and non-Christians. Among Christians, nearly 62 percent believed that being Christian is either “very important” (44 percent) or “fairly important” (18 percent) for being truly American, whereas nearly 66 percent of non-Christians felt that Christian faith was “not at all important,” and almost one in five (19 percent) thought it was “not very important.”

TABLE 1 How important is being a Christian for being truly American? (percent) N Not at all Not very Fairly Very Total All respondents 1,283 25.2 21.4 15.4 38.0 100.0 Christian 1,049 16.0 21.9 17.6 44.4 100.0 Evangelical Prot. 346 6.6 19.1 16.8 57.5 100.0 Mainline Protestant 247 21.1 22.3 20.6 36.0 100.0 Black Protestant 111 5.4 13.5 12.6 68.5 100.0 Roman Catholic 318 24.5 28.0 17.3 30.2 100.0 Other Christian 26 34.6 19.2 26.9 19.2 Non-Christian 224 66.2 19.2 5.1 9.4 100.0 Jewish/other faith 84 66.7 20.2 2.4 10.7 100.0 Nonaffiliated 148 66.2 18.9 6.8 8.1 100.0 How important is being a Christian for being truly American? (percent) N Not at all Not very Fairly Very Total All respondents 1,283 25.2 21.4 15.4 38.0 100.0 Christian 1,049 16.0 21.9 17.6 44.4 100.0 Evangelical Prot. 346 6.6 19.1 16.8 57.5 100.0 Mainline Protestant 247 21.1 22.3 20.6 36.0 100.0 Black Protestant 111 5.4 13.5 12.6 68.5 100.0 Roman Catholic 318 24.5 28.0 17.3 30.2 100.0 Other Christian 26 34.6 19.2 26.9 19.2 Non-Christian 224 66.2 19.2 5.1 9.4 100.0 Jewish/other faith 84 66.7 20.2 2.4 10.7 100.0 Nonaffiliated 148 66.2 18.9 6.8 8.1 100.0 View Large

TABLE 1 How important is being a Christian for being truly American? (percent) N Not at all Not very Fairly Very Total All respondents 1,283 25.2 21.4 15.4 38.0 100.0 Christian 1,049 16.0 21.9 17.6 44.4 100.0 Evangelical Prot. 346 6.6 19.1 16.8 57.5 100.0 Mainline Protestant 247 21.1 22.3 20.6 36.0 100.0 Black Protestant 111 5.4 13.5 12.6 68.5 100.0 Roman Catholic 318 24.5 28.0 17.3 30.2 100.0 Other Christian 26 34.6 19.2 26.9 19.2 Non-Christian 224 66.2 19.2 5.1 9.4 100.0 Jewish/other faith 84 66.7 20.2 2.4 10.7 100.0 Nonaffiliated 148 66.2 18.9 6.8 8.1 100.0 How important is being a Christian for being truly American? (percent) N Not at all Not very Fairly Very Total All respondents 1,283 25.2 21.4 15.4 38.0 100.0 Christian 1,049 16.0 21.9 17.6 44.4 100.0 Evangelical Prot. 346 6.6 19.1 16.8 57.5 100.0 Mainline Protestant 247 21.1 22.3 20.6 36.0 100.0 Black Protestant 111 5.4 13.5 12.6 68.5 100.0 Roman Catholic 318 24.5 28.0 17.3 30.2 100.0 Other Christian 26 34.6 19.2 26.9 19.2 Non-Christian 224 66.2 19.2 5.1 9.4 100.0 Jewish/other faith 84 66.7 20.2 2.4 10.7 100.0 Nonaffiliated 148 66.2 18.9 6.8 8.1 100.0 View Large

CA was also higher among evangelical Protestants than most other Christian groups, although it was even more pronounced among black Protestants. Fifty-eight percent of evangelicals and 69 percent of black Protestants regarded being a Christian as “very important” for being truly American, while 36 percent of mainline Protestants, and 30 percent of Roman Catholics held the same view. The remaining Christians (chiefly quasi-Protestant denominations and sects) were more ambivalent. Only 19 percent thought of their faith as essential to American identity, while just over a third felt the opposite way.

Multivariate Analysis

Analytic strategy

To ensure that these religious differences are not due to other sources of variation among religious groups, we use a multivariate procedure to estimate the effect of religious affiliation on CA while controlling for our other independent variables. Because AMCHRSTN is highly skewed (with “very important” as the modal category), we have dichotomized our dependent variable to highlight the threshold between “very important” and all other responses. We use binary logistic regression (Long 1997) to estimate the net effects of the independent variables on the log-odds that being a Christian is regarded as “very important” for being truly American. For ease of interpretation, our discussion focuses on the exponentiated regression coefficients (exp[β]).

Results

The first model in table 2 tests whether religious affiliation (Christians vs. non-Christians) significantly influences CA in the 1996 sample. As expected (H1), Christians are considerably more likely to regard being a Christian as “very important” for being truly American. If fact, the odds of such a response are more than three and a half times as great for this group than for non-Christians, controlling for other differences between them. The test statistic (ΔX2) indicates that the model fits the data significantly better than a comparison model without religious affiliation.

TABLE 2 All respondents (N = 1,005) Christian respondents (N = 829) Model 1 Model 2 Model 3 Model 4 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Christian 1.307 0.296 3.694*** – – – – – – – – – (Mainline) Evangelical – – – 0.916 0.193 2.499*** 0.749 0.203 2.116*** 0.871 0.218 2.398*** Black Protestant – – – 1.417 0.290 4.123*** 1.335 0.304 3.796*** 0.654 0.474 1.923 Roman Catholic – – – −0.091 0.201 0.913 −0.122 0.209 0.894 0.196 0.227 1.217 Other Christian – – – −1.409 0.734 0.244† −1.552 0.751 0.212* −1.023 0.760 0.359 Religious attendance 0.550 0.085 1.733*** – – – 0.513 0.084 1.671*** 0.466 0.090 1.594*** National attachment 0.358 0.163 1.431* – – – 0.597 0.160 1.817*** 0.425 0.171 1.529* Immigration Reduce some 0.022 0.194 1.022 – – – 0.072 0.193 1.074 −0.036 0.205 0.965 Reduce a lot 0.577 0.185 1.780** – – – 0.611 0.183 1.842*** 0.475 0.194 1.607* Education −0.409 0.087 0.665*** – – – – – – −0.329 0.090 0.720*** Cohort (<1940) 1940–59 −0.595 0.203 0.551** – – – – – – −0.654 0.214 0.520** ≥1960 −1.197 0.209 0.302*** – – – – – – −1.279 0.224 0.278*** Female 0.083 0.153 1.086 – – – – – – 0.109 0.162 1.115 South 0.647 0.159 1.909*** – – – – – – 0.502 0.172 1.652** Race (white) Black 1.061 0.252 2.891*** – – – – – – 0.874 0.413 2.396* Other race 1.307 0.296 3.694*** – – – – – – 0.606 0.356 1.833† Constant −2.091 0.373 0.124*** −0.701 0.151 0.496*** −1.909 0.243 0.148*** −1.023 0.319 0.360*** −2LL X2 1053.581 1075.422 1009.378 944.366 ΔX2 23.711*** 68.509*** 66.045*** 65.012*** df 1 4 4 7 All respondents (N = 1,005) Christian respondents (N = 829) Model 1 Model 2 Model 3 Model 4 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Christian 1.307 0.296 3.694*** – – – – – – – – – (Mainline) Evangelical – – – 0.916 0.193 2.499*** 0.749 0.203 2.116*** 0.871 0.218 2.398*** Black Protestant – – – 1.417 0.290 4.123*** 1.335 0.304 3.796*** 0.654 0.474 1.923 Roman Catholic – – – −0.091 0.201 0.913 −0.122 0.209 0.894 0.196 0.227 1.217 Other Christian – – – −1.409 0.734 0.244† −1.552 0.751 0.212* −1.023 0.760 0.359 Religious attendance 0.550 0.085 1.733*** – – – 0.513 0.084 1.671*** 0.466 0.090 1.594*** National attachment 0.358 0.163 1.431* – – – 0.597 0.160 1.817*** 0.425 0.171 1.529* Immigration Reduce some 0.022 0.194 1.022 – – – 0.072 0.193 1.074 −0.036 0.205 0.965 Reduce a lot 0.577 0.185 1.780** – – – 0.611 0.183 1.842*** 0.475 0.194 1.607* Education −0.409 0.087 0.665*** – – – – – – −0.329 0.090 0.720*** Cohort (<1940) 1940–59 −0.595 0.203 0.551** – – – – – – −0.654 0.214 0.520** ≥1960 −1.197 0.209 0.302*** – – – – – – −1.279 0.224 0.278*** Female 0.083 0.153 1.086 – – – – – – 0.109 0.162 1.115 South 0.647 0.159 1.909*** – – – – – – 0.502 0.172 1.652** Race (white) Black 1.061 0.252 2.891*** – – – – – – 0.874 0.413 2.396* Other race 1.307 0.296 3.694*** – – – – – – 0.606 0.356 1.833† Constant −2.091 0.373 0.124*** −0.701 0.151 0.496*** −1.909 0.243 0.148*** −1.023 0.319 0.360*** −2LL X2 1053.581 1075.422 1009.378 944.366 ΔX2 23.711*** 68.509*** 66.045*** 65.012*** df 1 4 4 7 View Large

