Content area
Those who have believed in a "Catholic vote" as some unified bloc or as a consistent "swing" component of the U.S. electorate should pay close attention to the results of the 2004 presidential election. The failure of John F. Kerry, the first major party Catholic candidate since John F. Kennedy, to definitively win the votes of those in the electorate who share his faith should finally put to rest the myth of a Catholic presidential vote. The research presented here indicates that it was also a myth that most Catholics were primarily motivated in 2004 by "moral values" in making their presidential choice at the ballot box. Using survey data from the 1960, 2000, and 2004 elections, we show that partisanship has grown to trump faith for Catholic voters due to a combination of demographic factors, changes within the Catholic Church, and changes within the U.S. party system. The defining features of anything that one might call the Catholic vote are in its fractures, not its wholeness. [PUBLICATION ABSTRACT]
Full text
Those who have believed in a "Catholic vote" as some unified bloc or as a consistent "swing" component of the U.S. electorate should pay close attention to the results of the 2004 presidential election. The failure of John F. Kerry, the first major party Catholic candidate since John F. Kennedy, to definitively win the votes of those in the electorate who share his faith should finally put to rest the myth of a Catholic presidential vote. The research presented here indicates that it was also a myth that most Catholics were primarily motivated in 2004 by "moral values" in making their presidential choice at the ballot box. Using survey data from the 1960, 2000, and 2004 elections, we show that partisanship has grown to trump faith for Catholic voters due to a combination of demographic factors, changes within the Catholic Church, and changes within the U.S. party system. The defining features of anything that one might call the Catholic vote are in its fractures, not its wholeness.
Before you drift to sleep upon your cot, think back on all the tales that you remember. . . . That once there was a fleeting wisp. . . . Don't let it be forgot that once there was a spot for one brief shining moment that was known as Camelot.
-Lyrics from the 1960 Broadway musical Camelot
Despite the volume of ink and sound bite noise generated in discussions of the "Catholic vote" during the 2004 presidential campaign, there is but one simple empirical lesson that the election results reinforced: on the long scale of modern U.S. electoral history the Catholic vote, much like the legend of Camelot, was a "fleeting wisp" most evident with the election of John F. Kennedy in 1960.1 However, it ceased to be a cohesive bloc more than forty years ago and, as a group, Catholic voters have not always been the consistent "swing vote" group that they are often portrayed to be.
The election of John F. Kennedy was truly an extraordinary and liberating event for U.S. Catholics (Dolan 1992, 422). As Crews (1993) notes, "An invisible barrier had been shattered. Across the nation, Catholics sensed that they had finally achieved unquestioned first-class status as loyal citizens" (p. 139). After enduring more than 170 years of political anti-Catholicism in various forms and degrees,2 eight in ten Catholic voters supported Kennedy at the ballot box in 1960. At the time, Converse et al. (1961) noted "the vote polarized along religious lines in a degree which we have not seen in the course of previous sample survey studies" (p. 273). Slightly more than 118,000 votes separated Kennedy from Nixon nationally and it would have been a daunting challenge for him to win with anything less than 80 percent of the Catholic vote. Converse et al. (1961) estimate that Kennedy's vote among Catholics was 17 percentage points higher than what an average non-Catholic candidate could have expected (p. 275).
As John F. Kerry's candidacy began to take shape in late 2003, it was an open question as to how Catholics might react if he were to be the first Catholic since Kennedy to win one of the two major parties' nominations. Few expected that Kerry would be able to attract eight in ten Catholic voters, yet most would not have bet that he might lose the Catholic vote. This article puts that latter result-which will likely always be in question-into context by first introducing existing theory and evidence and then proceeding to comparisons of the Catholic vote in 1960, 2000, and 2004.
The U.S. Presidency and the "Catholic Vote"
Although John F. Kennedy's election is often considered a breakthrough moment overcoming a long tradition of overt political anti-Catholicism, it is a misconception that Catholics were somehow systematically kept out of U.S. government, even as it began and the Catholic population was very small.3 In 1790, just 1 percent of the U.S. population was Catholic and as of 1840 it was still only 4 percent (Prendergast 1999, 2). Yet, things began to change in the 1850s as waves of immigration began from many predominantly Catholic countries4-first from Europe and then followed by Latin America as well as Asia and Africa in the twentieth century (ongoing)-which would result in Catholics never again being less than 10 percent of the U.S. population.
Although Kennedy was the first Catholic to win, he was of course not the first to run. Al Smith was the first, getting the Democratic party's nomination in 1928 when Catholics still made up less than 20 percent of the population. Smith's Catholicism was the central issue of that campaign and similar to how Catholics would support Kennedy three decades later, he benefited from a huge Catholic turnout. Although Prendergast (1999) estimates Catholics made up only a third of Smith's votes, he concludes that "in no presidential election, before or since, have Catholics been so close to unanimity in their choice of a candidate" (p. 96).5 However, this is largely an ecological assertion made without the benefit of surveys that might allow for comparison to I960.
