Summarize 4 research articles due anytime before tomorrow!! Must be in own words

Wiley and Society for the Scientific Study of Religion are collaborating with JSTOR to digitize, preserve and extend access to Journal for the Scientific Study of Religion.

http://www.jstor.org

Through Tinted Glasses: Religion, Worldviews, and Abortion Attitudes Author(s): Michael O. Emerson Source: Journal for the Scientific Study of Religion, Vol. 35, No. 1 (Mar., 1996), pp. 41-55 Published by: on behalf of Wiley Society for the Scientific Study of Religion Stable URL: http://www.jstor.org/stable/1386394 Accessed: 06-11-2015 21:30 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

Through Tinted Glasses: Religion, Worldviews, and Abortion Attitudes*

MICHAEL 0. EMERSONt

We lack knowledge on factors intervening between religion and abortion attitudes. Consequently, how religion is associated with abortion attitudes, the full impact of religion, and the most important dimension of religion are all uncertain. Informed by theoretical considerations, I test the plausibility that worldview dimensions serve as intervening factors. Additional hypotheses are also made. All hypotheses are evaluated using General Social Survey data. The findings indicate that worldviews serve as a vital link between reli- gion and abortion attitudes. Moreover, previous work has seriously underestimated the influence of religion, and the degree to which beliefs and decisions are transcendently focused is religion's most central dimension in shaping abortion attitudes.

INTRODUCTION

From confrontations outside abortion clinics, to mass pro- and antiabortion rallies, to in- terpersonal conflicts, abortion tears at the fabric of American life. Not surprisingly, much research has focused on why people diverge on abortion attitudes, and the research agrees on the principal correlates. Roughly, these may be summarized as sociodemographics, reli- gion, and values or ideologies. Despite this agreement, a divide exists between theoretical and empirical work. As a result, significant questions remain about abortion attitudes.

Specifically, theoretical and quantitative works (e.g., Hunter 1991; Luker 1984) dis- cuss factors influencing abortion attitudes as more than a mere collection of variables, each with its own direct impact. Rather, the rich descriptions of these works more closely parallel a process, complete with paths of influence. Yet, with one exception (Tamney, Johnson, and Burton 1992), empirical studies have not examined these paths. In short, the intervening influences on abortion attitudes are marked with uncertainty.

Consequently, our understanding of religion's influence is incomplete. Although the association between religion and abortion attitudes is well established, the process by which this association exists is not. Moreover, because only the direct influence of religion has been tested, its full impact is unknown. As such, the most influential dimension of religion on abortion attitudes remains a mystery.

To address these gaps, theoretical and qualitative works point to the worldview con- cept.1 Many studies on attitudes in general and abortion in particular have discussed the importance of worldviews in shaping activists' positions (Fried 1988; Hunter 1991; Luker

I thank Richard Albares, Jason Dillon, Joni Emerson, Roger Finke, Janet Hope, David Sikkink, Christian Smith, Sylvester Theisen, and the anonymous reviewers for their helpful comments and suggestions.

M Michael 0. Emerson is an assistant professor of sociology in the Department of Cultural Studies, Bethel College, St. Paul, MN 55112.

1. Though I use the term "worldview,' others have been used, such as "symbolic politics' (Fried 1988; Welch, Leege, Cavendish 1995).

? Journal for the Scientific Study of Religion, 1996, 35 (1): 41-55 41

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

42 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

1984; Neitz 1981; Oberschall 1993), yet empirical research has either omitted worldview di- mensions or improperly modeled them. The central goals of this paper, then, are to extend the concept of worldviews to the U.S. public, to test the plausibility of worldview dimensions as mediating factors for this population, and to more fully understand religion's impact on abortion attitudes.

TOWARD A PATH MODEL OF ABORTION ArTrrUDES

Worldviews and the Dimensions Important to Abortion Attitudes

Luker (1984:193) defines worldview as "a set of assumptions about how the world is and ought to be organized.' A worldview allows us to make sense of the world and operate in it. By shaping our interpretations, it predisposes us to, and guides, our specific attitudes and actions.

At least among activists, the abortion debate is animated by those holding worldviews at great odds with each other. I label these ideal-type worldviews as "liberal" and "conservative." The liberal worldview holds that "each human being, born with practically unlimited potential for growth and development, must be allowed ... maximum freedom for self-development and the fullest self-expression" (McNamara 1985:453). Conversely, the conservative worldview holds that each human being, prone to selfishness and harm, best develops and expresses in the context of service and restraint, external regulation, and hier- archical structures (McNamara 1985). Clearly, the liberal worldview emphasizes the innate goodness and primacy of the individual. By contrast, the conservative worldview emphasizes that goodness is best achieved in the appropriate context of others.

According to the qualitative literature on abortion attitudes, worldviews are multidi- mensional (e.g., Luker 1984: ch. 7). The literature is in basic agreement on dimensions shap- ing abortion attitudes: moral reasoning, gender roles, sexuality, and the centrality of chil- dren and of parenthood. Because nonactivists often hold worldviews between the liberal and conservative extremes, and because they may hold contradictory views on different dimen- sions (e.g., see Asher 1988), delineating the separate dimensions is important.

The most general worldview dimension, and serving as a partial root of the others, is moral reasoning (Granberg 1981; Granberg and Denny 1982; Luker 1984; Neitz 1981). Many terms are used to differentiate the polar opposites of moral reasoning, such as transcenden- tal/utilitarian, objective/subjective, and tight-bounded/loose-bounded. No matter the terms, at one end are those whose moral reasoning is rooted in absolutes that specify behavior. These principles are transcendent and valid across space and time. For example, divorce was wrong for all people in 1600, is wrong for all people now, and will be wrong for all people in 2400. At the opposite end are those who use situation ethics. Morality consists of applying a few general ethical principles (such as "be happy" or "choose the greatest good") to each situation. The decision arrived at varies by person and time. In short, morality is relative.