TABLE 2 All respondents (N = 1,005) Christian respondents (N = 829) Model 1 Model 2 Model 3 Model 4 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Christian 1.307 0.296 3.694*** – – – – – – – – – (Mainline) Evangelical – – – 0.916 0.193 2.499*** 0.749 0.203 2.116*** 0.871 0.218 2.398*** Black Protestant – – – 1.417 0.290 4.123*** 1.335 0.304 3.796*** 0.654 0.474 1.923 Roman Catholic – – – −0.091 0.201 0.913 −0.122 0.209 0.894 0.196 0.227 1.217 Other Christian – – – −1.409 0.734 0.244† −1.552 0.751 0.212* −1.023 0.760 0.359 Religious attendance 0.550 0.085 1.733*** – – – 0.513 0.084 1.671*** 0.466 0.090 1.594*** National attachment 0.358 0.163 1.431* – – – 0.597 0.160 1.817*** 0.425 0.171 1.529* Immigration Reduce some 0.022 0.194 1.022 – – – 0.072 0.193 1.074 −0.036 0.205 0.965 Reduce a lot 0.577 0.185 1.780** – – – 0.611 0.183 1.842*** 0.475 0.194 1.607* Education −0.409 0.087 0.665*** – – – – – – −0.329 0.090 0.720*** Cohort (<1940) 1940–59 −0.595 0.203 0.551** – – – – – – −0.654 0.214 0.520** ≥1960 −1.197 0.209 0.302*** – – – – – – −1.279 0.224 0.278*** Female 0.083 0.153 1.086 – – – – – – 0.109 0.162 1.115 South 0.647 0.159 1.909*** – – – – – – 0.502 0.172 1.652** Race (white) Black 1.061 0.252 2.891*** – – – – – – 0.874 0.413 2.396* Other race 1.307 0.296 3.694*** – – – – – – 0.606 0.356 1.833† Constant −2.091 0.373 0.124*** −0.701 0.151 0.496*** −1.909 0.243 0.148*** −1.023 0.319 0.360*** −2LL X2 1053.581 1075.422 1009.378 944.366 ΔX2 23.711*** 68.509*** 66.045*** 65.012*** df 1 4 4 7 All respondents (N = 1,005) Christian respondents (N = 829) Model 1 Model 2 Model 3 Model 4 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Christian 1.307 0.296 3.694*** – – – – – – – – – (Mainline) Evangelical – – – 0.916 0.193 2.499*** 0.749 0.203 2.116*** 0.871 0.218 2.398*** Black Protestant – – – 1.417 0.290 4.123*** 1.335 0.304 3.796*** 0.654 0.474 1.923 Roman Catholic – – – −0.091 0.201 0.913 −0.122 0.209 0.894 0.196 0.227 1.217 Other Christian – – – −1.409 0.734 0.244† −1.552 0.751 0.212* −1.023 0.760 0.359 Religious attendance 0.550 0.085 1.733*** – – – 0.513 0.084 1.671*** 0.466 0.090 1.594*** National attachment 0.358 0.163 1.431* – – – 0.597 0.160 1.817*** 0.425 0.171 1.529* Immigration Reduce some 0.022 0.194 1.022 – – – 0.072 0.193 1.074 −0.036 0.205 0.965 Reduce a lot 0.577 0.185 1.780** – – – 0.611 0.183 1.842*** 0.475 0.194 1.607* Education −0.409 0.087 0.665*** – – – – – – −0.329 0.090 0.720*** Cohort (<1940) 1940–59 −0.595 0.203 0.551** – – – – – – −0.654 0.214 0.520** ≥1960 −1.197 0.209 0.302*** – – – – – – −1.279 0.224 0.278*** Female 0.083 0.153 1.086 – – – – – – 0.109 0.162 1.115 South 0.647 0.159 1.909*** – – – – – – 0.502 0.172 1.652** Race (white) Black 1.061 0.252 2.891*** – – – – – – 0.874 0.413 2.396* Other race 1.307 0.296 3.694*** – – – – – – 0.606 0.356 1.833† Constant −2.091 0.373 0.124*** −0.701 0.151 0.496*** −1.909 0.243 0.148*** −1.023 0.319 0.360*** −2LL X2 1053.581 1075.422 1009.378 944.366 ΔX2 23.711*** 68.509*** 66.045*** 65.012*** df 1 4 4 7 View Large

Models 2–4 test whether CA varies significantly by denomination, religious attendance, national attachment, and outgroup attitudes among Christian respondents in the sample (H2–H4). As expected (H2), the appeal of CA is quite a bit stronger among evangelicals and especially black Protestants compared to mainline Protestants (Model 2), though it is substantially weaker among smaller Christian sects and denominations. Independently of denominational differences, Christians who attend religious services more frequently (H3a), display very strong national attachment (H3b), or favor a large reduction in immigration (H4) are significantly more likely to see their faith as essential to American identity (Model 3). When the control variables are added (Model 4), most of these effects remain statistically significant and similar in size.4

THE GROWING SALIENCE OF CHRISTIAN FAITH TO AMERICAN IDENTITY

Descriptive Results

Having confirmed that most of our hypothesized predictors significantly influence CA in the expected direction, we turn to the question of whether there has been any significant increase in CA during our period of study and, if so, whether the patterns of change are consistent with the kinds of mechanisms our theory envisions. We begin by cross-tabulating our dichotomous measure of CA by survey year. As a preliminary test of compositional change (H5) and selective intensification (H6, H6a), we examine the distribution of respondents by religious affiliation, as well as the rates of change in CA by religious affiliation, over time (table 3).

