Paper presented at the annual meeting of the American Sociological Association, Chicago, IL, August 2002. We are grateful to Jim Davidson, Kenneth Ferraro, John Stahura, and Christian Smith for comments on a previous version. Additional thanks go to Nicole Klotz for research assistance.
Please cite as: Deflem, Mathieu and Christoph Weismayer. 2002. "Religion and Abortion Attitudes: Findings from the Religion and Politics Survey." Unpublished paper. Presented at the annual meeting of the American Sociological Association, Chicago, IL, August 2002.
We investigate abortion attitudes on the basis of the “Religion and Politics Survey, 1996,” and particularly focus on religious affiliation and religiosity. Results of our analysis suggest that: 1) important differences exist in abortion attitudes across religious affiliations, with Jews and certain Protestants denominations (such as Episcopalians, Presbyterians, Methodists, and Lutherans) holding more lenient, and other Protestants (especially, Pentecostals) more restrictive views on abortion than do Catholics; and 2) religiosity strongly decreases leniency towards abortion. While some other variables also reveal variation in abortion attitudes, they cannot account for the predictive qualities of religious affiliation and religiosity.
Religion plays an important role in accounting for variation in attitudes towards the morality of abortion. Research on abortion attitudes has often found that religion is among the most important factors shaping attitudes towards abortion. Religious belief, especially of Catholics and fundamentalist Protestants, has been discovered to be among the strongest predictors of attitudes towards the morality of abortion (Welch, Leege, and Cavendish 1995). In addition, the strength of religious conviction (religiosity) has been found to increase the likelihood of a more restrictive opinion on abortion (Cochran et al. 1996; Cook, Jelen, and Wilcox, 1993). In this paper, we analyze abortion attitudes on the basis of data from the “Religion and Politics Survey, 1996,” commissioned by the Pew Center for the People and the Press. Findings suggest the centrality of religion in the abortion debate by revealing important differences across religious affiliations, additionally showing that religiosity strongly decreases leniency in attitudes towards abortion. Examining the significance of these findings by estimating their impact under conditions of control of other variables, the Religion and Politics Survey allows measurements of religious affiliation that are precise, especially in respects of membership in various Protestant denominations. Without such precise measurements, some variation in abortion attitudes would have been masked.
ABORTION ATTITUDES: THE RELIGION AND POLITICS SURVEY, 1996
We investigate the impact of religious affiliation and religious importance on abortion attitudes, estimated with ordinary least squares. Based on prior research (Wilcox 1992; Sullins 1999), we hypothesize that some religious affiliations, such as Pentecostals, will have relatively restrictive views of abortion, while other faith groups, including Episcopalians and Jews, will hold comparatively lenient views on abortion. Based on the detailed categories of religious affiliation provided in the Religion and Politics Survey, our analysis is able to estimate more precise measures than the broad categories that have often been used in previous research.
We also expect that religiosity (subjective religious importance) has a significant inverse relationship with leniency towards abortion (Boggess and Bradner 2000; Woodrum and Davison 1992). The salience of religion in the abortion debate should furthermore be shown from the fact that correlations of religious variables remain significant even when controlling for other variables. Among these non-religious independent variables, we do expect some —especially, education, income, and political variables— to correlate significantly with abortion attitudes, but these factors will not account for the influences of religious affiliation and religiosity.
The “Religion and Politics Survey, 1996” was conducted by the Princeton Survey Research Associates. Relying on nation-wide phone interviews with 1,975 adults, 18 years of age or older, the survey measures 152 variables related to religion and politics in the United States. The respondents were selected through proportional stratification by county and telephone exchange within county (i.e., the number of telephone numbers randomly sampled from within a given county is proportional to the county’s share of telephone households in the United States). To analyze the data, we replaced missing data using grand mean substitution for all those variables with less than three percent missing data. For the income variable, of which 10.8% of the responses were missing, predicted mean replacement was employed using gender, employment status, education, age, and region (being from the South) as predictors. We thus analyzed the following measures using weights.
Abortion Attitudes. The dependent variable of our analysis, attitudes toward abortion, is examined using responses to the following question: ‘Which comes closer to your view? Abortion should be generally available to those who want it; Abortion should be available but under stricter limits than it is now; Abortion should be against the law except in cases of rape, incest, and to save the woman’s life; Abortion should not be permitted at all.’ The dependent variable is examined as a continuous variable in the direction of increasing restrictiveness.
Independent Variables. Religious affiliation is based on respondent’s answer to the question, ‘What is your religious preference?,’ with separate questions to probe for more specific affiliations in the case of Christians and Protestants, and is measured as a dummy variable with Catholics as the omitted category. Religious importance or religiosity is measured on the basis of the question, ‘How important would you say religion is in your own life?,” and has three response categories (“very important”, “fairly important,” and “not very important”), in which direction we also measure this continuous variable.
