Religious Influence and the Politics of Family Decline Concern: Trends, Sources, and U.S. Political Behavior

by Clem Brooks
Religious Influence and the Politics of Family Decline Concern: Trends, Sources, and U.S. Political Behavior
Clem Brooks
American Sociological Review
Start Page: 
End Page: 
Select license: 
Select License





Indiana University, Bloornington

One of the most widely debated issues in contemporary sociology has been how to interpret patterns of family change in the United States during the past four de- cades. Much of these debates focus on a thesis advanced by a number of scholars and political activists-that such features of family structure as high divorce rates and an increasing proportion of single-parent families have led to the decline of the family. Although past research has examined the causes and historical trends in- volved in family change, scholars have neglected important questions about family attitudes that have been raised in recent debates. Have levels of public concern with family decline increased over time? If so, what are the sources of these attitudes? And have changing levels of public concern with family decline led to the emergence of a new political cleavage? This study applies a theory of religious influence to answer these questions. Results show that public concern with family decline in- creased steadily after 1980, leading to a new and increasingly large cleavage in presidential elections. The analyses also find that high levels of concern with family decline are concentrated among evangelical Protestants who attend church regu- larly. In addition to extending sociological research on family change in new and fruitful directions, these results bear productively on theory and research in political sociology and the sociology of religion. I discuss their relevance to long-standing debates over political trends among evangelical Protestants and the influence of

Christian Right movement organizations.

0NE OF THE MOST widely debated issues in contemporary sociology has been how to interpret patterns of family change in the United States during the past four decades. At the center of recent contro- versies is the claim that trends through the early 1980s such as the growing proportion of single-parent families and rising divorce

Direct all correspondence to Clem Brooks, De- partment of Sociology, 1020 E. Kirkwood Ave., Indiana University, Bloomington, IN 47405- 7103 ( Data and code- books from the American National Election Stud- ies were provided by the Inter-university Consor- tium for Political and Social Research and by the Center for Political Studies. The usual disclaim- ers apply. I thank Elizabeth Armstrong, David Brady, Andrew Greeley, Alex Hicks, Michael

rates contributed to a decline of the tradi- tional nuclear family. In addition to state- ments by social scientists (see Gill 1997; Glenn 1987a, 1993; Popenoe 1993, 1994; Wilson 1993), interpretations of family de- cline have also been advanced by a number of Christian Right leaders (Christian Coali- tion 1995; Dobson and Bauer 1990; Falwell 1992; Reed 1994, 1996). While not neces- sarily agreeing on all normative assumptions or policy recommendations, both literatures offer interpretations of family decline that

Hout, Jeff Manza, Douglas Maynard, Patricia McManus, Brian Powell, Robert Robinson, Judith Stacey, Mitchell Stevens, Robert Woodberry, and the ASR Editors and reviewers for their comments and suggestions.

have been widely debated in contemporary sociology.'

Although far from achieving resolution, scholarly debates over family decline have usefully highlighted the role of married women's declining fertility in explaining the growing proportion of children born into single-parent households (Bane and Jargowsky 1988; Luker 1996), the relevance of long-term historical trends in family struc- ture and divorce (Cherlin 1988; Furstenberg 1998; Greeley 1991; Spain and Bianchi 1996), and the mediating role of public poli- cies that provide for the welfare of women and children (Lichter 1997; Skolnick 1991; Strober and Dornbusch 1988). However, questions relating to public attitudes toward family decline have been neglected. The fo- cus of research on demographic or economic aspects of family change (e.g., structure, fer- tility, and marital dissolution) has not been complemented by investigations of attitudes toward family decline in the United States. This lack of research is significant on two counts.

First, while some proponents of the fam- ily decline thesis hypothesize a growing public concern with family decline (Popenoe 1993; Wilson 1993), they provide no direct evidence. Second, it is likely that changes in family structure are not isomorphic with re- spect to (subjective) changes in attitudes and perceptions. If public attitudes toward the is- sue of family decline have consequences for the behavior of individuals or the organiza- tion of institutions independently of the in- fluence of demographic and economic fac- tors, their cultural or political meanings are worthy of detailed study.

In this study, I seek to extend research on family change to address questions about public attitudes. I emphasize that my goal is not to adjudicate ongoing debates over the effects of family change on the welfare of members or the stability of the institution. Instead, my focus is on the subjective di-

Four major symposia on family change have been published in social scientific journals: Journal of Family Issues (1987, vol. 8, pp. 347-476); Journal of Marriage and the Family (1993, vol. 55, pp. 23-38; and vol. 55, pp. 527-55); and Contemporary Sociology (1998, vol. 27, pp. 223- 37).

mensions of family change. I ask the follow- ing questions: (1) Are there trends in the level of public concern with family decline?

(2) What are the causal sources of concern with family decline? (3) Have changing lev- els of concern over family decline led to the emergence of a new political cleavage?