TABLE 3 All Respondents All years 1996 2004 Δ 1996–2004 LR X2 Δ CA (percent very important) 42.9 37.5 48.9 +11.4 33.404*** Religious affiliation (percent) Christian 79.9 81.1 78.5 −2.6 2.664 Non-Christian 20.1 18.9 21.5 +2.6 100.0 100.0 100.0 0.0 Evangelical Protestant 26.4 26.7 26.2 −0.5 26.367*** Mainline Protestant 16.7 18.8 14.3 −4.5 Black Protestant 7.8 8.7 6.8 −1.9 Roman Catholic 25.8 24.9 26.9 2.0 Other Christian 3.1 2.0 4.3 2.3 Jewish/other faith 7.1 7.3 6.8 −0.5 Nonaffiliated 13.0 11.6 14.7 +3.1 100.0 100.0 100.0 0.0 CA by religious affiliation (percent) Christian 51.2 44.4 59.3 +14.9 42.313*** Non-Christian 10.1 9.4 10.7 +1.3 0.204 Evangelical Protestant 62.7 57.5 68.7 +11.2 8.039** Mainline Protestant 41.6 36.0 50.3 +14.3 8.087** Black Protestant 73.0 68.5 79.5 +11.0 2.883† Roman Catholic 39.4 30.2 49.0 +18.8 23.120*** Other Christian 47.9 19.2 63.8 +44.6 14.107*** Jewish/other faith 9.5 10.6 8.2 −2.4 0.259 Nonaffiliated 10.0 8.1 11.9 +3.8 1.211 Sample size (N) 2,486 1,311 1,175 All Respondents All years 1996 2004 Δ 1996–2004 LR X2 Δ CA (percent very important) 42.9 37.5 48.9 +11.4 33.404*** Religious affiliation (percent) Christian 79.9 81.1 78.5 −2.6 2.664 Non-Christian 20.1 18.9 21.5 +2.6 100.0 100.0 100.0 0.0 Evangelical Protestant 26.4 26.7 26.2 −0.5 26.367*** Mainline Protestant 16.7 18.8 14.3 −4.5 Black Protestant 7.8 8.7 6.8 −1.9 Roman Catholic 25.8 24.9 26.9 2.0 Other Christian 3.1 2.0 4.3 2.3 Jewish/other faith 7.1 7.3 6.8 −0.5 Nonaffiliated 13.0 11.6 14.7 +3.1 100.0 100.0 100.0 0.0 CA by religious affiliation (percent) Christian 51.2 44.4 59.3 +14.9 42.313*** Non-Christian 10.1 9.4 10.7 +1.3 0.204 Evangelical Protestant 62.7 57.5 68.7 +11.2 8.039** Mainline Protestant 41.6 36.0 50.3 +14.3 8.087** Black Protestant 73.0 68.5 79.5 +11.0 2.883† Roman Catholic 39.4 30.2 49.0 +18.8 23.120*** Other Christian 47.9 19.2 63.8 +44.6 14.107*** Jewish/other faith 9.5 10.6 8.2 −2.4 0.259 Nonaffiliated 10.0 8.1 11.9 +3.8 1.211 Sample size (N) 2,486 1,311 1,175 View Large

TABLE 3 All Respondents All years 1996 2004 Δ 1996–2004 LR X2 Δ CA (percent very important) 42.9 37.5 48.9 +11.4 33.404*** Religious affiliation (percent) Christian 79.9 81.1 78.5 −2.6 2.664 Non-Christian 20.1 18.9 21.5 +2.6 100.0 100.0 100.0 0.0 Evangelical Protestant 26.4 26.7 26.2 −0.5 26.367*** Mainline Protestant 16.7 18.8 14.3 −4.5 Black Protestant 7.8 8.7 6.8 −1.9 Roman Catholic 25.8 24.9 26.9 2.0 Other Christian 3.1 2.0 4.3 2.3 Jewish/other faith 7.1 7.3 6.8 −0.5 Nonaffiliated 13.0 11.6 14.7 +3.1 100.0 100.0 100.0 0.0 CA by religious affiliation (percent) Christian 51.2 44.4 59.3 +14.9 42.313*** Non-Christian 10.1 9.4 10.7 +1.3 0.204 Evangelical Protestant 62.7 57.5 68.7 +11.2 8.039** Mainline Protestant 41.6 36.0 50.3 +14.3 8.087** Black Protestant 73.0 68.5 79.5 +11.0 2.883† Roman Catholic 39.4 30.2 49.0 +18.8 23.120*** Other Christian 47.9 19.2 63.8 +44.6 14.107*** Jewish/other faith 9.5 10.6 8.2 −2.4 0.259 Nonaffiliated 10.0 8.1 11.9 +3.8 1.211 Sample size (N) 2,486 1,311 1,175 All Respondents All years 1996 2004 Δ 1996–2004 LR X2 Δ CA (percent very important) 42.9 37.5 48.9 +11.4 33.404*** Religious affiliation (percent) Christian 79.9 81.1 78.5 −2.6 2.664 Non-Christian 20.1 18.9 21.5 +2.6 100.0 100.0 100.0 0.0 Evangelical Protestant 26.4 26.7 26.2 −0.5 26.367*** Mainline Protestant 16.7 18.8 14.3 −4.5 Black Protestant 7.8 8.7 6.8 −1.9 Roman Catholic 25.8 24.9 26.9 2.0 Other Christian 3.1 2.0 4.3 2.3 Jewish/other faith 7.1 7.3 6.8 −0.5 Nonaffiliated 13.0 11.6 14.7 +3.1 100.0 100.0 100.0 0.0 CA by religious affiliation (percent) Christian 51.2 44.4 59.3 +14.9 42.313*** Non-Christian 10.1 9.4 10.7 +1.3 0.204 Evangelical Protestant 62.7 57.5 68.7 +11.2 8.039** Mainline Protestant 41.6 36.0 50.3 +14.3 8.087** Black Protestant 73.0 68.5 79.5 +11.0 2.883† Roman Catholic 39.4 30.2 49.0 +18.8 23.120*** Other Christian 47.9 19.2 63.8 +44.6 14.107*** Jewish/other faith 9.5 10.6 8.2 −2.4 0.259 Nonaffiliated 10.0 8.1 11.9 +3.8 1.211 Sample size (N) 2,486 1,311 1,175 View Large

The results in the first line confirm that the salience of Christianity to American identity has significantly increased. Between 1996 and 2004, the percentage of “very important” responses rose by more than 11 percentage points, from about 38 percent to almost 50 percent. As we suspected, the overall rise in CA was accompanied by a slight decrease in the proportion of Christians (by about 3 percentage points) between survey years. Although Roman Catholics and Other Christians grew by about 2 percentage points each, their gains were offset by a 5-point loss on the part of mainline Protestants, from 19 percent to just 14 percent. Nor was their any proportionate growth in the two bastions of belief in a CA. Whereas evangelicals were equally represented at both time points, black Protestants suffered a 2-point loss. Among non-Christians, Americans with no religious affiliation expanded their relative numbers by 3 percentage points, from 12 percent to 15 percent, while adherents of non-Christian faiths held steady (for a more detailed analysis of these trends, see Kosmin and Keysar 2009; Pew Forum 2008).

Given the gradual diminution of Christians (and the lack of proportionate growth in denominational bastions), it is unlikely compositional change was responsible for the observed increase in CA between survey years (H5). A more likely explanation, we hypothesized, is that the rise in CA represents a type of period effect, or societal reaction to “unsettling” events. Specifically, we predicted that major events which focus attention on matters of religion and nationality should lead adherents of the dominant religion to accentuate the importance of their faith as a component of their national identity (H6). As we see in the last set of results, the rise in CA was indeed concentrated in the Christian majority. Whereas the percentage of “very important” responses increased by an average of 15 points in the Christian category, there was no significant change for the non-Christian groups.

In substantive terms, the fact that CA intensified where it was already the most pronounced means that the existing divide between Christians and non-Christians grew even wider during our period of study. The breakdown by subgroup confirms this impression. While there was no significant change among religious “nones” and adherents of other faiths, there was a significant increase in CA for nearly all categories of Christians, though it is greater in some categories than in others. Before drawing firm conclusions from these results, however, we must first determine whether they remain significant when other variables are taken into account.