Smith's candidacy is certainly significant in that he broke one barrier in getting the nomination. Yet it was also significant in that the opposition to him was able to evoke the political anti-Catholicism that overtly insinuated a Catholic president would put pope and church before country and constitution.6 In the years that followed, the disappointment in Smith's failure was stemmed by the continued growth in the number of Catholics in the electorate. In the decade before Kennedy was elected, the percentage of the U.S. population self-identifying as Catholic had crossed the 20 percent barrier and has remained approximately 22 to 24 percent (among adults) ever since.7 With Catholics making up nearly a quarter of the electorate, the prospect of a Catholic candidate on the ticket was attractive and culminated in Kennedy's I960 nomination.
To counteract these potential attacks against his religion, the Kennedy campaign engineered a strategy that Kerry would utilize more narrowly in 2004 on the issues of abortion, stem cell research, and same-sex marriages-emphasize the personal and individual importance of faith yet also promise to always govern in a manner entirely independent of any influence from the Catholic Church or its doctrine.8 As important as Kennedy's eventual election was for many U.S. Catholics, it may have been too important for Kerry and other Catholic candidates of the future.9 It had mattered for Smith, who won the first nomination, and Kennedy, who won the first election, however we argue here that both of these elections may have created a "shattered barrier" effect, thereby neutralizing the unifying impact of a candidate's Catholicism.10
Even as Catholics had showed strong support for Catholic Democrats vying for the presidency in the past, the propensity for Catholics to self-identify as Democrats or to vote for Democratic presidential candidates consistently weakened after 1964. As Table 1 indicates using the most widely utilized surveys, Catholics did not vote in the majority for a Democrat in 1972, 1980, 1984, and possibly in 1992.
Table 1 also shows the often precarious nature of measuring the vote of Catholics as there is not always agreement among Gallup surveys, the National Election Studies (NES), and the major media exit polls. In 1956, 1988, 1992, and 2004, the winner of the Catholic vote cannot be unanimously determined. The Gallup and NES surveys include many fewer Catholic respondents than the exit polls and thus have higher margins of sampling error. However, exit poll methodology and cluster sampling introduce additional complications-especially in more recent years, with greater numbers of voters having access to and selecting absentee or vote-by-mail options.11 It is unlikely that anyone will ever know whether Kerry actually "won" the Catholic vote given the discrepancies shown in Table 1.
The large number of interviews conducted for the state-level exit polls does allow for more localized explorations of these data that provide some insight into how Catholics voted, however. As Table 2 indicates, Kerry won an estimated 58 percent of the Catholic vote in Western states. Differences in the estimated Catholic vote percentages between Kerry and Bush are 1 percentage point or less in the Midwest, non-swing states, and in states that did not have a bishop make statements about possibly denying Kerry Communion at Mass in their diocese if Kerry did not change his position on some aspect of Catholic doctrine-usually abortion.
Overall, fifteen bishops in fourteen states (see Appendix) made such statements between November 2003 and the election.'2 Within those states, Kerry only won 44 percent of the vote. At this level of analysis there can be no sound argument of causality as it could be coincidental, a function of overlapping geography, or some reverse causality in that bishops in states where Catholics were already predisposed to vote against Kerry felt more able to make these statements.
It is clear that there are both regional variations in the current Catholic vote and some considerable election-to-election volatility in the vote choices of Catholics, with majorities switching back and forth between Democratic and Republican candidates over time." However, as Figure 1 indicates, there has been more stability in Catholic party identification although it has trended slightly away from the Democrats in recent years. Several researchers indicate that there has been a Catholic shift toward the Republican party (e.g., Wattenberg and Miller 1981; Abramson, Aldrich, and Rhode 1994; Wagner 2001). Three pieces of evidence are clear: (1) declining Democratic identification among Catholics from the mid 1960s to the late 1980s, (2) a rather slight decrease relative to Protestants in the likelihood of Catholics voting for Democrats in national elections, and (3) the increasing tendency since 1980 of active regular Mass-attending Catholics to vote Republican in presidential elections.14
In 2004, Catholics were still more likely to self-identify as Democrat (including "leaners") than anything else, though this identification has dropped below majority status for the first time in a presidential election year since 1952 (49 percent compared to 37 percent Republican). In 2002, the percentage of Catholics who considered themselves to be Democrats was even lower. These party identification shares do put the 2004 vote in greater perspective. If Bush won a majority, or nearly a majority, of the Catholic vote, he did so by attracting some Catholics who do not consider themselves Republican. This realization led Democratic political consultants (i.e., Democracy Corps) to issue a widely circulated post-election memo entitled "Reclaiming the White Catholic Vote" (Greenberg and Hogan 2005). They note, "The drop in Catholic support is a big part of the 2004 election story" (p. 1). They focus their argument on a group they call the "democratic defectors"-Catholics who consider themselves Democrats but who have voted for Republican presidential candidates in recent elections.