Gender views also contrast (Conover and Grey 1983; Cook, Jelen, and Wilcox 1992; Luker 1984). For some, men and women are intrinsically difference and complementary, meaning each is better suited for particular social roles. Men are better suited for the public world of work, and women, for the private world of home and children. The world is inher- ently divided both socially and emotionally into male and female spheres. This does not sig- nify, however, that such people are opposed to women working. Rather, it means that moth- erhood and homemaking, if one is married and has children, takes precedence over other roles. Moreover, the different yet complementary roles of the sexes entail that raising chil- dren with two parents is better than with one.

For others, the sexes are equal, meaning they are similar. Far from glorifying the role of mother, such a role is seen simply as one among many, and one that is potentially a bar-

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 43

rier to full equality. Moreover, similarity of the sexes means it is not necessarily better that children are raised by two parents of the opposite sex.

Partially because gender ideologies differ (Luker 1984: ch. 7), people also diverge in their views of sexuality (Cochran and Beeghley 1991; Cook et al. 1992; Ginsburg 1984). At the conservative end, sex expresses love between husband and wife and enables procreation. Sex is thus appropriate when people are committed to each other for life (by marriage) and to accepting the possibility of new life being created (even if using contraception). The trust involved in this expression and the potential for creating new life mean sex is inherently transcendent. Persons believing this oppose any form of intercourse outside marriage, in- cluding all forms of homosexual relations.

At the liberal end, value is placed on amative sex, the primary purposes being plea- sure, human contact, and intimacy. Whereas for those at the conservative end sex is inher- ently transcendent, for those at the liberal end it is potentially transcendent. It reaches transcendence "when it dissolves the boundaries between self and other, when it brings one closer to one's partner, and when it gives one a sense of the infinite" (Luker 1984:178). Because this is only achieved through experience, premarital sex is not only acceptable but desirable. Moreover, such a view does hold that sexual relations can be outside male/female relationships.

The fmal dimension is the centrality of parenthood and children (Cook et al. 1992). At the conservative end, a parent is one who has children; at the liberal end, a parent is one who gives a child the best set of emotional, social, and financial resources possible. For hold- ers of the liberal worldview, one should plan when to have children and limit family size to maximize the available resources. Though parenthood can be a satisfying part of life, it need not be central.

Conversely, persons holding a conservative worldview believe a pervasive anti-child sentiment in American society exists. They see evidence for this through such norms as the two-child family and opposition to families with more than three children. Furthermore, they perceive these norms and the strict attention to planning births as indicative of devalu- ing children (Luker 1984:170). Because people holding this worldview dimension highly value children, they are more accepting of alternative family sizes (D'Antonio and Stack 1980; Renzi 1975) and do not support strict planning of the number of or the timing of births.

The Origin of Worldviews

Religion provides answers to questions of ultimate meaning. It also provides guide- lines for how to live in the present world. More broadly, it in large part provides a world- view. By outlining what ought to be (Hunter 1991:58) and by creating and reinforcing group norms through interaction (White 1968), religion has a substantial influence on persons' worldviews. Religion establishes moral rules (Durkheim 1954), serving as a main source for determining "right and wrong, good and bad, just and unjust" (Hunter 1991:59). As Hunter (1991:58) suggests, religion shapes reasoning, provides the meaning, importance, and properness of different social arrangements and institutions, and infuses these "with univer- sal if not transcendent significance." Religious beliefs and practices differ, of course, and this is precisely why we find people possessing diverse worldviews. Persons holding more conser- vative worldviews are active in religious practices and cultures that base their moral author- ity in the transcendent. On the contrary, those holding more liberal worldviews participate in religious and secular cultures that root moral authority in the human being and the pow- ers of reason and revelation (Granberg and Denny 1982; Hunter 1991; Luker 1984).

Of secondary importance, persons' positions in the social structure - be the mecha- nism common life situations or socialization or both -also shape worldviews: "Reasonable

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

44 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

people who are located in very different parts of the social world find themselves differently exposed to diverse realities, and this differential exposure leads each of them to come up with different ... constructions of the world" (Luker 1984:19 1).

How Worldviews Shape Attitudes toward Legalized Abortion

For most of the twentieth century, abortion was removed from public scrutiny by defining it as a question of medical judgment. According to Luker (1984:48), asking people their attitudes toward legalized abortion "would have been as implausible as asking the American public today about its opinions on the efficacy of coronary bypass operations." In short, to have an attitude toward an issue, it must first be defined as a public issue (something people should have an attitude toward). This implies the importance of public debate and "moral entrepreneurs" in shaping attitudes toward abortion (e.g., see Staggenborg 1991: ch. 2).

Hence, an abortion attitude is held not only because it is a logical consequence of a worldview dimension, but also due to frame alignment (Snow, Rochford, Worden and Benford 1986).2 By frame alignment I mean that a worldview dimension leads to an attitude only as far as "moral entrepreneurs" have successfully linked an attitude to a worldview.