Among our control variables, age and sex are based on self-reported information. We estimate the influence of the variable region to measure differences between respondents living in the South and the other U.S. states. Education is measured on the basis of the question, ‘What is the last grade or class that you completed in school?’. Respondent’s income is measured as a continuous variable in categories based on total family income from all sources before taxes in the previous year. Political views is measured in terms of respondents’ self-description as very conservative, conservative, moderate, liberal, or very liberal, while the variable party affiliation has three categories (Republican, Democrat, and Independent), and is measured as two dummy variables with Independent as the omitted category.
ANALYSIS: RELIGION AND ABORTION ATTITUDES
Briefly reporting the distributions across the categories of the dependent variable for all respondents in the survey, it is found that about a third of the respondents (33.5%) feel that abortion should be generally available, while about a fourth of the sample (24.2%) find that abortion should be available but under stricter limits than are currently in place. Turning to the more restrictive attitudes, about a third of the respondents (31.8%) feel that abortion should be illegal except in cases of rape, incest, and to save the pregnant woman’s life, while 8.8% of those surveyed argue that abortion should not be permitted at all.
Our regression analysis focuses only on those 1,715 respondents in the survey who named their religious affiliation and answered the question on abortion attitudes. Of the 233 respondents excluded from the analysis, 151 were unable or unwilling to name their religious affiliation. An additional 71 respondents were not included in the analysis because their religious background was very broadly declared as “other.” The potential composition of this group could be very diverse, including a wide variety of religious groups such as Muslims, Buddhists, Scientologists, and Hindus, so that analyzing them as a separate category is theoretically not meaningful. An additional 27 respondents were excluded because they did not answer the question measuring their abortion attitude.
The Religion and Politics Survey has the advantage of providing a very precise measure of religious affiliation. Prior research on religion and abortion attitudes sometimes relies on very broad categories, such as Christians and Protestants or categorizations of Protestants, such as Liberal, Moderate, and Evangelical Protestants. There are theoretical arguments not to rely on such broad categories that group the various Protestant denominations (Woodberry and Smith 1998; Steensland et al. 2000). Because denominations generate unique worldviews through socialization into particular ideological orientations (faith) as well as by participation in church gatherings, formal preachings and informal discussions, it is recommended, whenever possible, to use either a more sophisticated categorization based on theological criteria related to creeds and associational criteria related to organizational membership, or to ask respondents for the name of the church and/or denomination to which they belong. The Religion and Politics Survey allows for the latter option.
Turning to the findings from the multivariate analysis, results show that there are considerable differences in abortion attitudes among religious affiliations (see Table 1). Overall, we find that, without controls, Jewish and Orthodox respondents hold considerably more lenient views on abortion than Catholics, as do —to a somewhat lesser extent— Episcopalian, Methodist, Lutheran, and Presbyterian respondents (Model 1). It is striking that both some of the so-called Liberal Protestant groups (Episcopalians and Presbyterians) as well as some denominations of the Moderate Protestants group (Lutherans and Methodists) hold relatively lenient attitudes towards abortion, with some additional variation among the Liberal Protestants. The relative degree of their leniency towards abortion does not correspond to the results we would expect in terms of their categorization into Liberal, Moderate, and Evangelical groups. Methodist and Lutheran respondents, in particular, are considered Moderate Protestants, but, without additional control variables, they are found to hold attitudes about as lenient to abortion than Presbyterians, who are considered Liberal Protestants (as revealed by the unstandardized coefficients of -0.31 and -.030 for Methodists and Lutherans, respectively, and -0.28 for Presbyterians). Furthermore, within the group of Evangelical Protestants, there is considerable variation in abortion attitudes. Especially noteworthy is the relatively high degree of restrictiveness towards abortion among Pentecostals (whose unstandardized coefficient is 0.57). These variations confirm the findings from research by Roof and McKinney (1987:208-212), who similarly found that abortion issues provided the greatest spread in attitudes toward morality issues among religious groups and, also, that there were not only strong divisions among the various categorizations of Protestants (Liberal, Moderate, Black, and Conservative), but also among the various denominations that these categories contain.
Results further show that the second religious variable in our analysis, religiosity, is a highly significant predictor of attitudes towards abortion. The less important religion is to a respondent, the more lenient the respondent’s attitude towards abortion (as indicated by a statistically significant unstandardized coefficient of -0.48 in Model 1).
