In the last two decades studies on the correlates of have proliferated (Hertel, et al., 1974; Renzi, 1975; Granberg, 1978; McIntosh, et al., 1979; D'Antonio and Stack, 1980; Benin, 1985; Harris and Mills, 1985; McCutcheon, 1987; Jelen, 1988). Although these studies have clarified which variables are related to , the reasons for these relationships have been largely left unexplained. Thus, much of the literature lacks theoretical coherence.
In this study, an effort is made to contribute to the understanding of by clarifying and adding to the theoretical base underlying the existing literature. More generally, the study attempts to contribute to an understanding of (a) attitudinal consistency, (b) structural factors, and (c) social interaction in attitude formation. Specifically, in addition to examining the structural effects of demographic control variables on , the notion of ritual interaction, stemming from the writings of Collins (1981), Durkheim (1912/1954), and Goffman (1967), will be used to examine the effects of religiosity, feminism, and conservativism on . I will show that there is not only a direct relationship, as many prior studies assert, but also that there is an interaction effect of the amount of interpersonal contact with others. That is, interaction with others intensifies the effect of other attitudes on .
A THEORETICAL MODEL

Attitudinal Effects

In the literature on abortion attitudes (e.g., Granberg, 1978; Barnartt and Harris, 1982; Granberg and Granberg, 1981; Benin, 1985), several general clusters of attitudes--religiosity, conservativism, and feminism--show consistent correlations with specific attitudes toward abortion. Research has shown that religious and conservative attitudes lead to opposition to abortion, and feminist attitudes lead to higher levels of approval of abortion.

It is important to distinguish between religion and religiosity. Both variables tend to be strongly correlated with abortion attitudes. Several studies have examined the relationship between religion (identification or affiliation with a specific religion or denomination) and different attitudinal and behavioral variables using reference group theory (Jacques and Chason, 1978; Hill and Dodder, 1978; Bock, et al., 1983, 1987; Cochran, et al., 1988; Beeghley, et al., 1990). This theory predicts that the groups to which people belong will serve as a source of norms guiding attitudes and behavior (Singer, 1981). Bock, et al. (1987) and Cochran et al. (1988) found that the effect of religiosity on alcohol use is strongest among individuals who are affiliated with religious denominations which forbid alcohol use, arguing that an individual's religion serves as a reference group guiding behavior. The authors argue that

The degree to which a group or collectivity serves as a reference group for an individual is a positive and additive function of:

the degree of similarity between the status attributes of an individual and other members;
the degree to which an individual's values and beliefs agree with those of other group members;
the degree of clarity in a group's values and beliefs;
the degree to which an individual is in sustained interaction with other group members;
the degree to which an individual defines group leaders as significant others. (Cochran, et al., 1988, p. 258)

They further argue that churches tend to meet these conditions and should, therefore, serve as reference groups for individuals identifying themselves with a certain religion. In these studies on alcohol use, the split on the independent variable, religious preference, was made based on the dependent variable (i.e., whether or not the religion/denomination forbids drinking); however, that split is not appropriate here because the division among churches on the issue of abortion is not as clear. Very few religions, if any, explicitly condone abortion. Either they take a strong stand against it or they remain non-committal. In the studies on alcohol use, the researchers found that religion was not a strong predictor of alcohol abuse because there are only norms against alcohol abuse; no churches support alcohol abuse. In addition, Ebaugh and Haney (1978) compared attitudes of members of conservative and liberal churches toward abortion and found a relationship between frequency of attendance and attitude toward abortion only in conservative churches. The relationship was not significant for liberal churches. Yet another problem with using religious identification as a major variable in this particular model is that people may identify themselves with a religion, even if they hold few of the beliefs and never go to church. For these reasons most abortion attitude research includes both religious identification and religiosity as independent variables (e.g. Clayton and Tolone, 1973; Peterson and Mauss, 1976; Granberg and Granberg, 1981; Benin, 1985; Jelen, 1988).