To illustrate, let us look more closely at these mechanisms for the moral reasoning dimension. For those who tend toward dichotomous moral reasoning, abortion can be opposed because it takes the life of an innocent human being. In the words of a pro-life tract: "An unborn baby is a being because it exists, and it is human because it has human parents." Thus, "the only choice confronting a pregnant woman is whether to have a live baby or a dead one," and "every time an abortion is performed, an unborn baby is killed" (Minnesota Citizens Concerned for Life, no date). If one reasons absolutely, these arguments are convincing and lead to opposition to abortion. The moral reasoning/abortion attitude connection can be individually created (an individual reasons and arrives at an abortion attitude) or socially created (frame alignment). In the context of frame alignment, moral entrepreneurs successfully link absolute moral reasoning and opposition to abortion. The existence of persuasion literature suggests the importance of frame alignment.

For relative/utilitarian reasoners, the arguments above are foreign and ineffective. Contrast them with an argument in a pro-choice pamphlet: "Personhood at conception is a religious belief.... The religious community will never reach consensus on the definition of 'person' or when abortion is morally justified" (Rothstein and Williams 1983). The argument for legalized abortion rests on the fact that groups differ on the topic. Because we lack a moral consensus, the abortion decision should be left to the individual. For relative reason- ers, this argument is convincing and leads to support for legalized abortion. Again, this con- nection can be individually or socially created. The same holds true for the remaining di- mensions.

THE MODEL AND HYPOTHESEs

From the above discussion, I developed a causal model (Figure 1) and generated fur- ther hypotheses (discussed below). Although the antecedent variables have paths to each

2. Snow et al. (1986:464) define frame alignment as 'the linkage of individual and SMO [social movement organiza- tion] interpretive orientations, such that some set of individual interests, values and beliefs and SMO activities, goals, and ideology are congruent and complementary." I expand this slightly to include not only SMOs but individuals and the media as well.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 45

of the subsequent variables, they are not shown for simplicity's sake.3 I also modeled the expected relationships between the worldview dimensions. As hypothesized, I expect a direct effect of each worldview dimension on abortion attitudes. Furthermore, because the moral reasoning dimension is more general than the others (Hunter 1991), it serves as a partial root of the remaining dimensions. Finally, Luker (1984: 163) intimates that attitudes toward sexuality are rooted in views on the social roles of men and women; I therefore include gen- der roles as a predictor of attitudes toward sexuality.

It its current social context, abortion threatens all of the discussed conservative worldview dimensions (so there is vested interest in opposing legalized abortion), but it is principally seen as wrong from the moral reasoning dimension. Further, from my review of the primary literature, moral reasoning is the most common reference point used by moral entrepreneurs in generating frame alignment. It therefore follows that of the worldview di- mensions, moral reasoning should have the largest direct impact on attitudes toward legal- ized abortion.

The social location control variables listed in Figure 1 are those suggested by qualita- tive researchers and considered significant in empirical research.4 Their impacts are ex- pected to be primarily through the worldviews.

The main antecedent block is religion. The two most common measures of religion in the abortion attitudes literature are affiliation (Catholic, Conservative Protestant, etc.) and frequency of attendance at religious services. These measures make an implicit connection between a theological culture and individual beliefs. A problem with such an assumption, however, is that cultural anthropologists (e.g., Sperber 1975) have found that religious cul- tures do not govern thought and conduct in the manner of legal code (Albares 1992). Such measures assume a radically localized network, overlooking ways in which individual beliefs and practices might be influenced by religious and secular media, other organizations, neighbors, colleagues, and so on. Recent works (e.g., Guth, Smidt, Kellstedt, and Green 1993; Cook et al. 1992; Wilcox 1992) have been much more sophisticated, and I attempt to follow this pattern.

Religion is multidimensional (Roof 1979:40; Johnstone 1992:62). Informed by the work of Himmelfarb (1975), Hunter (1991), Guth et al. (1993), Kellstedt and Smidt (1991), Peek, Lowe, and Williams (1991), and Wilcox (1992), I conceptualize religion along two dimen- sions, religiosity and orthodoxy, each with two subdimensions, public and private (Figure 1). I define the first dimension, religiosity, as "the intensity and consistency of persons' practice to their religion" (Johnstone 1992:58), including the stated strength of their faith. The sec- ond dimension, orthodoxy, is specified as the degree to which beliefs and decisions are cen- tered on transcendent authority (such as Scriptures and the Church). In short, and oversim- plified, religiosity refers to practices and behaviors, orthodoxy, to beliefs and decisions.

The two dimensions of religion are subdivided into public and private. These aspects of one's religious faith have been found distinct (King and Hunt 1975; Peek et al. 1991; Wilcox 1992). Furthermore, considering the above discussion on the error in implicitly as- suming a perfect congruence between public aspects (e.g., affiliation and attendance) and private aspects (beliefs and practices), including both subdimensions is vital.

3. One reviewer suggested that the model may be misspecified by omitting attitudes toward euthanasia, a significant predictor of abortion attitudes in prior research (Cook et al. 1992; Jelen 1984). However, to incorporate this variable as a predictor of abortion attitudes is causally incorrect. I reached this conclusion after review of the primary literature from abortion activist groups. Either the attitude toward euthanasia was linked to the same source as abortion atti- tudes (the worldviews and religion), or it was argued to be the logical consequence of one's abortion attitude.

4. Income, included in many previous studies, was omitted because it is highly variable from year to year, is known to be subject to a high degree of misreporting, and review of the literature revealed that in models including more sophis- ticated measures of religion, income has no significant effect (Cook et al. 1992; Dugger 1991; Tamney et al. 1992; Wilcox 1992).