As the NES survey data presented in Table 3 indicate, however, Kerry's Catholicism may have decreased this rate of defection among Catholics from the previous election. Seven percent of the Catholic electorate in 2004 was made up of voters who had cast ballots for Bush in 2000 and then voted for Kerry in 2004. Only 3 percent of Catholic voters had voted for Gore in 2000 and then voted for Bush in 2004." Although this gain is nowhere near the 31 percent Catholic voters who had voted for Eisenhower in 1956 and then voted for Kennedy in I960, it is an indication that Kerry's Catholicism may have been at least a slight short-term force upon the vote of Catholics.
One of the interesting aspects of Table 3 is that Kerry, although unable to attract any groundswell of Catholic votes, was slightly more able to hold the non-Catholic · Democratic vote from the previous election than Kennedy was. In a very close election, Kennedy likely needed the substantial shift from Catholic Republicans to overcome some defections among his own party voters from the previous election who did not share his religion. This result underlies the fundamental aspect of difference between I960 and 2004: for Catholics, partisanship is substantially more important than religion now than it was! in I960. Catholics are not so attracted to a candidate of their own religion that they ·", defect from their partisanship in large numbers, and non-Catholics do not fear the potential impact of a Catholic candidate winning so much that they vote against their partisanship for the presidency.16
Greenberg and Hogan note that Catholics are divided not just on partisanship but on many issues (2005, 17), and there is research indicating a great deal of internal ideological diversity among Catholics (Jelen 1996; Olson 1997, 250; Bendyna 2000). The largest differences have developed on so-called cultural or moral issues-abortion, gay and lesbian rights, euthanasia, stem cell research, and government funding for birth control-all issues that Kennedy and the Catholic voters of I960 did not face.
A Global Church within a Nation's Politics
The premise that Catholics may vote differently from others is based on the notion that there is something about this faith and belief system that affects Catholics' political attitudes, ideology, and behavior. Catholicism is unique in some respects in that it is a truly global faith with a common set of doctrine and canon law that binds dioceses, bishops, clergy, parishes, and the Catholic laity together worldwide. Although there is considerable autonomy in many areas for bishops and clergy within their dioceses and parishes, official Church doctrine on many social and political issues is not an area where many are allowed to stray too far afield.
The Catholic Church in the United States is not able to campaign or advocate' specifically for or against one party/candidate or another without risking the Church's tax status. However, the United States Conference of Catholic Bishops (USCCB) has regularly issued pre-election documents since 1975 that outline the Church's position on issues of importance in light of Church doctrine. In 2004, this document, which serves as kind of a political platform for the Church, was entitled Faithful Citizenship: A Catholic Call to Political Responsibility. This document, as well as Church doctrine more generally, tends to pull Catholic voters toward opposite ends of the political spectrum.
In Table 4, the "moral priorities in public life" from this document are listed along with checkmarks indicating the authors' estimations of Republican or Democratic party advantage on each issue. This is a somewhat subjective and imprecise demonstration. However, it is also one that almost anyone informed about the Church and the U.S. party system could replicate with few differences.
For example, under "protecting human life," one could argue that the Republican party platform is more consistent with the Church's teachings on abortion, euthanasia, and cloning, whereas the Democratic party platform is more in line with restrictions and restraint in using the death penalty and with serious moral concerns regarding preemptive use of force. In each area, other than "practicing global solidarity," there is a noticeable party split. If a Catholic voter were to take this statement, as well as the Church's teachings in general, into close consideration while voting, they would likely be divided on whom they could vote for. One of the notable aspects of Table 4 is that for most issues listed there was either no party difference or there was no such existing issue in I960. As a Catholic candidate, John Kerry faced a much greater dilemma in crafting his campaign than Kennedy did.
The U.S. Catholic bishops admit "a Catholic moral framework does not easily fit the ideologies of 'right' or 'left' nor the platforms of any party" (USCCB 2004). Yet Catholics in America do have to fit themselves within the two-party system. Perhaps in 960 they fit better. Yet as most of the issues in Table 4 emerged in the years that followed, many Catholics made a choice to follow their partisanship and focus on the issues that their party supported that were consistent with Church teachings.17 This divided Catholic electorate has played an important role. Bendyna (2000) argues that Catholics have provided what might be considered a "Goldilocks effect," ensuring that the overall political discourse in the United States never reaches too far left or right. She shows that Catholic Republicans are more liberal than their non-Catholic counterparts on issues of social welfare (Bendyna et al. 2001). And Catholic Democrats are more conservative than their non-Catholic counterparts on cultural issues. The implications of Bendyna's findings are that Catholicism may have a moderating role in public-policy debate, with each group helping to keep its respective party from advocating or enacting the most extreme policies in those areas.
The extent to which the advocacy of Church leaders can successfully influence the political opinions of lay Catholics within this context remains an issue of great debate (Welch et al. 1993; Leege and Mueller 2000). Clear evidence of influence exists for only a few issues, and the strength of that influence varies.18 What, then, can be concluded about Catholics' contributions to public discourse on politics? One view is that the influence is minimal, with Catholics much more likely to be politically influenced by the society about them than the other way around. Leege has emphasized that factors such as Catholics' gender, area of geographic residence, and socioeconomic status remain better predictors of their voting and their attitudes on noncultural issues than their religious commitment (Leege and Welch 1989; Leege 2000; Leege and Mueller 2000).