Next, we find that the influences of both religious variables (affiliation and religiosity) are fairly stable under conditions of control of other independent variables, although some of the latter do account for variation in abortion attitudes (see Model 2). Jewish respondents remain the most lenient in their attitudes towards abortion. The leniency towards abortion among Presbyterians is more pronounced under conditions of control, but it remains close to the average attitudes of Methodists and Lutherans (the unstandardized coefficients are -0.36 for Presbyterians and -0.32 for both Methodists and Lutherans). Under conditions of control, also, Episcopalians and Pentecostals continue to be the two most extreme and opposing Protestant groups on abortion, as the most and the least lenient Protestant denomination, respectively (with respective unstandardized coefficients of -0.38 and 0.46). The influence of religiosity on abortion attitudes remains almost unaffected under conditions of control (as indicated by an unstandardized coefficient of -0.43 in Model 2).
In terms of the control variables, respondent’s region, age, and race are found to not significantly influence attitudes towards abortion. Although respondent’s sex is related significantly to abortion attitudes, with males holding slightly less lenient views on abortion than women do, the difference is statistically significant only at the lowest level of significance, confirming that the relationship between gender and abortion attitudes is ambiguous and weak at best (Hertel and Russell 1999; Strickler and Danigelis 2002). The absence of racial differences in abortion attitudes corresponds to research findings that in recent years, unlike the 1970s and 1980s, racial disparities in abortion attitudes have been disappearing under influence of a trend towards increasing restrictiveness among whites —especially white males— and increasing leniency among Black females (Boggess and Bradner 2000; Gay and Lynxwiler 1999).
Turning to the other control variables, level of education and income relate as expected positively to leniency on abortion, as does holding liberal political views. The influence of party affiliation is more ambiguous, in that being a Democrat does not influence abortion attitudes significantly in comparison to the Independent group, while self-reported Republicans comparatively hold more restrictive attitudes towards abortion.
Corroborating evidence of our perspective that merging religious affiliations in broader categories hides variation in abortion attitudes is provided by an additional multivariate analysis that groups the Protestant denominations into Liberal, Moderate, and Evangelical categories (Models 3 and 4 in Table 1). Results of this analysis generally show the unsurprising fact that Liberal Protestants are more lenient towards abortion than Moderate Protestants, who, in turn, are more lenient than Evangelical Protestants. However, many of the differences that exist among the various Protestant denominations are no longer shown. Without additional control variables, it is no longer shown that Methodists and Lutherans, who make up the group of Moderate Protestants, are significantly more lenient towards abortion than Catholics are. Under conditions of control, the relative leniency among Moderate Protestants becomes statistically significant, but without much of an increase in strength of the relationship (as indicated by a change in the unstandardized coefficient from -0.10 to -0.14 for the Moderate group).
Thus, while the influences of religiosity and the control variables are confirmed, and while the variance explained for the models as a whole remain stable, the analysis with broad religious categories masks some of the substantive influences from religious affiliation, specifically the fact that some of the denominations that make up the various Protestant categories hold views substantially different from the average value of their respective group. In particular, when Methodists and Lutherans are combined into the broader category of Moderate Protestants, it is no longer shown that, compared to Catholics, they are almost as lenient in their abortion attitudes as are Presbyterians. And although the coefficient of the Moderate group reaches statistical significance under conditions of control (-0.14), there remains a substantial difference with the Liberal Protestant category (-0.26), although the values of the various denominations that make up these categories are not that far apart. Furthermore, there is considerable variation in abortion attitudes among the denominations that make up the Evangelical group. Whereas Evangelical Protestants as a group are found to hold more restrictive views on abortion than Catholics at a statistically significant level, this difference masks the much more restrictive attitudes of Pentecostals, as compared to Catholics, and the absence of any such statistically significant differences among the other Protestant denominations in the Evangelical group.
Finally, the frequency distributions of abortion attitudes per religious affiliation (which are meaningful given their stability under conditions of control) further illustrate the differences in abortion attitudes that exist among religious groups, in general, and Protestant denominations, in particular (Table 2). These findings particularly show the relative leniency towards abortion among Moderate Protestant groups and additionally reveal variation within the Evangelical group, with a striking polarization in the Congregational/United Church of Christ group.
This analysis relied on the “Religion and Politics Survey, 1996” to examine the influence of two important religious variables, religious affiliation and religiosity, on abortion attitudes. The results of our analysis suggest the salience of religion in shaping abortion attitudes. First, there are important differences in abortion attitudes across religious affiliations. Jewish and Orthodox believers hold among the most lenient views on abortion, with Episcopalians, Presbyterians, Methodists, and Lutherans also being more lenient towards abortion than Catholics. Pentecostal Protestants hold very restrictive viewpoints on abortion, while other Protestant denominations hold views not statistically different from Catholics. These differences in abortion attitudes across religious affiliations are more substantial than a crude categorization would suggest, especially inasmuch as there is considerable disparity in attitudes among the Protestant denominations, which in recent years may even have further polarized (Evans 2002; Sullins 1999).