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

46 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

0~~~~~~~~~~~ 0~~~~~~~~~~

0~~~~~~~

0~~~~~~~~~~~

0

0~~~~

- 0

0

(U 4 E 0 0~~~~~~~~~0

0~~~~~~~~~~~~~~~~~~~~~~

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 47

All four aspects of religion should be positively associated with opposition to abortion. More important, if worldviews intervene between religion and abortion attitudes, the main impact of religion will be through the worldview dimensions. Further, Hunter (1991:58-59) suggests which dimension will be most powerful. According to him, private orthodoxy is the primary source of one's worldview, and, by extension, abortion attitude. Private orthodoxy, then, should have the largest impact of the religion measures.5

To recap, three hypotheses, besides those summarized in Figure 1, are suggested from the theoretical discussion:

(1) Insofar as the world view dimensions are intervening variables of substantive significance, the principal influence of religion (and social location) will be indi- rect, through the worldview dimensions.

(2) Of the worldview dimensions, moral reasoning will have the largest direct im- pact on abortion attitudes.

(3) Of the religion measures, private orthodoxy will have the largest impact on abortion attitudes.

RESEARCH DESIGN

Data

I employ data from the 1988 General Social Survey (GSS), a national, full-probability sample drawn from noninstitutionalized English-speaking persons 18 years of age or older (Davis and Smith 1990). Only the 1988 survey provides the multiple measures needed to op- erationalize the concepts.

The GSS has utilized a split-ballot design since 1987 (Davis and Smith 1990: app. P). To maximize the number of questions included in the survey and yet avoid excessive inter- view length, selected blocs of variables are omitted from a randomly selected subset of ap- proximately one-third of the respondents. The dependent variable and many of explanatory variables are asked of only two-thirds of the respondents. Thus, the potential sample size is reduced from 1481 to 977. Others have found this design not to be a source of bias (Ellison and Sherkat 1993), and auxiliary analysis has confirmed this.

Only 530 of the 977 cases had complete data on all model variables.6 Due to the signif- icant drop in sample size, I conducted further ancillary analysis to test for bias between cases with complete and missing data. Comparisons by race, sex, region, marital status, ur- ban/rural residence, and number of children (variables for which complete data were avail- able) revealed no significant differences (p > .10).

Dependent Variable: Support for Legalized Abortion

Consistent with previous work, I operationalized opposition to legalized abortion with the following questions. Respondents were asked if they think obtaining a legal abortion

5. Though the four aspects of religion undoubtedly influence one another, the literature is not in agreement on causal- ity. For example, does group affiliation influence beliefs (e.g., Ellison and Sherkat 1993; White 1968), or do beliefs in- fluence affiliation (e.g., McIntosh, Alston, and Alston 1979; Finke and Stark 1988)? Because it is beyond the scope of this investigation, I do not attempt to model these relationships.

6. Perhaps due to the sensitive nature of many questions, some respondents did not answer at least one of the queries. Rather than impute their answers, I omitted these respondents.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

48 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

should be possible if (1) there is a strong chance of a serious defect in the baby, (2) the woman is married and does not want any more children, (3) the woman's own health is seri- ously endangered by the pregnancy, (4) the family has a very low income and cannot afford any more children, (5) the woman becomes pregnant from rape, and (6) she is not married and does not want to marry the man. Answers to the questions were summed (1 for no, 0 for yes) to form a 0-to-6 scale that is unidimensional (Gillespie, Ten Vergert and Kingma 1988) and highly valid and reliable (Granberg 1991; Granberg 1984). The alpha reliability (Cronbach 1951), a measure of the internal consistency of a scale ranging from 0 to 1, with higher scores reflecting greater reliability, is .860.

Intervening Variables: Worldview Dimensions

As discussed previously, four worldview dimensions require measurement: appropri- ate gender roles (Gender Roles), proper expression of sexuality (Sexual Relations), the rela- tive importance of children and parenthood (Children), and moral codes (Moral Reasoning). Worldviews, religiosity, and orthodoxy are all complex constructs and, using single variable measures, are highly prone to unreliability. Further, recent research (Tourangeau, Rasinski, and D'Andrade 1991) suggests that such concepts are best operationalized using scales.

For the worldview dimensions and the three religion dimensions, then, I used the fol- lowing strategy. First, relying on face validity, I listed all variables judged to measure the dimension as discussed in the literature. Second, variable responses were reordered so that higher scores reflect greater conservatism, orthodoxy, or religiosity. Next, because "internal- consistency reliability is an indicator of how well the individual items of a scale reflect a common, underlying construct" (Spector 1992:65), I conducted an item analysis to identify a set of questions that formed an internally consistent scale. From the item-analysis results, I created a summated scale. Because the coefficient alphas were less than perfect, I employed the additional strategy of using the alphas to incorporate estimates of measurement error (discussed more fully in the model estimation section). Listed below are the variables used in each worldview scale and the coefficient alpha.

Gender Roles: (1) A husband's job is to earn money, a wife's is to look after home/family; (2) a working mother can establish as warm and secure a relationship as other mothers; (3) a preschool child is likely to suffer if mother works; (4) family life suffers when the woman has a full-time job; and (5 and 6) a single father/mother can bring up a child as well as a married couple. Alpha reliability = .823.

Sexual Relations: (1) Appropriateness of sexual relations between same-sex adults; (2) appropriate of homosexual couples marrying; (3) acceptability of a married person having sexual relations with someone other than his/her marriage partner; and (4) acceptability of a man and a woman having sexual relations before marriage.7 Alpha reliability = .684.

Children: (1) Desirability of a family with four or more children; (2) ideal number of children; (3) watching children grow up is life's greatest joy; (4) a marriage without children is not fully complete; (5) children are more trouble than they are worth; (6) having children interferes with the freedom of parents; (7) it is better not to have children because they are such a heavy financial burden. Alpha reliability = .645.