On Closer Examination: Multivariate Analysis
To explore further how Catholics cast their votes with a Catholic candidate on the ballot, we conduct a series of multivariate logistic regressions using the NES and media exit poll data. For added context we make comparisons to 1960 and 2000.
There is considerable existing research that compares Catholic political attitudes and behavior with that of Evangelical and Mainline Protestants. We depart from these approaches here to pose a minimalist test by comparing the 1960 and 2004 elections where a Catholic candidate was on the ballot. We simply ask, do Catholics stand out at all compared to non-Catholics, and if so, how? The dependent variable for each model presented is the two-party vote choice, where a vote for the Democratic candidate is coded as "1" and a vote for the Republican candidate is coded as "0."
We begin with a basic demographic model using NES data19 and including variables for gender, age, race and ethnicity, and education, with the addition of a dummy variable for respondents who self-identify as Catholic. The reported logistic regression exp(b) coefficients can be interpreted as such: if exp(b) is greater than 1 then an increase in a statistically significant explanatory variable is associated with the odds of the dependent variable outcome occurring (i.e., voting for the Democrat). If the exp(b) is less than 1 then as the explanatory variable increases the odds of the outcome event occurring decreases (i.e., respondents vote Republican).
As displayed in Model 1 for 2004 in Table 5, this very basic estimation does indicate the existence of a small Catholic effect, with respondents self-identifying as such having been slightly more likely to vote for Kerry than Bush as compared to non-Catholics. Additionally, Kerry voters were also more likely to be female, under 30, African American/black, Hispanic/Latino, and have attended graduate school.
In the second NES 2004 model, the addition of measures for party identification and the disaggregation of Catholics by their frequency of Mass attendance provide additional insight and a substantial increase in explanatory power.20 Self-identifying Catholics who report that they "never" attend Mass were more likely than non-Catholics to vote for Kerry, yet none of the other categories for Mass-attending Catholics achieve statistical significance. The increase in explanatory power for this model is primarily driven by the variables measuring party identification. Self-identifying Democrats (including leaners) were substantially more likely than Republicans to vote for Kerry, and those reporting that they are Independents, affiliated with a third party, or who are apolitical were more likely to vote for Kerry as well-albeit with substantially lower odds than Democrats.
We are able to reproduce the same estimations for voters in the 1960 election with one exception. The NES did not ask a question regarding Hispanic/Latino ethnicity in 1960.21 The differences in results for the basic model are noticeable. The odds of a Catholic voter casting a ballot for Kennedy were seven times larger than those of nonCatholics. In the second model, the disaggregation of Catholics by their frequency of Mass attendance indicates a substantially different pattern of results compared to 2004. Catholics attending Mass once a week or more or even at least once a month were substantially more likely to vote for Kennedy than Catholics who rarely or never attended Mass. This is in part reflective of changes in Catholic Mass attendance (Gray 2005) during the period, with substantially more Catholics attending weekly in 1960 than in 2004. Yet it also represents a shift in the choices of those who are at Mass on a weekly basiseven if this is a significantly smaller share of all Catholics in 2004 as compared to I960. Once again the addition of measures of party identification is highly important to predicting vote choice, albeit at a smaller magnitude compared to 2004, with Democrats and Independents/third-party/apolitical respondents favoring Kennedy.
Next, in Table 6, we report results for replications of these models, to the degree that it is possible, with the major media exit polls for 2004 and 2000.22 The results of the NES models indicated a minimal Catholic effect in 2004, but it is unclear whether this is the same effect that any recent non-Catholic Democratic candidate could have expected. Comparisons to 2000 are especially relevant as they include the same challenger, George W. Bush.
In the basic model, women, younger voters (under 30), respondents identifying as something other than white (non-Hispanic), and those who have attended graduate school were more likely to have voted for Kerry than Bush. Once again, a small Catholic effect is evident, with the odds of self-identifying Catholics voting for Kerry having been slightly higher than those of non-Catholics. The addition of party identification and the disaggregation of Catholics by their frequency of Mass attendance provide partial support for the results of the NES models. Once again, the addition of partisanship dramatically increases the amount of variance explained and cases correctly predicted. Among Catholics, those attending Mass once or twice a year-perhaps only at Christmas, Easter, and/or Ash Wednesday-were more likely to vote for Kerry than non-Catholics.
The basic models for the 2000 vote indicate the results of the Kerry model are not exceptional in regard to Catholics. In fact, the odds that Catholics voted for Gore instead of Bush in 2000 are slightly larger than that for Catholics choosing Kerry instead of Bush in 2004. After controlling for partisanship and disaggregating Catholics by Mass attendance even greater differences emerge. Catholics who rarely or never attended Mass in 2000 were more supportive of Gore than they were of Kerry in 2004.