Multivariate Analysis

For a more rigorous assessment of composition effects versus selective intensification, we turn again to multivariate logistic regression. For this purpose, we have combined the 1996 and 2004 surveys into a single sample and added a variable for GSS year to estimate the amount and direction of change in CA between surveys (table 4). The first model estimates the “effect” of survey year on our dependent variable. (We include sex, region, and race in this model because their distribution is the same in both surveys.) The exponentiated coefficient for survey year (1.712) in this model confirms that that the odds of a “very important” response were significantly higher, by about 70 percent, in 2004 than in 1996.

TABLE 4 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Year: 2004 0.537 0.095 1.712*** 0.799 0.116 2.224*** −0.110 0.388 0.896 Christian – – – 1.744 0.207 5.722*** 1.160 0.295 3.189*** Attendance – – – 0.704 0.062 2.021*** 0.704 0.062 2.023*** Attachment – – – 0.430 0.115 1.538*** 0.431 0.115 1.539*** Immigration Reduce some – – – 0.073 0.134 1.076 0.075 0.134 1.078 Reduce a lot – – – 0.479 0.137 1.614*** 0.487 0.137 1.628*** Education – – – −0.367 0.060 0.693*** −0.368 0.060 0.692*** Cohort (<1940) 1940–59 – – – −0.648 0.161 0.523*** −0.651 0.161 0.521*** ≥1960 – – – −1.095 0.162 0.335*** −1.107 0.162 0.331*** Female 0.274 0.095 1.315** 0.184 0.110 1.201† 0.180 0.110 1.198 South 0.743 0.100 2.103*** 0.471 0.116 1.769*** 0.590 0.116 1.803*** Race (white) Black 0.920 0.155 2.508*** 1.073 0.184 2.923*** 1.071 0.184 2.919*** Other −0.900 0.218 0.406*** −0.543 0.249 0.581* −0.547 0.249 0.578* Christian × Year – – – – – – 0.987 0.404 2.683* Constant −1.016 0.093 0.362*** −2.791 0.279 0.061*** −2.245 0.337 0.106*** −2LL X2 2555.824 2032.524 2026.714 ΔX2 184.895*** 523.300*** 5.810* df 5 8 1 Nagelkerke R2 0.118 0.399 0.402 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Year: 2004 0.537 0.095 1.712*** 0.799 0.116 2.224*** −0.110 0.388 0.896 Christian – – – 1.744 0.207 5.722*** 1.160 0.295 3.189*** Attendance – – – 0.704 0.062 2.021*** 0.704 0.062 2.023*** Attachment – – – 0.430 0.115 1.538*** 0.431 0.115 1.539*** Immigration Reduce some – – – 0.073 0.134 1.076 0.075 0.134 1.078 Reduce a lot – – – 0.479 0.137 1.614*** 0.487 0.137 1.628*** Education – – – −0.367 0.060 0.693*** −0.368 0.060 0.692*** Cohort (<1940) 1940–59 – – – −0.648 0.161 0.523*** −0.651 0.161 0.521*** ≥1960 – – – −1.095 0.162 0.335*** −1.107 0.162 0.331*** Female 0.274 0.095 1.315** 0.184 0.110 1.201† 0.180 0.110 1.198 South 0.743 0.100 2.103*** 0.471 0.116 1.769*** 0.590 0.116 1.803*** Race (white) Black 0.920 0.155 2.508*** 1.073 0.184 2.923*** 1.071 0.184 2.919*** Other −0.900 0.218 0.406*** −0.543 0.249 0.581* −0.547 0.249 0.578* Christian × Year – – – – – – 0.987 0.404 2.683* Constant −1.016 0.093 0.362*** −2.791 0.279 0.061*** −2.245 0.337 0.106*** −2LL X2 2555.824 2032.524 2026.714 ΔX2 184.895*** 523.300*** 5.810* df 5 8 1 Nagelkerke R2 0.118 0.399 0.402 View Large

TABLE 4 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Year: 2004 0.537 0.095 1.712*** 0.799 0.116 2.224*** −0.110 0.388 0.896 Christian – – – 1.744 0.207 5.722*** 1.160 0.295 3.189*** Attendance – – – 0.704 0.062 2.021*** 0.704 0.062 2.023*** Attachment – – – 0.430 0.115 1.538*** 0.431 0.115 1.539*** Immigration Reduce some – – – 0.073 0.134 1.076 0.075 0.134 1.078 Reduce a lot – – – 0.479 0.137 1.614*** 0.487 0.137 1.628*** Education – – – −0.367 0.060 0.693*** −0.368 0.060 0.692*** Cohort (<1940) 1940–59 – – – −0.648 0.161 0.523*** −0.651 0.161 0.521*** ≥1960 – – – −1.095 0.162 0.335*** −1.107 0.162 0.331*** Female 0.274 0.095 1.315** 0.184 0.110 1.201† 0.180 0.110 1.198 South 0.743 0.100 2.103*** 0.471 0.116 1.769*** 0.590 0.116 1.803*** Race (white) Black 0.920 0.155 2.508*** 1.073 0.184 2.923*** 1.071 0.184 2.919*** Other −0.900 0.218 0.406*** −0.543 0.249 0.581* −0.547 0.249 0.578* Christian × Year – – – – – – 0.987 0.404 2.683* Constant −1.016 0.093 0.362*** −2.791 0.279 0.061*** −2.245 0.337 0.106*** −2LL X2 2555.824 2032.524 2026.714 ΔX2 184.895*** 523.300*** 5.810* df 5 8 1 Nagelkerke R2 0.118 0.399 0.402 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Year: 2004 0.537 0.095 1.712*** 0.799 0.116 2.224*** −0.110 0.388 0.896 Christian – – – 1.744 0.207 5.722*** 1.160 0.295 3.189*** Attendance – – – 0.704 0.062 2.021*** 0.704 0.062 2.023*** Attachment – – – 0.430 0.115 1.538*** 0.431 0.115 1.539*** Immigration Reduce some – – – 0.073 0.134 1.076 0.075 0.134 1.078 Reduce a lot – – – 0.479 0.137 1.614*** 0.487 0.137 1.628*** Education – – – −0.367 0.060 0.693*** −0.368 0.060 0.692*** Cohort (<1940) 1940–59 – – – −0.648 0.161 0.523*** −0.651 0.161 0.521*** ≥1960 – – – −1.095 0.162 0.335*** −1.107 0.162 0.331*** Female 0.274 0.095 1.315** 0.184 0.110 1.201† 0.180 0.110 1.198 South 0.743 0.100 2.103*** 0.471 0.116 1.769*** 0.590 0.116 1.803*** Race (white) Black 0.920 0.155 2.508*** 1.073 0.184 2.923*** 1.071 0.184 2.919*** Other −0.900 0.218 0.406*** −0.543 0.249 0.581* −0.547 0.249 0.578* Christian × Year – – – – – – 0.987 0.404 2.683* Constant −1.016 0.093 0.362*** −2.791 0.279 0.061*** −2.245 0.337 0.106*** −2LL X2 2555.824 2032.524 2026.714 ΔX2 184.895*** 523.300*** 5.810* df 5 8 1 Nagelkerke R2 0.118 0.399 0.402 View Large

Compositional change

The next model adds main effects for the remaining independent variables to see if distributional changes in these predictors as a whole help account for the rise in CA between survey years. As noted above, the modest decline in the ratio of Christians to non-Christians would be expected to favor a slight decrease in the prevalence of CA over time (H5). In addition, two of our control variables that had strong negative effects in table 2 seem likely to have changed in ways that would exert downward pressure on the growth of CA. In particular, the appeal of CA was considerably weaker among better-educated respondents (Regnerus and Smith 1998; Schwadel 2005) and respondents belonging to more recent birth cohorts (Edgell et al. 2006; Kunovich 2006). Over time, the rising level of educational attainment and gradual replacement of older by younger cohorts should have tended to suppress the rate of increase in CA.5 On the other hand, we confirmed (in table 2) that religious attendance, national attachment, and restrictive immigration preferences all had significant positive effects on CA at the beginning of our study. If these have been on the rise since then (H6), then they might help explain some of the increase in CA over time, perhaps even compensating for suppressing influence of demographic trends. On inspecting the results in Model 2, however, it appears that the impact of any positive trends has been overshadowed by countervailing winds of compositional change. Here, the estimated effect of survey year (2.224) is even more positive than in the previous model, once all the independent variables are included. This suggests that the rise in CA might have been even sharper if there had been no changes in the predictor variables.6