7. The abortion questions were asked on ballots A and C; the premarital question, on ballots A and B. Because these variables shared only one ballot, the case loss was too great for the premarital sex question to be used without modifi- cation. Though I avoided assigning values on all other variables, due to this vaniable's importance, I imputed responses for the premarital sex question for people on ballot C. Commonly, the mean would be substituted for the missing value. However, because information was available on the other variables in the index, I cross-tabulated the known values of the premarital sex variable with an additive scale of the other three variables. The mean premarital sex score was then calculated by categories of the additive scale.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATrITUDES 49

Moral Reasoning: (1) Morality is a personal matter; society should not force everyone to follow one standard; (2) right and wrong are not usually simple matters of black-and- white - there are many shades of gray; (3) those who violate God's rules must be punished; (4) immoral actions by one person can corrupt society in general; (5) one bad effect of science is that it breaks down people's ideas of right and wrong; and (6) couples do not take mar- riage seriously enough when divorce is easily available (reflecting absolute vs. relative rea- soning). Alpha reliability = .522.8

Control Variables: Social Location

Binary variables are Black, Female, Southern Native, and Married. Persons are clas- sified as southern natives if they lived in the South at age 16. The Married variable is coded as 1 for married and widowed persons, 0 otherwise. To reduce skewness, those with more than 16 siblings were collapsed into one category (16 or more). The Number of Children variable ranges from 0 to 8 or more.

Exogenous Variables: Religion

Recall from the model and previous discussion that I conceptualized religion in four dimensions. Below, I operationalize each.

Private Religiosity. Religiosity is "the intensity and consistency of persons' practice to their religion" (Johnstone 1992:58), including the strength of their faith. The following vari- ables were used in the scale measuring the private aspect: (1) frequency of prayer; (2) fre- quency of Bible reading; (3) closeness felt to God most of the time; (4) degree to which faith is free of doubts. Alpha coefficient = .77 1.

Public Religiosity. By public, I mean that which is believed or conducted within a group context. I operationalized this construct using (1) frequency of attendance at religious services; (2) church member, or not; (3) involvement in church activities other than atten- dance at religious services, or not. Alpha coefficient = .783.

Private Orthodoxy. Previously, I defined orthodoxy as the degree to which beliefs and decisions are centered on transcendent authority. I operationalized the private aspect with a scale containing the following: (1) degree of literal interpretation of the Bible; (2) the Bible's importance in making personal decisions; (3) the importance of the church's or synagogue's teachings in making personal decisions; and (5) the importance of believing in God without doubt. Alpha coefficient = .823.9

Public Orthodoxy. Individuals publicly express their degree of orthodoxy through their religious affiliation. Though it is not without weaknesses, I use Smith's (1986) catego- rization because of greater parsimony and completeness compared with other available mea- sures. He classifies the respondent's religious group as liberal, moderate, or fundamental.

8. Using the worldview variables, I conducted a principal-components analysis. From the results, the dimensions are clearly distinct, except for the empirical suggestion that the moral reasoning and sexual relations items are better viewed as one factor. However, such an analysis operates under the postulates of factorial causation (the correlations between observed variables result from sharing a common factor and not from some being the direct cause of others) and parsimony (the smallest number of factors is preferred). In light of theoretical arguments and qualitative evidence that moral reasoning and sexual relations are distinct dimensions causally linked, both assumptions are violated, thus rendering the empirical suggestion of one factor inappropriate (see Kim and Mueller 1978:22-45).

9. As the reader may note, variable 4 under private religiosity and variable 5 of private orthodoxy appear similar. There are at least two important differences, however. The former asks for actual strength of faith (in this sense, a practice), whereas the latter asks for the importance of faith (a belief). Second, the former asks about faith (it does not specify to what or whom the faith is directed); the latter specifically asks about faith in God. Because Scripture un- equivocally states that God exists, the latter variable is included as a measure of belief in the authority of Scripture.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

50 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

This measure is consistent with Hunter's (1991) contention that people are becoming less concerned with a congregation's specific denominational affiliation and more concerned with its orientation on the progressive/orthodox continuum. The argument is buoyed by one quar- ter of American places of worship having no denominational affiliation (Navigators 1994). I first explored using a series of binary variables, but this provided no information above sim- ply using Smith's categorization. Thus, in the interest of parsimony, I treat this measure as a single variable. Support for a liberal/moderate/fundamentalist continuum has been found by others with respect to issues related to the public and the family (Ellison and Sherkat 1993; Hertel and Hughes 1987).10

Model Estimation

The model is estimated using recursive path analysis. Such classic econometric models assume perfect reliability of the measures. However, the reliability estimates (coefficient al- phas) reported for the scales are far from perfect. Of particular concern are those that are below .7 (Nunnally 1978). Recent research has shown the significant impact on model esti- mates that may result by incorporating coefficient alpha information (e.g., Emerson and Van Buren 1992).