The NES data include too few Catholic interviews to appropriately explore the relation of specific attitudes and ideology to the votes of Catholics specifically. However, the exit poll data do include a sufficient number of interviews to do this research. Although post-election media reports highlighted the importance of "moral values" in Bush's election, as Table 7 indicates, these were not the top issue of concern for Catholic voters.
Catholics were less likely than non-Catholics to cite moral values as the "issue that mattered most in deciding" their presidential vote (17 compared to 24 percent). Catholic respondents were more likely to cite terrorism and the economy or jobs as a deciding factor.
We utilize responses to this issue that "mattered most" question in multivariate models of the Catholic vote in Table 8.23 These estimations only include respondents who self-identify as Catholic. The larger number of Catholic interviews in these surveys also allows for additional variables not included in previous estimations.
In Model 1, we replicate a basic model similar to previous estimations, with the addition of regional variables and a dummy variable measure about whether the respondent lives in a state where a Catholic bishop made a statement indicating the possibility of denying Kerry Communion. Among Catholics specifically, Kerry voters were more likely to be under the age of 30, African American/black or Hispanic/Latino, and to reside in the West (relative to the Northeast and South). Ceteris paribus, those Catholics living in a state where a bishop did make a statement were not more or less likely to have voted for Kerry.24 Catholics attending Mass once a week or more were significantly more likely than Catholics never attending Mass to have voted for Bush.
In Model 2, we add measures for party identification and political ideology. As in previous models, the addition of party identification dramatically increases the explained variance and indicates that Catholic Democrats and Independents are more likely to have voted for Kerry. Parallel results are evident with political ideology as liberal and moderate Catholics voted disproportionately for Kerry as well. The addition of these factors does not significantly alter the results for regions. Results related to Mass attendance no longer achieve statistical significance in this model or the next.
In the final model, we include dummy variables for the respondents' identification of the issue that mattered most for their presidential vote. The excluded reference category is "moral values," and thus the magnitude and statistical significance of each of these is measured against those Catholics who said that moral values was the issue that mattered most to them.
The inclusion of these issue variables strengthens the results for party identification and slightly weakens those for ideology. Controlling for party identification and ideology, Catholics who considered the issues of education, the war in Iraq, taxes, or the economy/jobs as mattering most to them relative to those who selected moral values voted disproportionately for Kerry. Those Catholics who selected terrorism as the issue mattering most-a plurality of Catholics-are no more or less likely to have voted for Kerry than those who were most concerned with moral values. Thus, despite the media coverage regarding moral issues, Kerry may have lost the Catholic vote and possibly the election because he did not sufficiently convince Catholics that he could better deal with the issue considered most important to them-terrorism. If he could have attained a greater edge among Catholics on this issue perhaps the final results might have been different.
Conclusion
John F. Kerry's candidacy came too late for any revival of Camelot. The magic wore off long ago. The Catholic voters' Camelot was likely a brief and one-time event initiated in 1928 and waning a bit until the next Catholic candidate emerged in 1960. However, after 1964 and the reelection of Kennedy's former vice president, no unified Catholic vote on a national level can be empirically identified-even when one of the candidates self-identifies as Catholic.25
In our multivariate models for 2000 and 2004, Catholics are slightly more likely to vote for Democrats; however, Kerry was no more of an attractive choice for Catholics than Gore. John F. Kennedy and John F. Kerry share much in terms of background, yet these two candidates faced starkly different political issues and discourse within the United States and within the Catholic Church during their campaigns. Now with the two most important barriers having been shattered (nomination and election), a Catholic presidential candidate of the Democratic party, like Kerry, can no more count on the vote of those in the electorate who share his or her faith than those who do not. Although the test of a Republican Catholic candidate has not yet been broached, the research presented here indicates that the issues divide among Catholics runs in both directions. Although a Catholic Republican who opposes abortion would likely get more support from the bishops than Kerry did, it is still unclear whether this support matters much to the average Catholic voter. The results presented here also indicate that issues related to social economic justice and foreign policy are equally important predictors of the Catholic vote.
Although there may not be something one could call "the Catholic vote" anymore, this does not mean that the vote of Catholics is not important. The fractures in the vote of Catholics are what makes it so enticing to both Republicans and Democrats seeking an advantage to move a near "even split" in one direction or another. However, as the research presented here indicates, the Catholicism of a presidential candidate is not likely to create any new gains among Catholic voters.
1. Although the Catholic vote for Lyndon Johnson was nearly as high in 1964 as it was for Kennedy in 1960, one must consider that Johnson's landslide percentage of the presidential popular vote (61.1 percent) is the highest ever achieved. Thus, it was not just Catholics who voted for him in large numbers. At the same time, it is important to consider that Johnson was Kennedy's vice president, less than a year removed from his assassination, and many Catholics may have cast their vote in part as a continuation of Kennedy's legacy.