Selective intensification

Having confirmed that the observed rise in CA was indeed significant, even though formidable demographic trends should have favored a net decline, we now wish to see whether selective intensification on the part of the religious majority was sufficiently powerful to overcome the attenuating effects of compositional change. In the final model, we add a multiplicative term for the interaction between religious affiliation (Christian/non-Christian) and survey year. If selective intensification explains the overall increase in CA over time, then the coefficient for the interaction term should be significant and positive, while the main effect of survey year should be non-significant. On inspecting the results in the last column, we see that both expectations are confirmed. Net of compositional change, the increase in odds of a “very important” response was 2.7 times greater for Christians than for non-Christians. Because the coefficient for survey year is not significant, we can infer that there was no discernable net change in CA among non-Christians, corroborating our earlier inference that the existing gap between Christians and non-Christians has grown even larger over time.

PATTERNS OF CHANGE WITHIN THE CHRISTIAN COMMUNITY

As our findings up to this point make clear, the idea of a Christian America is an issue about which Christians and non-Christians remain deeply divided, and all the more so in times of international and religious conflict. But how do these “unsettling” times affect previous differences within the Christian community itself? Does the rising belief in a Christian America promote greater consensus around religious conceptions of national identity, or does it only widen existing divisions between various segments of the Christian community? To answer these questions, we conclude our empirical analysis by examining the patterns of change in CA among different groups of Christians. Although we know that CA was higher among Christians, on average, at the end of our period than at the beginning, Hypothesis 6a predicted that the rate of intensification should be significantly greater in those categories of Christians where CA was already stronger. Thus, we expected that CA would increase more among evangelicals and black Protestants than among mainline Protestants and other Christian groups, as well as among Christians who attend service frequently, display strong national attachment, and advocate greater restrictions on immigration.

As before, we use logistic regression to estimate the expected change in odds that Christianity is regarded as very important for being truly American, holding other factors constant. To see whether the change in odds differed significantly between categories of Christians, we add multiplicative terms representing the interaction between survey year and the other independent variables of interest. (For ease of interpretation, the independent variables with hypothesized interaction effects are coded as dichotomous.) To ensure that positive coefficients for each interaction term represent positive changes in the dependent variable over time, categories of each independent variable with the smallest observed increase in CA are used as reference categories. The results are presented in table 5.

TABLE 5 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Survey year 0.714 0.104 2.041*** 0.900 0.121 2.460*** 0.243 0.288 1.275 Religion (Evangelical) Mainline – – – −0.568 0.161 0.567*** −0.868 0.215 0.420*** Black Protestant – – – −0.054 0.336 0.947 −0.212 0.385 0.809 Catholic – – – −0.488 0.144 0.614*** −0.607 0.199 0.545** Other – – – −0.832 0.313 0.435** −2.008 0.748 0.134** Attendance: frequent – – – 1.231 0.132 3.424*** 0.789 0.179 2.201*** Attachment: weak – – – −0.412 0.119 0.662*** −0.424 0.168 0.654* Immigration: liberal – – – −0.331 0.126 0.718** −0.397 0.164 0.672* Education – – – −0.288 0.061 0.750 −0.300 0.062 0.741*** Cohort (<1940) 1940–59 – – – −0.734 0.169 0.480*** −0.744 0.171 0.475*** ≥1960 – – – −1.180 0.172 0.307*** −1.210 0.174 0.298** Female 0.189 0.104 1.208† 0.215 0.115 1.240† 0.222 0.116 1.248† South 0.731 0.110 2.076*** 0.543 0.125 1.721*** 0.520 0.126 1.682*** Race (white) Black 0.810 0.172 2.249*** 0.895 0.282 2.447** 0.925 0.286 2.521** Other −0.640 0.239 0.527** −0.412 0.261 0.662 −0.349 0.264 0.706 Interactions: Religion (Evangelical) Mainline – – – – – – 0.674 0.321 1.962* Black Potestant – – – – – – 0.273 0.475 1.313 Catholic – – – – – – 0.267 0.281 1.307 Other – – – – – – 1.682 0.843 5.376* Attendance: frequent – – – – – – 0.994 0.272 2.703*** Attachment: weak – – – – – – 0.021 0.236 1.021 Immigration: liberal – – – – – – 0.137 0.252 1.147 Constant −0.741 0.101 0.476*** 0.804 0.212 2.234*** 1.144 0.248 3.139*** −2LLX2 2110.320 1847.794 1825.681 ΔX2 145.118*** 262.526*** 22.113** df 5 10 7 Nagelkerke R2 0.114 0.295 0.310 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Survey year 0.714 0.104 2.041*** 0.900 0.121 2.460*** 0.243 0.288 1.275 Religion (Evangelical) Mainline – – – −0.568 0.161 0.567*** −0.868 0.215 0.420*** Black Protestant – – – −0.054 0.336 0.947 −0.212 0.385 0.809 Catholic – – – −0.488 0.144 0.614*** −0.607 0.199 0.545** Other – – – −0.832 0.313 0.435** −2.008 0.748 0.134** Attendance: frequent – – – 1.231 0.132 3.424*** 0.789 0.179 2.201*** Attachment: weak – – – −0.412 0.119 0.662*** −0.424 0.168 0.654* Immigration: liberal – – – −0.331 0.126 0.718** −0.397 0.164 0.672* Education – – – −0.288 0.061 0.750 −0.300 0.062 0.741*** Cohort (<1940) 1940–59 – – – −0.734 0.169 0.480*** −0.744 0.171 0.475*** ≥1960 – – – −1.180 0.172 0.307*** −1.210 0.174 0.298** Female 0.189 0.104 1.208† 0.215 0.115 1.240† 0.222 0.116 1.248† South 0.731 0.110 2.076*** 0.543 0.125 1.721*** 0.520 0.126 1.682*** Race (white) Black 0.810 0.172 2.249*** 0.895 0.282 2.447** 0.925 0.286 2.521** Other −0.640 0.239 0.527** −0.412 0.261 0.662 −0.349 0.264 0.706 Interactions: Religion (Evangelical) Mainline – – – – – – 0.674 0.321 1.962* Black Potestant – – – – – – 0.273 0.475 1.313 Catholic – – – – – – 0.267 0.281 1.307 Other – – – – – – 1.682 0.843 5.376* Attendance: frequent – – – – – – 0.994 0.272 2.703*** Attachment: weak – – – – – – 0.021 0.236 1.021 Immigration: liberal – – – – – – 0.137 0.252 1.147 Constant −0.741 0.101 0.476*** 0.804 0.212 2.234*** 1.144 0.248 3.139*** −2LLX2 2110.320 1847.794 1825.681 ΔX2 145.118*** 262.526*** 22.113** df 5 10 7 Nagelkerke R2 0.114 0.295 0.310 View Large