Thus, the estimated model incorporates information provided by the coefficient al- phas. Because perfect reliability implies zero-error variance, the variance of the error term using the coefficient alpha reliability estimates (p) is equal to 1 minus coefficient alpha mul- tiplied by the variance of the measure (i.e., (1-p)*(the measure's variance). One minus the alpha reliability is an estimate of the proportion of the measure's variance due to random measurement error (Bollen 1989: ch. 5). Such information can be incorporated into the model using a structural equation program (I use LISREL). This estimation technique, though, does not necessarily eliminate measurement error (coefficient alpha is a lower- bound estimate of reliability, i.e., a worst-case scenario, and thus tends to overstate the error in the measures). Yet, this technique gives a more accurate picture of the relationships than simply assuming perfect reliability (Bollen 1989: ch. 501

RESULTS

The theoretical model posits that, for the U.S. population, worldviews shape abortion attitudes, and that worldviews are in turn influenced by religion and social location. I tested the model using path analysis incorporating reliability information.12 Figure 2 shows the re- sults of the path analysis, excluding paths not significantly different from zero (p > .05). The

10. The zero-order correlations among the measures of religion range from .218 (public orthodoxy and public religios- ity) to .663 (Private orthodoxy and private religiosity). Removing variable 5 from private orthodoxy (which, on the sur- face, appears similar to variable 4 of private religiosity) does not significantly reduce the latter correlation (.663 to .648). As with the worldview variables, I conducted a principal-components analysis using the religion variables (except public orthodoxy because it was measured with a single variable). The results provided strong support for the empirical independence of the dimensions.

11. I estimated the models both with and without, estimates of measurement error, and the results were substantively similar. The main differences were the attenuation of coefficients in the model assuming perfect reliability, a significant effect of sexual relations attitudes on abortion attitudes, and nearly equal standardized betas for sexual relations attitudes and moral reasoning. Further, I employed the maximum likelihood estimator. To ensure that results were not highly dependent on the selection of an estimator, I also estimated the model using generalized least squares. No substantive conclusions were altered.

12. To improve overall model fit, paths that did not approach statistical significance (coefficients less than their stan- dard error) were set to zero. This did not alter the substantive findings. I explored the possibility of including interac- tions among the antecedent variables; high collinearity precluded their incorporation into the analysis. Additionally, I examined partial plots for nonlinearity, but none was revealed.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 51

raw~~~~~~~~

4,~~~~~~r

r4~~~~~~~~r

0~~~~~~~~~~~~

z~~~~~~~~~~~~~~~' 0~~~~~~~~~~~~~~~~~~~~~~~~~~~I

0 H4

[4 W 2~~~~~~~ [4r r

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

52 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

path coefficients are in standardized form. Overall, the results lend solid support to the hy- pothesized model and confirm the core hypothesized relationships. The model accounts for 47% of the variation in opposition to legalized abortion and fits the data well.13 The world- view dimensions, except attitudes toward sexual relations, significantly influence abortion attitudes. On average, the more one views the sexes as having distinct social roles, sees par- enthood and children as central, and tends toward absolute (rather than relative) moral rea- soning, the greater the opposition to legalized abortion. The influence of sexuality, signifi- cant in a bivariate relationship with abortion attitudes (and in the path model without esti- mates of measurement error) is evidently spurious, disappearing primarily upon controlling for moral reasoning.

The moral reasoning dimension, because it is more general, was expected to have an indirect impact through the remaining worldviews, and this confirmed. Also, of the world- view dimensions, moral reasoning was posited to have the largest direct impact on attitudes toward legalized abortion. Its standardized coefficient (.357) is more than double the next largest dimension coefficient (.154), thus supporting the hypothesis.

The gender roles dimension is explained solely by social location measures; the moral reasoning dimension, except for the race variable, is explained solely by the religion dimen- sions. From the theoretical discussion, that measures of religion have the primary impact on moral reasoning is not surprising. The lack of any religion to gender roles paths, however, appears to contradict recent research showing that fundamentalism increases conservatism on this issue (e.g., Peek at al. 1991). Rather than rebut these findings, though, the present results merely elaborate the relationship. Namely, the impact of religion on gender roles is indirect, being mediated by individuals' moral codes. Lastly, considering the influence of the antecedent measures on opposition to abortion, the impact appears mainly indirect, through the worldview dimensions. Table 1 provides a more direct assessment.

The standardized direct, indirect, and total effects for each variable in the model (except Married and Education, which do not have significant effects on any of the endoge- nous variables) are contained in Table 1. From the table, we may assess two hypotheses. First, if the worldview dimensions are important intervening variables, the main influence of religion and social location will be indirect, through the worldviews. As the table indi- cates, though all 10 exogenous variables listed have a significant influence on opposition to legalized abortion (Total Effect column), only 3 have a direct impact, and 9 of 10 have an in- direct impact. Based on these results, the worldview dimensions are indeed intervening variables of substantive significance.

The results also support the hypothesis that religion's primary impact on abortion at- titudes is indirect. Only public religiosity has a direct impact on abortion attitudes. If the current model estimated direct effects only, religion's impact would be just one-fifth the im- pact found from the path model (.116 vs. the summed .566).

From prior discussion on religion dimensions, the degree to which personal beliefs and decisions are centered on transcendent authority (private orthodoxy) was posited to have the largest impact on abortion attitudes. Examining the total effects, this is clearly supported. The standardized coefficient is three times larger than any other religion coefficient.

13. Overall fit measures are provided at the bottom of the figure. Generally, a model fits the data well when the Goodness of Fit Index (GFI) and the Adjusted Goodness of Fit Index (AGFI) exceed .9. For the remaining measure (22/df), though there is no consensus on what represents a good fit, recommendations range from 3 or less to as high as 5 (see Bollen 1989:278).