Thus, our research indicates that the significance of religion in accounting for abortion attitudes is even more pronounced than prior studies have suggested. Our findings also suggest that religiosity, measured as the subjective importance attributed to religion, is related positively with more restrictive attitudes towards abortion. Religiosity may crosscut the influence from religious affiliation, as previous research has found (Cochran et al. 1996; Petersen 2001). Finally, the salience of religion in attitudes towards abortion is additionally confirmed by the fact that while some other variables also exert an influence, they cannot account for the influences of religious affiliation and religiosity.
The relevance of religion has been found to be a persistent factor in the abortion debate. Research from the early 1970s onwards has found that religion is the single most important predictor of attitudes toward abortion among the factors considered, and most recent studies have confirmed the centrality of religion as an influence on abortion attitudes, especially relative to the much lower predictive powers of gender and feminist ideology (Evans 2002). Most strikingly perhaps is the fact that even when religion is not the central focus of research on abortion attitudes, religious variables are nonetheless often found to be the strongest among all predictors (Misra and Hohman 2000). What further research will need to focus on are the mechanisms at the institutional and ideological levels that might account for the influences of religious variables and their variable meaning across different faith groups.
Boggess, Scott, and Carolyn Bradner. 2000. “Trends in Adolescent Males’ Abortion Attitudes, 1988-1995: Differences by Race and Ethnicity.” Family Planning Perspectives 32:118-123.
Cochran, John K., Chamlin, Mitchell B., Beeghley, Leonard, Harnden, Angela, Blackwell, and Brenda Sims. 1996. “Religious Stability, Endogamy, and the Effects of Personal Religiosity on Attitudes toward Abortion.” Sociology of Religion 57:291-308.
Cook, Elizabeth A., Jelen, Ted G., and Clyde Wilcox. 1993. “Catholicism and Abortion Attitudes in the American States.” Journal for the Scientific Study of Religion 32:223-230.
Emerson, Michael O. 1996. “Through Tinted Glasses: Religion, Worldviews, and Abortion Attitudes.” Journal for the Scientific Study of Religion 35:41-55.
Evans, John H. 2002. “Polarization of Abortion Attitudes in U.S. Religious Traditions, 1972-1998.” Sociological Forum 17(3):397-422.
Gay, David, and John Lynxwiler. 1999. “The Impact of Religiosity on Race Variations in Abortion Attitudes.” Sociological Spectrum 19:359-377.
Hertel, Bradley R., and Mark C. Russell. 1999. “Examining the Absence of a Gender Effect on Abortion Attitudes: Is There Really no Difference?” Sociological Inquiry 69(3):364-381.
Miller, Alan. 1996. “The Influence of Religious Affiliation on the Clustering of Social Attitudes.” Review of Religious Research 37:123-36.
Petersen, Larry R. 2001. “Religion, Plausibility Structures, and Education’s Effect on Attitudes toward Elective Abortion.” Journal for the Scientific Study of Religion 40(2):187-204.
Roof, Wade C., and William McKinney. 1987. American Mainline Religion: Its Changing Shape and Future. New Brunswick, NJ: Rutgers University Press.
Steensland, Brian, Park, Jerry Z., Regnerus, Mark D., Robinson, Lynn D., Wilcox, W. Bradford, and Robert D. Woodberry. 2000. “The Measure of American Religion: Toward Improving the State of the Art.” Social Forces 79(1):291-318.
Strickler, Jennifer, and Nicholas L. Danigelis. 2002. “Changing Frameworks in Attitudes Toward Abortion.” Sociological Forum 17(2):187-201.
Sullins, Paul. 1999. “Catholic/Protestant Trends on Abortion: Convergence and Polarity.” Journal for the Scientific Study of Religion 38(3):354-369.
Welch, Michael R., Leege, David C., and James C. Cavendish. 1995. “Attitudes toward Abortion among U.S. Catholics: Another Case of Symbolic Politics?” Social Science Quarterly 76:142-157.
Wilcox, Clyde. 1992. “Race, Religion, Region and Abortion Attitudes.” Sociological Analysis 53:97-105.
Woodberry, Robert D., and Christian S. Smith. 1998. “Fundamentalism at al.: Conservative Protestants in America.” Annual Review of Sociology 24:25-56.
Woodrum, Eric, and Beth L. Davison. 1992. “Reexamination of Religious Influences on Abortion Attitudes.” Review of Religious Research 33:229-243.