TABLE 5 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Survey year 0.714 0.104 2.041*** 0.900 0.121 2.460*** 0.243 0.288 1.275 Religion (Evangelical) Mainline – – – −0.568 0.161 0.567*** −0.868 0.215 0.420*** Black Protestant – – – −0.054 0.336 0.947 −0.212 0.385 0.809 Catholic – – – −0.488 0.144 0.614*** −0.607 0.199 0.545** Other – – – −0.832 0.313 0.435** −2.008 0.748 0.134** Attendance: frequent – – – 1.231 0.132 3.424*** 0.789 0.179 2.201*** Attachment: weak – – – −0.412 0.119 0.662*** −0.424 0.168 0.654* Immigration: liberal – – – −0.331 0.126 0.718** −0.397 0.164 0.672* Education – – – −0.288 0.061 0.750 −0.300 0.062 0.741*** Cohort (<1940) 1940–59 – – – −0.734 0.169 0.480*** −0.744 0.171 0.475*** ≥1960 – – – −1.180 0.172 0.307*** −1.210 0.174 0.298** Female 0.189 0.104 1.208† 0.215 0.115 1.240† 0.222 0.116 1.248† South 0.731 0.110 2.076*** 0.543 0.125 1.721*** 0.520 0.126 1.682*** Race (white) Black 0.810 0.172 2.249*** 0.895 0.282 2.447** 0.925 0.286 2.521** Other −0.640 0.239 0.527** −0.412 0.261 0.662 −0.349 0.264 0.706 Interactions: Religion (Evangelical) Mainline – – – – – – 0.674 0.321 1.962* Black Potestant – – – – – – 0.273 0.475 1.313 Catholic – – – – – – 0.267 0.281 1.307 Other – – – – – – 1.682 0.843 5.376* Attendance: frequent – – – – – – 0.994 0.272 2.703*** Attachment: weak – – – – – – 0.021 0.236 1.021 Immigration: liberal – – – – – – 0.137 0.252 1.147 Constant −0.741 0.101 0.476*** 0.804 0.212 2.234*** 1.144 0.248 3.139*** −2LLX2 2110.320 1847.794 1825.681 ΔX2 145.118*** 262.526*** 22.113** df 5 10 7 Nagelkerke R2 0.114 0.295 0.310 Model 1 Model 2 Model 3 β sd(β) exp(β) β sd(β) exp(β) β sd(β) exp(β) Survey year 0.714 0.104 2.041*** 0.900 0.121 2.460*** 0.243 0.288 1.275 Religion (Evangelical) Mainline – – – −0.568 0.161 0.567*** −0.868 0.215 0.420*** Black Protestant – – – −0.054 0.336 0.947 −0.212 0.385 0.809 Catholic – – – −0.488 0.144 0.614*** −0.607 0.199 0.545** Other – – – −0.832 0.313 0.435** −2.008 0.748 0.134** Attendance: frequent – – – 1.231 0.132 3.424*** 0.789 0.179 2.201*** Attachment: weak – – – −0.412 0.119 0.662*** −0.424 0.168 0.654* Immigration: liberal – – – −0.331 0.126 0.718** −0.397 0.164 0.672* Education – – – −0.288 0.061 0.750 −0.300 0.062 0.741*** Cohort (<1940) 1940–59 – – – −0.734 0.169 0.480*** −0.744 0.171 0.475*** ≥1960 – – – −1.180 0.172 0.307*** −1.210 0.174 0.298** Female 0.189 0.104 1.208† 0.215 0.115 1.240† 0.222 0.116 1.248† South 0.731 0.110 2.076*** 0.543 0.125 1.721*** 0.520 0.126 1.682*** Race (white) Black 0.810 0.172 2.249*** 0.895 0.282 2.447** 0.925 0.286 2.521** Other −0.640 0.239 0.527** −0.412 0.261 0.662 −0.349 0.264 0.706 Interactions: Religion (Evangelical) Mainline – – – – – – 0.674 0.321 1.962* Black Potestant – – – – – – 0.273 0.475 1.313 Catholic – – – – – – 0.267 0.281 1.307 Other – – – – – – 1.682 0.843 5.376* Attendance: frequent – – – – – – 0.994 0.272 2.703*** Attachment: weak – – – – – – 0.021 0.236 1.021 Immigration: liberal – – – – – – 0.137 0.252 1.147 Constant −0.741 0.101 0.476*** 0.804 0.212 2.234*** 1.144 0.248 3.139*** −2LLX2 2110.320 1847.794 1825.681 ΔX2 145.118*** 262.526*** 22.113** df 5 10 7 Nagelkerke R2 0.114 0.295 0.310 View Large

Since we know that the only significant rise in CA took place among Christian Americans, it is not surprising that the coefficient for survey year in Model 1 is significantly positive. Overall, the observed odds of a “very important” response among Christians doubled in size between 1996 and 2004. Because it is still significant, and even more positive, when we add the other independent variables (Model 2), we can safely conclude that this result is not explained by compositional change. More striking is that the main effect of survey year is no longer significant when the interaction terms are included (Model 3). Thus, although there may have been a modest (28 percent) increase in odds for Christians as a whole, most of the net change in CA evidently occurred in those categories where the interactions with survey year are significant.

Consistent with our prediction, CA experienced fairly substantial growth in the category of frequent churchgoers. While the net odds of a “very important” response did not change significantly for less-than-weekly attendance, they increased by a factor of 2.7 if attendance is at least a weekly occurrence, almost tripling the previous gap in odds between frequent and less frequent attendees. In other cases, however, the patterns are more suggestive of upward convergence, or intensification “from below.” Among mainline Protestants, for example, the net odds of a “very important” response were only 42 percent as great as for evangelicals (and black Protestants) in 1996, but by 2004 the gap had all but disappeared. For the Other Christian category, the rise in CA is exceptionally dramatic. At the end of our period, the estimated odds were more than five times as great as at the beginning, and about 70 percent as great as for evangelicals (compared with an odds ratio of 13 percent in 1996).

Although these exceptions are inconsistent with our prediction of intensification “at the top,” we believe they are most likely related to compositional changes (H5) for which analysis was unable to control.7 In the case of mainline Protestants, we suspect that the much of relative increase in CA in this shrinking category was due to attrition by the weakly committed into the category of religious “nones” (cf., Hout and Fischer 2002). Although limitations of our data do not permit us to confirm this directly, it would mean that the remaining contingent of mainliners is increasingly composed of strong believers in a CA. If this is what occurred, then the finding would be a byproduct of compositional change, rather than genuine intensification “from below.”

Unlike that for mainline Protestants, the surge in CA among Other Christians coincided with a twofold increase in the relative size of the category. As it turns out, roughly 40 percent of this growth was made up by newcomers of foreign origin. In light of our theory, we would expect recent immigrants of Christian faith to be especially attracted by the idea that being a Christian makes one more “truly American.” Although limitations of sample size prevent us from testing this inference, it would mean that the disproportionate rise in CA among Other Christians represents a special case of compositional change, rather than selective intensification. That is, even though the percentage of Christians among recent immigrants is slightly lower than in the population (Pew Forum 2008:47), a disproportionate influx of foreign-born Christians into the Other Christian category may have increased the proportion of members with strong CA beliefs.8

If these conjectures are borne out in future research, then our findings indicate that the growing prevalence of CA among Christians resulted primarily from selective intensification among the most frequent churchgoers, regardless of denominational affiliation and other characteristics. In other words, just as highly committed Christians are more likely than others to embrace the idea of a Christian America, they are also the most prone to find comfort in this belief during unsettling times. It also means that, within any given Christian denomination, the most frequent churchgoers have drawn even further ahead of less active Christians in promoting religious conceptions of American identity. From the former's perspective, being a “pious” Christian may be increasingly viewed as more “American” than mere nominal membership in a Christian denomination.