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 53

TABLE 1

DIRECT, INDIRECT, AND TOTAL STANDARDIZED EFFECTS ON OPPOSITION TO LEGALIZED ABORTION (N = 530)a

Direct Indirect Total Variable Effect Effect Effect

Controls Black nsb -.057 -.057 Southern Native ns -.026 -.026 Age -.160 .037 -.127 Number of Siblings ns -.012 -.012 Number of Children .101 .030 .131 Female ns -.023 -.023

Religion Public Religiosity .116 ns .116 Private Religiosity ns .045 .045 Public Orthodoxy ns .048 .048 Private Orthodoxy ns .357 .357

Worldview Dimensions Gender Roles .127 ns .127 Sexual Expression ns _c ns Centrality of Children .154 - .154 Morality .345 .094 .439

a Paths significantly different from zero (p > .05), unless otherwise noted. Effects of education and marital status not

breported because neither had significant paths in the models. Not significantly different from zero (p > .05)

c Not applicable

Overall, how persons reason morally and their degree of private orthodoxy emerge as the most dominant variables. Moreover, three of the five variables with the largest influence on the dependent variable are the worldviews. Among the social location measures, the most potent are the number of children one has and the person's age. The noneffect of education is perhaps surprising, but precedented. Using sophisticated measures of religion, Guth et al. (1993) found no significant impact of education on abortion attitudes. However, as they point out, education probably influences abortion attitudes indirectly through religion.

DIscUSSION

To heighten understanding of the mediating influences on abortion attitudes and the full role of religion, I tested the plausibility that, for the U.S. population, worldview dimen- sions intervene between religion and abortion attitudes. The results were clear and decisive. The primary influence of the antecedent factors was via the worldview dimensions. Interestingly, this did not include the dimension of sexual relations. Once estimates for measurement error were incorporated, the influence of sexual relations deferred to the broader moral dimension: moral reasoning. This finding stands in contrast to previous re- search, and should be replicated and more fully explored.

The process by which religion affected abortion attitudes was largely indirect, through the worldview dimensions. Of the four religion measures, only public religiosity had a direct effect. The impact of the remaining religion measures, including the most powerful, was en- tirely indirect. Beyond showing that religion's impact is mediated by worldviews, these findings suggest that in direct-effect models, religion's influence is underestimated.

The analysis also provided strong indication that private orthodoxy (the degree to which beliefs and decisions are focused on transcendent authority) is religion's central factor

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

54 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION

in shaping abortion attitudes. This supports Hunter's (1991) thesis, but more research is needed to learn if private orthodoxy is a central dimension for other socio-moral attitudes, as he contends.

The empirical results and the concept of frame alignment (the social creation of the association between worldviews and abortion attitudes) illuminate the process by which pri- vate orthodoxy affects abortion attitudes. We may ask: Is private orthodoxy important pri- marily (1) because people literally apply specific messages about families and human life en- coded in the teachings of transcendent authority, or (2) because persons rooted in personal orthodoxy tend toward dichotomous moral reasoning, and on this basis can be mobilized by moral entrepreneurs? Frame alignment and the strong relationship between private ortho- doxy, moral reasoning, and abortion attitudes suggest that the latter, at a minimum, are not trivial. Yet, stating this more strongly is difficult, for though frame alignment serves as an important theoretical mechanism, its impact was not empirically tested - but it can be. If frame alignment plays a role, as the public debate and the sophistication of moral en- trepreneurs grow, so too should the strength of the relationship between worldviews and abortion attitudes. Some preliminary evidence for this is provided by Harris and Mills (1985). More generally, we can posit that the greater the public debate, the greater the relationship between worldviews and socio-moral attitudes.

Beyond examining the role of frame alignment, perhaps the most promising line of fu- ture study is to understand how people arrive at a moral authority and subsequent moral reasoning. Also, comparative and longitudinal studies are necessary to address key ques- tions. As persons' religion (and social location) change, do their worldviews change? As frame alignment is cast in alternative lights, do the relationships between worldview dimen- sions and socio-moral attitudes change? Moreover, do the relationships vary by subgroup, such as sex and race (Dugger 1991; Wilcox 1992, 1990)? These important questions should be answered. In the meantime, this study indicates that the abortion controversy will not quickly fade from the American social landscape.

REFERENCES

Albares, R. 1992. Probing the "secret life" of the "last men." Symposium 10:115-22. Asher, H. 1988. Presidential elections and American politics: Voters, candidates, and campaigns since 1952. Chicago:

Dorsey. Bollen, K 1989. Structural equation models with latent variables. New York: Wiley. Cochran, J. and L. Beeghley. 1991. The influence of religion on attitudes toward nonmarital sexuality: A preliminary

assessment of reference group theory. Journal for the Scientific Study of Religion 30:45-62. Conover, P. and V. Grey. 1983. Feminism and the New Right: Conflict over the American family. New York: Praeger. Cook, E., T. Jelen, and C. Wilcox. 1992. Between two absolutes: Public opinion and the politics of abortion. Boulder:

Westview. Cronbach, L. 1951. Coefficient alpha and the internal structure of tests. Psychometrika 16:297-334. Current thoughts and trends. Vol. 10. 1994. Colorado Springs: Navigators. D'Antonio, W. and S. Stack. 1980. Religion, ideal family size, and abortion: Extending Renzi's hypothesis. Journal for

the Scientific Study of Religion 19:397-408. Davis, J. and T. Smith. 1990. General social surveys: Cumulative Codebook, 1972-1990. Chicago: National Opinion

Research Center. Drugger, K. 1991. Race differences in the determinants of support for legalized abortion. Social Science Quarterly

72:570-87. Durkheim, E. 1954. The elementary forms of the religious life. New York: Free Press. Ellison, C. and D. Sherkat. 1993. Conservative Protestantism and support for corporal punishment. American

Sociological Review 58:131-44. Emerson, M. and M. Van Buren. 1992. Conceptualizing public attitudes toward the welfare state: A comment on

Hasenfeld and Rafferty. Social Forces 71:503-10. Finke, R. and R. Stark. 1988. Religious economies and sacred canopies: Religious mobilization in American cities,

1906. American Sociological Review 53:41-49. Fried, A. 1988. Abortion politics as symbolic politics: An investigation into belief systems. Social Science Quarterly

69:137-63.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions

RELIGION, WORLDVIEWS, AND ABORTION ATTITUDES 55

Gillespie, M., E. Ten Vergert, and J. Kingma. 1988. Secular trends in abortion attitudes. The Journal of Psychology 122:323-41.