2. In the Anglo tradition, dating to Blackstone's common law treatises, "But while they [Catholics] acknowledge a foreign power [the Pope], superior to the sovereignty of the kingdom, they cannot complain if the laws of that kingdom will not treat them upon the footing of good subjects" (1769, IV: 54).
3. In the First U.S. Congress, seated in 1789, one member of the Senate was Catholic, as well as two members of the House. The first Catholic Supreme Court justice, Roger Taney of Maryland, was confirmed in 1836 as chief justice.
4. This rapid increase in immigration spawned several nativist anti-Catholic movements beginning in the 1850s including the America party, which is better known as the Know Nothings.
5. Prendergast relies on the estimate that between 85 and 90 percent of Catholics voted for Smith.
6. More ridiculous assertions were made, including that Smith would make Protestant marriages invalid and that Catholicism would be made the official U.S. state religion (DeGregorio 2001, 469).
7. Although the percentage of the U.S. population self-identifying as Catholic has remained consistent during this period, a smaller share of these Catholics reports regular attendance at Sunday Mass. Gallup survey-based estimates indicate 70 percent attending Mass weekly in 1960 as compared to 45 percent in 2004.
8. In his own words Kennedy said, "I do not accept the right of any ecclesiastical official to tell me what to do in the sphere of my public responsibility as an elected official" (Dolan 1992, 422).
9. One indication of how important Kennedy's religion was as an issue in I960 can be drawn from Converse et al. (1961), who report that nearly four in ten survey respondents "voluntarily introduced" some discussion of Kennedy's Catholicism without being prompted by a question about this issue (p. 276).
10. By 2004, overt political anti-Catholicism was nearly nonexistent and Catholics could certainly have no claim today to underrepresentation. In the current 109th Congress, 29 percent of representatives self-identify as Catholic (128 representatives and 24 senators; the largest religious group in Congress). Five of the current Supreme Court justices are Catholic: Antonin Scalia, Anthony Kennedy, Clarence Thomas, and John Roberts, as is the recently confirmed justice, Samuel Alito (Zapor 2005).
11. The U.S. Catholic population includes segments of naturalized citizens-who are primarily Hispanic/Latino-who may speak English as a second language and prefer to be interviewed in Spanish or some other language (Perl, Greely, and Gray 2005). In political studies, any underrepresentation of these voters due to available interview languages, poor sampling, or complete reliance on telephone interviews may skew results. For example, Catholic Hispanics/Latinos (with the exception of Cubans) are more likely to selfidentify as Democrats and describe themselves as liberal (Uhlaner, Gray, and Garcia 2000) and be registered as Democrats (Cain, Kiewiet, and Uhlaner 1991).
12. Although it would be preferable to look at Catholics within the specific diocesan boundaries where a bishop made such a statement, this is not possible using the exit poll data. The cluster sampling and lack of county codes do not allow for a reorganization of the data within Catholic diocese. We use a proxy for states that is reasonable in the sense that these are areas where Catholic voters may have heard about such a statement through the local media or by a reaction from their own bishop to such a statement in a neighboring diocese.
13. Despite claims by the pundits and media, Catholics are not a "kingmaker" or "swing-bloc" segment of the electorate in that they have not always voted for the winning candidate since 1952. It is only clear that a majority of Catholics supported the candidate who would go on to become president in the following years: 1960 (Kennedy), 1964 (Johnson), 1972 (Nixon), 1976 (Carter), 1980 (Reagan), 1984 (Reagan), and 1996 (Clinton).
14. Again it is important to note that the Gallup poll estimated percentage of Catholics reporting that they attended Mass in the last week has declined significantly. Thus, the Republican leanings of this group are becoming a smaller portion of the overall potential self-identifying Catholic electorate.
15. However, Bush won more of the votes of Catholics who did not vote in 2000.
16. Although it is not that Kennedy's Catholicism was more important than partisanship in I960 (Converse 1967, 123), in retrospect it is clear that Kennedy's Catholicism had a much greater impact on the votes of Catholics and non-Catholics than Kerry's did.
17. Many conservative Catholics accept that no party represents Church teachings and doctrine and instead argue that the primacy of some issues-primarily abortion-should lead U.S. Catholics to vote for candidates who are pro-life. Perhaps the most widespread effort to make this case in 2004 was by a group called Catholic Answers and their Voter's Guide for Serious Catholics. This document cites the following five "non-negotiable" issues that must be opposed: abortion, euthanasia, embryonic stem cell research, human cloning, and homosexual marriage. When faced with candidates who do not oppose each, one is instructed to "choose the candidate likely to do the least harm" (2004, 5). Also, this guide instructs, "Do not vote for candidates simply because they declare themselves to be Catholic" (p. 5). Some Catholic Church leaders have echoed these sentiments although rarely with the same level of specificity.