CONCLUSION

Conceptions of national belonging inevitably fashion borders and boundaries between “people like us” and “others” who do not fully belong (Edgell et al. 2006; Kunovich 2006; Lamont and Molnár 2002; Smith et al. 1998; Zimmer 2003). Just as some men and women, Caucasians and African Americans, or Christians and Jews will be thought more or less prototypical of their sex, race, or religion than others, the symbolic boundaries of national belonging recognize shades and gradations of authenticity among those they designate as insiders with respect to the national community. Thus, while the acquisition of U.S. citizenship has traditionally sufficed to make an Italian, Indonesian, or Ivorian immigrant into an American by nationality, the question of who is more “truly” American also depends on how understandings of national belonging are aligned with other dimensions of social membership, including the boundaries of religious faith.

Because symbolic boundaries can have practical consequences—affecting social prestige or even political clout (Kunovich 2006; Lamont and Molnár 2002)—it is perhaps only “natural” that the dominant groups in society would seek to align the boundaries of national belonging with membership in their particular communities. Thus, because America has always been a predominantly Christian country, many American Christians no doubt feel justified in viewing their religion as characteristic of the “true” American. Among individual Christians, the degree of alignment between religion and national identity also tends to reflect the salience of both religious and national boundaries. Within the religious majority, belief in a Christian America (CA) grows significantly stronger with intensity of religious commitment and national attachment.

Substantial variation also exists between different segments of the Christian community itself. For example, CA appears to be especially prevalent in evangelical denominations, perhaps because they tend to position their own doctrines and practices as more orthodox—and more “American”—than those of other Christian traditions (Smith et al. 1998, 2000). In other instances, religious definitions of national identity may function primarily as a strategy for combating social and symbolic exclusion. Such a strategy might explain the special appeal of CA within the black Protestant community.

Whatever their meaning for religious insiders, however, any religious understanding of American identity is likely to meet with skepticism, if not resentment, among those they symbolically marginalize. Understandably, non-Christian believers, as well as those with no religious affiliation, overwhelmingly reject the notion that being “truly” American requires adherence to the dominant faith. Such opposition to a Christian America need not be construed as denying the objective importance of Christianity in American life. More likely, it represents a desire to downplay the symbolic significance of religion for defining American identity, in favor of religious pluralism and tolerance for diversity—values that also have deep roots in U.S. history.

Given the diminishing ratio of Christians to non-Christians in the United States (Kosmin and Keysar 2009; Smith and Kim 2005), one might have expected the overall prevalence of CA beliefs to decline over time. Yet as we saw, Americans were even more likely to espouse these beliefs after the turn of the century that they had been just a decade before. As of 1996, 38 percent of our respondents regarded being a Christian as very important for being truly American; by 2004, the figure had risen to nearly 50 percent (see also Pew Research Center 2006). This trend was all the more surprising given the weaker salience of religious boundaries among better educated Americans, as well as among those born in more recent decades. As Americans became better educated, and as older generations gave way to younger ones, the appeal of a Christian American should have been on the decline.

Although demography alone would not have predicted a surge in CA, our reasoning was basically correct. When we took these trends into account, our models predicted an even greater increase in the absence of compositional change. At the same time, the fact that CA experienced a significant rally at all makes it clear that demography in this instance is not destiny, at least in the near term. For example, it may be, as some have argued, that certain cataclysmic events since the turn of the century have enhanced the importance of religion to the self-understandings of many Americans (Huntington 2004). Such period effects also have precedent in the literature on religious politicization and deprivatization (Regnerus and Smith 1998; Schwadel 2005).

Yet, if the traumatic events of 9/11 or the war in Iraq have accentuated the connection between religion and national identity, we suspected that any resulting growth in CA would come primarily from American Christians. This hunch was well supported by the data. Among Christians, the proportion who saw Christianity as very important to American identity increased substantially between 1996 and 2004, assisted in part by a surge in religious commitment and national attachment. Yet, among non-Christians there was no significant change in either direction, even when other predictors were taken into account. Within the religious majority, moreover, we found the largest proportionate increase in CA among Christians with at least weekly attendance at religious services. This was in line with our prediction that intensification will be especially pronounced, in “unsettled times,” where the appeal of CA had been strongest in the past. It also confirmed that events which provoke a growing religious divide can accentuate the moral boundary between more and less devout members of the dominant faith.

Given the limitations of our data, we cannot be certain that any particular set of events was responsible for the rise in CA during our period of study. All we can state with confidence is that the increase was not a direct byproduct of changes in the religious composition of society. At the same time, there are indications that the indirect impact of growing religious diversity has been more complex. On the one hand, the expansion of cultural tolerance and religious diversity in the last few decades may explain why younger and better-educated Christians tend to downplay the relevance of religion as a component of American identity. It may also account for the fact that outgroup hostility (as measured by immigration preferences) was on the decline during our period of study. On the other hand, Christians who did favor stronger restrictions on the rate of immigration were also more prone to erect religious barriers against non-Christian minorities. This suggests that the appeal of Christian America discourse may be related, not only to unsetting international events, but also to the salience of cultural issues in domestic policy debates, such as those involving immigration. Moreover, we speculated that recent immigrants who are Christians may be disproportionately attracted by the idea that their faith makes them more “truly” American. If so, then it is possible that the overall prevalence of CA in the population will decline more slowly than the percentage of Christians among newly minted Americans.

More generally, our findings suggest that, when a controversial belief becomes more pronounced in the population, it may be symptomatic of a growing divergence of opinion around perennial cleavages. If such polarization raises the potential for future social conflict, as some have argued (Bolzendahl and Brooks 2005; DiMaggio et al. 1996), then the extent of disagreement over American identity may be as at least as consequential as the prevalence of any particular definition. In practice, of course, the likelihood of conflict also depends on differences in attitudes line up with other politically relevant characteristics and commitments (DiMaggio et al. 1996). In the present case, it is likely that belief in a Christian America often varies as much within traditional coalitions as between them. An older African American Protestant, for example, would likely agree on many social and political issues with a young, secular college graduate who happens to be white. (Both would likely have voted for the Democratic nominee in the most recent presidential election.) Yet, as our findings suggest they would probably have very different views about the relevance of religion to matters of national identity. (We would expect a similar difference between, say, an economic conservative who is Jewish and a socially conservative evangelical, even if both typically vote Republican.)

At the same time, there are reasons to believe that religious boundaries of national identity can have major political consequences, even rivaling those of race and national origin. In the 2008 presidential election, for example, the racial identity of the Democratic candidate seemed less of an obstacle to electoral success than certain (mistaken) beliefs about his religious convictions. Given his relatively narrow margin of victory, it seems doubtful that Barack Obama would have become the nation's first black President had he indeed been a Muslim instead of a Christian. It is also worth noting how questions about the candidate's religion called attention to issues of race, and vice versa. Thus, rumors that Obama was privately a Muslim helped focus the media spotlight on the Northwest African side of his parental heritage. Meanwhile, Obama's long-standing membership in a West Chicago congregation not only substantiated his Christian convictions; it also brought to light some of the cultural differences between black Protestants and Christians of other races and denominations.

Such examples underscore our contention that the salience of any one component of identity often depends on that of other dimensions with which it is aligned. They also serve as a reminder that religion is only one of the many symbolic resources available for constructing the boundaries of American identity. If U.S. Christians tend to define America as a Christian nation, it would be of interest in future research to see whether other historically privileged categories exhibit similar proclivities—for instance, whether white Americans are more inclined than others to think of Caucasians as the quintessential Americans, or whether Americans who speak only English are more prone insist that being a “true” American requires fluency in English. If so, then a decline in the salience of religious boundaries over time would not, by itself, indicate that American identity has grown more pluralistic. It may instead signal that other axes of symbolic exclusion, such as language, ethnicity, or even patriotism, are on the ascendant.

SUPPLEMENTARY MATERIAL

A supplementary section is located with the electronic version of this article at Sociology of Religion online (http://socrel.oxfordjournals.org).