Ginsburg, F. 1984. The body politic: The defense of sexual restriction by antiabortion activists. In Pleasure and danger: Exploring Female Sexuality, edited by C. Vance, 173-88. Boston: Routledge and Kegan Paul.

Granberg, D. 1991. Conformity to religious norms regarding abortion. The Sociological Quarterly 32:267-75. . 1984. Attributing attitudes to members of groups. In Attitudinal Judgmenit, edited by R. Eiser, 85-108. New York: Springer-Verlag. . 1981. The abortion activists. Family Planning Perspectives 13:157-63.

Granberg, D. and D. Denny. 1982. The coathanger and the rose. Society 19:39-46. Guth, J., C. Smidt, L. Kellstedt, and J. Green. 1993. The sources of antiabortion attitudes: The case of religious politi-

cal activists. American Politics Quiarterly 21:65-80. Harris, R. and E. Mills. 1985. Religion, values and attitudes toward abortion. Journal for the Scientific Study of

Religion 24:137-54. Hertel, B. and M. Hughes. 1987. Religious affiliation, attendance, and support for "pro-family" issues in the United

States. Social Forces 65:858-82. Himmelfarb, H. 1975. Measuring religious involvement. Social Forces 53:606-18. Hunter, J. 1991. Culture wars: The struggle to define America. New York: Basic Books. Jelen, T. 1984. Respect for life, sexual morality and opposition to abortion. Review of Religious Research 25:220-31. Johnstone, R. 1992. Religion in society: A sociology of religion. 4th ed. Englewood Cliffs, NJ: Prentice-Hall. Kellstedt, L. and C. Smidt. 1991. Measuring fundamentalism: An analysis of different operational strategies. Journal

for the Scientific Study of Religion 30:259-78. Kim, J. and C. Mueller. 1978. Introduction to factor analysis. Beverly Hills: Sage. King, M. and R. Hunt. 1975. Measuring the religious variable: National replication. Journal for the Scientific Study of

Religion 14:13-22. Luker, K. 1984. Abortion and the politics of motherhood. Berkeley: University of California Press. McIntosh, W., L. Alston, and J. Alston. 1979. The differential impact of religious preference and church attendance on

attitudes toward abortion. Review of Religious Research 20:195-213. McNamara, P. 1985. The New Christian Right's view of the family and its social science critics: A study in differing

presuppositions. Journal of Marriage and the Family 47:449-68. Minnesota Citizens Concerned for Life. n.d. The abortion debate ... #1. Minneapolis: Minnesota Citizens Concerned

for Life. Neitz, M. 1981. Family, state, and God: Ideologies of the right-to-life movement. Sociological Analysis 42:265-76. Nunnally, J. 1978. Psychometric Theory. 2nd ed. New York: McGraw-Hill. Oberschall, A. 1993. Social movements: Ideologies, interests, and identities. New Brunswick, NJ: Transaction. Peek, C., G. Lowe, and L. Williams. 1991. Gender and God's word: Another look at religious fundamentalism and sex-

ism. Social Forces 69:1205-21. Renzi, M. 1975. Ideal family size as an intervening variable between religion and attitudes toward abortion. Journal

for the Scientific Study of Religion 14:23-27. Roof, W. 1979. Concepts and indicators of religious commitment: A critical review. In The religious dimension: New di-

rections in quantitative research, edited by R. Wuthnow, 17-45. New York: Academic Press. Rothstein, P. and M. Williams. 1983. Choice: Legal abortion - arguments pro & con. Washington, DC: National

Abortion Rights Action League. Smith, T. 1986. Classifying Protestant denominations. GSS technical report no. 67. Chicago: National Opinion

Research Center. Snow, D., E. Rochford Jr., S. Worden, and R. Benford. 1986. Frame alignment processes, micromobilization, and

movement participation. American Sociological Review 51:464-81. Spector, P. 1992. Summated reading scale construction: An introduction. Sage University Paper Series on Quantitative

Application in the Social Sciences, 07-082. Newbury Park: Sage. Sperber, D. 1975. Rethinking symbolism. Translated by A. Morton. Cambridge: Cambridge University Press. Staggenborg, S. 1991. The pro-choice movement: Organization and activism in the abortion conflict. New York: Oxford

University Press. Tamney, J., S. Johnson, and R. Burton. 1992. The abortion controversy: Conflicting beliefs and values in American so-

ciety. Journal for the Scientific Study of Religion 31:32-46. Tourangeau, R., K. Raskinski, and R. D'Andrade. 1991. Attitude structure and belief accessibility. Journal of

Experimental Social Psychokgy 27:48-75. Welch, M., D. Leege, and J. Cavendish. 1995. Attitudes toward abortion among U.S. Catholics: Another case of sym-

bolic politics? Social Science Quarterly 76:142-57. White, R. 1968. Toward a theory of religious influence. Pacific Sociological Review Spring:23-28. Wilcox, C. 1992. Race, religion, region and abortion attitudes. Sociological Analysis 53:97-105. .__ Race differences in abortion attitudes: Some additional evidence. Public Opinion Quarterly 54:248-55.

This content downloaded from 141.210.2.69 on Fri, 06 Nov 2015 21:30:31 UTC All use subject to JSTOR Terms and Conditions