Some research on abortion focuses on conservativism and feminism as well as religiosity. Granberg (1978) argues that a conservative sexual morality is the primary determinant of opposition to abortion, while Stinchcombe, et al. (1980) and Sawyer (1982) assert that attitudes toward abortion reflect underlying social and political philosophies. Recent research has shown that both factors are at work (Jelen, 1984, 1988; McCutcheon, 1987). The implicit explanation for these relationships is that people strive for consistency not only between attitudes and behavior, but also among their various attitudes. That is, if people have conservative attitudes in one area, their other attitudes will be conservative as well.

Ritual Interaction Effects

The idea of ritual interaction has been discussed in detail by Collins (1981), who concentrates mainly on the writings of Durkheim (1912/1954) and Goffman (1967). Collins begins with a general description of Durkheim's conception of rituals, noting that when people interact, especially ritual interaction like a church ceremony, intense feelings and emotions are generated. In particular:

Participating in rituals gives a feeling of strength and support, which individuals can then use in their daily lives....Rituals pay off in the form of emotional energy, which also carries with it feelings of morality about what one is doing. Rituals generate feelings of self-righteousness. (Collins, 1981, p. 192)

Hence, through common mood, shared focus, and symbolization, rituals generate emotional arousal and the development of moral righteousness. In discussing the work of Durkheim and Goffman, Collins distinguishes between macro and micro conceptions of interaction rituals (i.e., large versus small groupings) and also between institutionalized rituals (e.g. political rallies and weddings) and natural rituals (e.g. friends talking). Durkheim tended to focus more on formal macro rituals, whereas Goffman describes the effect of ritual on informal interactions. Although these "interpersonal rituals" tend to be short and temporary, they occur throughout everyday interactions and have a great effect on the flow of social life.

Carrying out casual talk... is the creation of a little temporary cult, worshiping the reality of what is being talked about for the moment. It assembles a little group, focuses attention, and builds up a common mood.... These fleeting rituals of everyday life, charged with Durkheimian symbolism, produce lines of social acceptability and unacceptability.... (Collins, 1981, p. 198)

Therefore, micro rituals stretched out over time may be just as important as macro rituals, even if the emotional energy produced by them is not as intense.

The importance of the amount of interaction on attitude formation has not been completely ignored in the abortion literature. Several researchers have explicitly discussed this variable in explaining the effect of religiosity on abortion (see, for example, Greeley, 1963; White, 1968). Most view frequency of church attendance as a proxy for rate of social interaction, arguing that it is interaction with others in the church that affects an individual's attitudes rather than religious socialization. However, I would argue that religiosity should be conceptualized as an attitude rather than as a behavior. If interaction has a strong effect, it should be directly incorporated into the model as a measured variable.

Religiosity, conservativism, and feminism have all been shown to influence attitudes toward abortion. The theoretical justifications offered in previous studies (e.g. attitudinal consistency and underlying values) may explain the direct relationships, but the effect of each of these three variables is amplified through interpersonal contact with others. That is, people may enter into interactions with a certain degree of intensity, or emotional energy, about an attitude; and through interaction, the strength of emotion increases and the effect of the original attitude is intensified. Therefore, the more people interact the greater will be the strength of their attitudes and the greater their effect on other attitudes. The focus here is on natural micro interaction rituals. Specifically, the more people interact with others, the stronger the effect of religiosity, conservativism, and feminism on attitudes toward abortion controlling for SES, religion, age, race, sex, rural/urban differences, and marital status.