18. After the release of the U.S. bishops' peace pastoral in 1983, the percentage of Catholic General Social Survey respondents who felt the United States spent too much on weapons increased sharply-a remarkable change for a group that had long supported a strong military. But these changes were short lived; Catholic support for reduced military spending returned to near previous levels and differences between active and inactive Catholics evaporated. Stronger influence of Church leaders appears evident in the case of capital punishment. Since the early 1980s or so, more committed Catholics, especially those who are anti-abortion, have become more opposed to the death penalty than other Catholics (Leege and Mueller 2000; Kelly and Kudlac 2000; Perl and McClintock 2001). There is evidence that parishioners may be influenced by their pastors, in particular, on the issue of capital punishment (Bjarnason and Welch 2003). There is little evidence that pastors successfully influence their parishioners by speaking out on abortion (Welch et al. 1993).
19. One major limitation of the NES data is that Catholics are a relatively small subgroup within the larger sample. The Catholic N is not significantly large for reliable study as an isolated subgroup. Given this limitation, our NES models attempt to isolate differences between Catholic respondents as compared to non-Catholics.
20. We are primarily interested in the differences among Catholics voting for a Catholic candidate relative to non-Catholics. Thus, we do not control for other Christian denominations nor do we report a model that measures an interaction effect of Catholic*Mass attendance relative to the attendance of nonCatholics at religious services.
21. At the time, the estimated percentage of the U.S. population self-identifying as Hispanic or Latino was probably less than 2 percent. According to Census estimates, Hispanics/Latinos made up 1.4 percent of the U.S. population in 1940, increasing to 4.5 percent in 1970 (Gibson and Jung 2002).
22. The media edit polls use substantially different questions and/or response sets for age and party identification.
23. Ideally, we would have liked to include responses to specific questions regarding abortion and same-sex marriage. However, the use of these questions for the exit polls was not widespread enough among Catholic respondents to provide a sufficient number of responses for estimation of a multivariate model.
24. In fact, the coefficient indicates a greater likelihood of voting for Kerry, although the result is not statistically significant.
25. Of course it is not only the Catholic voter who has changed. Catholics were in part mobilized to vote for Smith and Kennedy out of a perception of severe anti-Catholic bias in presidential politics from predominantly Protestant candidates and voters. However, overt expressions of anti-Catholicism waned in the 1960s and 1970s. The average difference in the feeling thermometers for "Protestants" and "Catholics" expressed by Evangelical and Mainline Protestants dropped from +27 and +18, respectively, in 1964 to +9 and +7 in 1976.
References
Abramson, Paul R., John H. Aldrich, and David W. Rhode. 1994. Change and continuity in the 1992 elections. Washington, DC: Congressional Quarterly Press.
Bendyna, Mary E. 2000. The Catholic ethic in American politics: Evidence from survey research. Ph.D. diss., Georgetown University, Washington, DC.
Bendyna, Mary E., John C. Green, Mark J. Rozell, and Clyde Wilcox. 2001. Uneasy alliance: Conservative Catholics and the Christian Right. Sociology of Religion 62: 51-64.
Bjarnason, Thoroddur, and Michael R. Welch. 2003. Father knows best: Parishes, priests and American Catholic parishioner attitudes toward capital punishment. Journal for the Scientific Study of Religion 43: 103-18.
Blackstone, Sir William. 1769. Of offences against God and religion. In Commentaries on the laws of England, Book IV, Chapter 4. Oxford, UK: Clarendon Press.
Cain, Bruce E., D. Roderick Kiewier, and Carole Uhlaner. 1991. The acquisition of partisanship by Latinos and Asian Americans. American Journal of Political Science 35: 390-422.
Catholic Answers. 2004. Voter's guide for serious Catholics. Summary. Retrieved July 20, 2005 from http://www.catholic.com/library/voters_guide.asp.
Converse, Philip E. 1967. Religion and politics: The 1960 election. In Elections and political order, edited by Angus Campbell, Philip E. Converse, Warren E. Miller, and Donald E. Stokes, 96-124. New York: Wiley.
Converse, Philip E., Angus Campbell, Warren E. Miller, and Donald E. Stokes. 1961. Stability and change in 1960: A reinstating election. American Political Science Review 45: 269-80.
Crews, Clyde F. 1993. American and Catholic: A popular history of Catholicism in the United States. Cincinnati, OH: St. Anthony Messenger Press.
DeGregorio, William. 2001. The complete book of U.S. presidents. New York: Gramercy Books.
Dolan, Jay P. 1992. The American Catholic experience: A history from colonial times to the present. Notre Dame, IN: University of Notre Dame Press.
Gibson, Campbell, and Kay Jung. 2002. Historical census statistics on population totals by race, 1790 to 1990, and by Hispanic origin, 1970 to 1990, for the United States, regions, divisions, and states. Working Paper Series No. 56. Washington, DC: U.S. Census Bureau.
Gray, Mark. 2005. Self-reported Mass attendance of U.S. Catholics unchanged since 2000. Review of Religious Research 46: 425-26.
Greenberg, Stan, and Matt Hogan. 2005. Reclaiming the white Catholic vote. Unpublished memorandum. Retrieved July 15, 2005 from http://www.democracycorps.com/reports/analyses/ Democracy_Corps_March_2005_Catholic_Analysis.pdf.