REFERENCES

Alwin Duane F. McCammon Ryan J. Mortimer J. T. Shanahan M. J. Generations, Cohorts, and Social Change , Handbook of the Life Course , 2003 New York Kluwer Academic/Plenum Publishers (pg. 23 - 49 ) (pg. Bail Christopher A. The Configuration of Symbolic Boundaries against Immigrants in Europe , American Sociological Review , 2008 , vol. 73 (pg. 37 - 59 ) , vol.(pg. Bolzendahl Catherine Brooks Clem Polarization, Secularization, or Differences as Usual? The Denominational Cleavage in US Social Attitudes Since the 1970s , The Sociological Quarterly , 2005 , vol. 46 (pg. 47 - 78 ) , vol.(pg. Casanova José Public Religions in the Modern World , 1994 Chicago University of Chicago Press Citrin Jack Wong Cara Duff Brian Ashmore R. Jussim L. Wilder D. The Meaning of American National Identity , Social Identity, Intergroup Conflict, and Conflict Resolution , 2001 New York Oxford University Press (pg. 71 - 100 ) (pg. Davis James A. Smith Tom W. Marsden Peter V. General Social Surveys, 1972–2006 [ICPSR04697-v2] , 2007 Chicago, IL National Opinion Research Center DiMaggio Paul Evans John H. Bryson Bethany Have Americans' Social Attitudes Become More Polarized? , American Journal of Sociology , 1996 , vol. 102 (pg. 690 - 755 ) , vol.(pg. Edgell Penny Gerteis Joseph Hartmann Douglas Atheists As ‘Other’: Moral Boundaries and Cultural Membership in American Society , American Sociological Review , 2006 , vol. 71 (pg. 211 - 34 ) , vol.(pg. Esses Victoria M. Dovidio John F. Jackson Lynne E. The Immigration Dilemma: The Role of Perceived Group Competition, Ethnic Prejudice, and National Identity , The Journal of Social Issues , 2001 , vol. 57 (pg. 389 - 412 ) , vol.(pg. Hoge D. R. de Zulueta E. Salience as a Condition for Various Social Consequences of Religious Commitment , Journal for the Scientific Study of Religion , 1985 , vol. 24 (pg. 21 - 38 ) , vol.(pg. Hout Michael Fischer Claude S. Why More Americans Have No Religious Preference: Politics and Generations , American Sociological Review , 2002 , vol. 67 (pg. 165 - 90 ) , vol.(pg. Huddy Leonie Khatib Nadia American Patriotism, National Identity, and Political Involvement , American Journal of Political Science , 2007 , vol. 51 (pg. 63 - 77 ) , vol.(pg. Huntington Samuel P. Who Are We? The Challenges to America's National Identity , 2004 New York Simon & Schuster Jones Frank L. Smith Philip Diversity and Commonality in National Identities: An Exploratory Analysis of Cross-National Patterns , Journal of Sociology , 2001 , vol. 37 (pg. 45 - 63 ) , vol.(pg. Kosmin Barry A. Keysar Ariela American Religious Identification Survey (ARIS 2008), Summary Report , 2009 Kunovich Robert M. An Exploration of the Salience of Christianity for National Identity in Europe , Sociological Perspectives , 2006 , vol. 49 (pg. 435 - 60 ) , vol.(pg. Lamont Michèle The Dignity of Working Men: Morality and the Boundaries of Race, Class, and Immigration , 2000 New York Russell Sage Foundation Lamont Michèle Molnár Virág The Study of Boundaries Across the Social Sciences , Annual Review of Sociology , 2002 , vol. 28 (pg. 167 - 95 ) , vol.(pg. Long J. Scott Regression Models for Categorical and Limited Dependent Variables , 1997 Thousand Oaks Sage Mummendey Amélie Wenzel Michael Social Discrimination and Tolerance in Intergroup Relations: Reactions to Intergroup Difference , Personality and Social Psychology Review , 1999 , vol. 3 (pg. 158 - 74 ) , vol.(pg. Pattillo-McCoy Mary Church Culture as a Strategy of Action in the Black Community , American Sociological Review , 1998 , vol. 63 (pg. 767 - 84 ) , vol.(pg. Pew Forum , U.S. Religious Landscape Survey , 2008 Washington, DC Pew Forum on Religion & Public Life Pew Research Center , Many Americans Uneasy with Mix of Religion and Politics , 2006 Washington, DC Pew Research Center for the People and the Press Regnerus Mark D. Smith Christian Selective Deprivatization among American Religious Traditions: The Reversal of the Great Reversal , Social Forces , 1998 , vol. 76 (pg. 1347 - 72 ) , vol.(pg. Schwadel Philip Education and Churchgoing Protestants’ Views of Highly Politicized Christianity , Review of Religious Research , 2005 , vol. 47 (pg. 150 - 61 ) , vol.(pg. Servin-Gonzales Mariana Torres-Reyna Oscar Trends: Religion and Politics , Public Opinion Quarterly , 1999 , vol. 63 (pg. 592 - 621 ) , vol.(pg. Smith Tom W. Kim Seokho The Vanishing Protestant Majority , Journal for the Scientific Study of Religion , 2005 , vol. 44 (pg. 211 - 23 ) , vol.(pg. Smith Tom W. Kim Seokho National Pride in Comparative Perspective: 1995/96 and 2003/04 , International Journal of Public Opinion Research , 2006 , vol. 18 (pg. 127 - 36 ) , vol.(pg. Smith Christian Emerson Michael Gallagher Sally Kennedy Paul Sikkink David American Evangelicals: Embattled and Thriving , 1998 Chicago University of Chicago Press Smith Christian Emerson Michael Gallagher Sally Kennedy Paul Sikkink David Christian America? What Evangelicals Really Want , 2000 Berkeley University of California Press Somers Margaret R. Bodemann Y. M. Yurdakul G. Citizenship, Statelessness and Market Fundamentalism: Arendtian Right to Have Rights , Migration, Citizenship, Ethnos , 2006 New York/Hampshire Palgrave-Macmillan (pg. 35 - 62 ) (pg. Steensland Brian Park Jerry Z. Regnerus Mark D. Robinson Lynn D. Wilcox W. Bradford Woodberry Robert D. The Measure of American Religion: Toward Improving the State of the Art , Social Forces , 2000 , vol. 79 (pg. 291 - 318 ) , vol.(pg. Swidler Ann Culture in Action: Symbols and Strategies , American Sociological Review , 1986 , vol. 51 (pg. 273 - 86 ) , vol.(pg. Voas David Crockett Alasdair Olson Daniel V.A. Religious Pluralism and Participation: Why Previous Research is Wrong , American Sociological Review , 2002 , vol. 67 (pg. 212 - 30 ) , vol.(pg. Weil Frederick D. Cohorts, Regimes, and the Legitimation of Democracy: West Germany Since 1945 , American Sociological Review , 1987 , vol. 52 (pg. 308 - 24 ) , vol.(pg. Wilcox Clyde Goldberg Rachel Public Opinion on Church-State Issues in a Changing Environment , Journal for the Scientific Study of Religion , 2002 , vol. 41 (pg. 369 - 76 ) , vol.(pg. Wimmer Andreas The Making and Unmaking of Ethnic Boundaries: A Multilevel Process Theory , American Journal of Sociology , 2008 , vol. 113 (pg. 970 - 1022 ) , vol.(pg. Wuthnow Robert America and the Challenges of Religious Diversity , 2005 Princeton, NJ Princeton University Press Zimmer Oliver Boundary Mechanisms and Symbolic Resources: Towards a Process-Oriented Approach to National Identity , Nations and Nationalism , 2003 , vol. 9 (pg. 173 - 93 ) , vol.(pg. Zolberg Aristide R. A Nation by Design: Immigration Policy in the Fashioning of America , 2006 Cambridge, MA Harvard University Press

© The Author 2010. Published by Oxford University Press on behalf of the Association for the Sociology of Religion. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org.