Control Variables

The control variables included in this model are religion, socioeconomic status (SES), gender, age, race, marital status, and size of place where the respondent lives. These variables are included for two reasons. First, because they give people specific interests and hence attitudes; that is, they have direct effects on attitudes toward abortion. Second, they not only have direct effects, but they also have exogenous correlations because (1) they predispose people to certain attitudes on other related issues (e.g. capital punishment or euthanasia), and (2) they predispose people to different ritual interaction patterns. Catholics are more likely to oppose abortion than are non-Catholics (Benin, 1985; Jelen, 1988); however, I also include a measure of fundamentalism because some Protestant denominations are also strongly anti-abortion and should not be included in the non-Catholic group. SES is also strongly associated with attitudes toward abortion. Most research has examined the separate effects of education, income and occupational prestige, finding that, at higher levels, these variables lead to a lower level of opposition to abortion (Hedderson, et al., 1974; Arney and Trescher, 1976; Skerry, 1978; Ebaugh and Haney, 1980). Most studies have shown that women are more favorable toward abortion than men (Hertel, et al., 1974; Blake, 1971), although some have found men to approve of abortions more than women (Rao and Bouvier, 1971). In general, individuals who are white, never-married, younger, or urban are more likely to approve of abortion than are individuals who are non-white, ever-married (married, divorced, or separated), older, or rural (Arney and Trescher, 1976; Barnartt and Harris, 1982; Evers and Mcgee, 1980; Benin, 1985).

The discussion above leads to the following hypotheses:

Men will oppose abortion more than women.
Respondents who are currently married or have ever been married will oppose abortion more than those who have never married.
Older respondents will oppose abortion more than younger respondents.
White respondents will oppose abortion more than non-white respondents.
The more rural an area respondents come from the more they will oppose abortion.
The greater respondents' SES, the greater their opposition toward abortion.
Catholics and/or Fundamentalists will oppose abortion.
More conservative individuals will oppose abortion.
The more religious respondents are, the more they will oppose abortion.
The less feminist respondents are, the greater their opposition.

The direct effect of interpersonal contact on attitude toward abortion will not be large, rather the interaction effects of interpersonal contact with religiosity, conservativism, and feminism, should be significant. That is, interpersonal contact and each of the three attitudinal independent variables work in combination to affect attitudes toward abortion. Interpersonal contact increases the effect of other attitudes on attitudes toward abortion.


METHODS

Sample

The data for this research come from the 1989 National Opinion Research Center General Social Survey, which is based on a full probability sample of English-speaking persons eighteen years of age or over in non-institutionalized settings in the continental United States. The entire sample includes 1537 respondents, but a sub-sample of the respondents (N=474) were used for the analysis, consisting of those who answered all the questions utilized for the model. T-tests of control variables show no significant differences between those who answered the abortion items and those who did not.

Measures

The survey included seven abortion questions asking respondents whether they think it should be possible for a woman to obtain a legal abortion "(1) if there is a strong chance of serious defect in the baby, (2) if she is married and does not want any more children, (3) if the woman's own health is seriously endangered by the pregnancy, (4) if the family has a very low income and cannot afford any more children, (5) if she became pregnant as a result of rape, (6) if she is not married and does not want to marry the man, or (7) if the woman wants it for any reason." (NORC, 1989). Although studies have shown that level of approval varies with the reason for abortion, the effects of attitudes and interpersonal contact should be the same regardless of the reason for abortion. Therefore, the abortion items were formed into one scale of opposition to abortion, ranging from 0 to 7, consisting of the number of disapprovals. Cronbach's alpha for the scale is .89, and the Guttman Split-Half Coefficient is .92, both indicating high reliability.

The measure for SES was created using factor analysis on four questions. Respondents were asked, "In which of these groups did your total family income, from all sources, fall last year before taxes?" The twenty response categories ranged from less than $1000 to $60,000 and over. Respondents were asked a series of questions to determine the highest year of education completed. Responses ranged from no formal schooling (00) to eight years of college (20). Respondents were asked a series of questions to determine occupational prestige based on the two digit Hodge, Siegel, and Rossi prestige score. Categories ranged from 10 to 89. Respondents were asked, "If you were asked to use one of four names for your social class, which would you say you belong in: the lower class (1), the working class (2), the middle class (3), or the upper ...