Jelen, Ted. 1996. Religion and public opinion in the 1990s: An empirical overview. In Understanding public opinion, edited by Barbara Norrander and Clyde Wilcox, 55-68. Washington, DC: Congressional Quarterly Press.
Kelly, James R., and Christopher Kudlac. 2000. Pro-life, anti-death penalty? America, April 1, pp. 6-8.
Leege, David C. 2000. Divining the electorate: Is there a religious vote? Commonweal, October 20, pp. 16-19.
Leege, David C., and Paul D. Mueller. 2000. American Catholics at the Catholic moment: An analysis of Catholic political patterns, 1952-1996. Paper presented at the 2000 annual meeting of the American Political Science Association, September 1, in Washington, DC.
Leege, David C., and Michael R. Welch. 1989. Religious roots of political orientations: Variations among American Catholic parishioners. Journal of Politics 51: 137-62.
Olson, Daniel V. A. 1997. Dimensions of cultural tension among the American public. In Cultural wars in American politics: Critical reviews of a popular myth, edited by R. H. Williams, 237-58. Hawthorne, NY: Aldine De Gruyter.
Perl, Paul, and Jaime S. McClintock. 2001. The Catholic "consistent life ethic" and attitudes toward capital punishment and welfare reform. Sociology of Religion 62: 275-99.
Perl, Paul, Jennifer Z. Greely, and Mark M. Gray. 2005. How many Hispanics are Catholic? A review of survey data and methodology. Washington, DC: Center for Applied Research in the Apostolate, Georgetown University. Retrieved July 1, 2005 from http://cara.georgetown.edu/Hispanic% 20Catholics.pdf.
Prendergast, William B. 1999. The Catholic voter in American politics. Washington, DC: Georgetown University Press.
Uhlaner, Carole J., Mark M. Gray, and F. Chris Garcia. 2000. Ideology, issues and partisanship among Latinos. Paper presented at the annual meeting of the Western Political Science Association, March 25, in San Jose, CA.
United States Conference of Catholic Bishops. 2004. Catholics in political life. Washington, DC: USCCB.
Wagner, Steve. 2001. Will America bury the hatchet? Craw, January, pp. 12-17.
Wattenberg, Martin P., and Arthur Miller. 1981. Decay in regional party coalitions: 1952-80. In Party coalitions in the 1980s, edited by Seymour M. Lipset, 341-67. San Francisco: Institute for Contemporary Studies.
Welch, Michael R., David C. Leege, Kenneth D. Wald, and Lyman A. Kellstedt. 1993. Are the sheep hearing the shepherds? Cue perceptions, congregational responses, and political communication processes. In Rediscovering the religious factor in American politics, edited by David C. Leege and Lyman A. Kellstedt. Armonk, NY: M. E. Sharp.
Women for Faith and Family. 2005. Statements by bishops on Catholics and political responsibility. Retrieved June 30, 2005 from http://www.wf-f.org/Bishops_Catholics_Politics.html.
Zapor, Patricia. 2005. Catholics, though few in number, have lengthy history on high court. Catholic News Service, July 21. Retrieved August 17, 2005 from http://www.catholicnews.com/data/stories/ cns/0504224.htm.
MARK M. GRAY
PAUL M. PERL
MARY E. BENDYNA
Georgetown University
Mark M. Gray is research assistant professor at Georgetown University's Center for Applied Research in the Apostolate. His research focuses on elections, voter turnout, and the political attitudes and behavior of Catholics within the United States and from comparative perspectives.
Paul M. Perl is research assistant professor at Georgetown University's Center for Applied Research in the Apostolate. His research focuses generally on the sociology of religion and specifically on the religious attitudes and behavior of U.S. Catholics, including the effects of Catholic schooling and changes in Catholic attendance at religious services.
Mary E. Bendyna is research assistant professor at Georgetown University's Center for Applied Research in the Apostolate, where she serves as the center's executive director. Her research focuses on the relationships between religious values and political orientations and includes studies of political attitudes and behaviors of American Catholics and their role in contemporary American politics.
Appendix
The following Catholic bishops, listed in Table A1, made statements during the campaign year that supported the denial of Communion to Catholic political candidates who supported a policy the bishops interpreted to be at odds with Catholic Church doctrine.
The U.S. bishops voted at a semiannual meeting held in June 2004 in Colorado on a statement regarding this issue. By a vote of 183-6, the bishops decided to maintain autonomy regarding the issue of denial of Communion up to the decision of each bishop within his own diocese. A joint statement indicated the following: "Given the wide range of circumstances involved in arriving at a prudential judgment on a matter of this seriousness, we recognize that such decisions rest with the individual bishop in accord with the established canonical and pastoral principles. Bishops can legitimately make different judgments on the most prudent course of pastoral action" (United States Conference of Catholic Bishops, Catholics in Political Life, June 18, 2004).
Copyright Center for the Study of the Presidency Jun 2006