I apply a theory of religious influence to answer these questions. In addition to ex- tending sociological research on family change in new and fruitful directions, my results bear on theory and research in politi- cal sociology and the sociology of religion. In the paper's first section, I develop a more precise conceptualization of the phe- nomenon of family decline, considering four elements discussed by proponents of the thesis. I then discuss a theoretical ap- proach to religious influence that identifies causal processes which potentially explain the rise and partisan political relevance of public concern with family decline. After discussing data, measures, and models, I present the analyses. The broader signifi- cance of the results is discussed in greater detail in conclusion.




What are the components of family decline in the United States? Proponents of the fam- ily decline thesis identify four social pro- cesses as indicators of the hypothesized de- cline of the family.

DIVORCE. The growing rate of marital dissolution from the 1960s through the early 1980s is widely viewed as central to the hy- pothesized decline of the family. Divorce is said to have negative effects on the stability of the nuclear family, leading to social prob- lems such as single-parent families and in- adequate child socialization. Underlying these concerns is the preeminent value that most proponents of the family decline thesis assign to the ideas of lifelong marriage and continuity in families (e.g., Dobson and Bauer 1990; Popenoe 1994).


FAMILIES. A second in- dicator of family decline is the prevalence of single-parent families. Proponents of the family decline thesis view this family form as less viable than the traditional two-parent institution on economic and oftentimes nor- mative grounds (e.g., Blankenhorn 1995; Murray 1993; Whitehead 1993). Policy pro- posals relating to changes in tax and divorce laws are often linked to the assumption that single-parent families are a major social problem to be addressed through govern- ment or legal reforms.


SOCIALIZATION. Many pro- ponents of the family decline thesis argue that children's socialization is less adequate than in the past, leading to social problems such as juvenile crime and poor school per- formance (Whitehead 1993). Inadequate child socialization is frequently viewed as a product of rising rates of divorce and a growing number of single-parent families (Gill 1997; Reed 1996), suggesting causal connections between these elements of fam- ily decline.

CHILDPOVERTY. Finally, proponents of the family decline thesis often view the eco- nomic welfare of children as linked to changes in family structure. Like children's socialization, child poverty is often hypoth- esized as both causing and being caused by other components of family decline such as single parenthood (Whitehead 1993; Wilson 1993). Some commentators identify child poverty as part of an hypothesized cycle of marital dissolution, inadequate child social- ization, and single-parenthood (Murray 1993).


Compared with other Western democracies, the United States continues to be character- ized by exceptionally high levels of religious commitment and religious-organizational strength (Ladd 1999; Sherkat and Ellison 1999; Verba, Schlozman, and Brady 1995, chap. 17). Religion shapes individuals' atti- tudes toward family and gender (Thornton

1985), and studies have found that denomi- national membership, church attendance, and belief in specific doctrines have large effects on behaviors associated with social- ization style, fertility, and the division of do- mestic work (Burns, Schlozman, and Verba 2001, chap. 9; Sherkat and Ellison 1999; Woodberry and Smith 1998).

Causal assumptions regarding the likely effects of religion on family processes and attitudes constitute a theory of religious in- fluence. This general theoretical approach provides a potential means of understanding the mechanisms behind changing levels of concern with family decline. Recent analyti- cal innovations have advanced the sociology of religion by identifying three distinct mechanisms transmitting the influence of re- ligion.

The first influence is denominational mem- bership. Affiliation with a denomination pro- vides individuals with a source of established ideas, authority, and organization (Steens- land et al. 2000; Wuthnow 1999). Compared with the social liberalism of mainline Protes- tant denominations, specific features of evangelical Protestantism-higher levels of belief in biblical inerrancy, a greater willing- ness to view the family as a sacred institu- tion, and broader support for traditional gen- der roles (Wald 1996; Woodberry and Smith 1998)-may predispose evangelicals to be concerned with family de~line.~

Such factors are likely to result in significant differences in levels of concern with family decline among members of different religious tradi- tions, and especially among members of evangelical versus non-evangelical denomi- nations.

A second way in which religion may in- fluence concern with family decline is through variable rates of church attendance. All religious institutions communicate ex- plicit normative recommendations about family behavior and organization (Stevens 2001; Thornton 1985). Attendance at reli- gious services exposes individuals to reli- gious communication involving clergy (and other attenders), while also establishing so- cial networks in which such ideas are rein- forced (Burns, Schlozman, and Verba 2001, chap. 3; Wald, Owen, and Hill 1988). If level of exposure to any religious communication is the main factor at work, it will be church attendance rather than denominational mem-

Sherkat and Ellison (1997) find that religious teachings associated with specific denominations (such as biblical inerrancy among conservative Protestants or evangelicals) can provide a source of cognitive structure which subsequently affects attitudes on specific issues.

bership that accounts for concern with fam- ily decline.

A third way in which religion may lead to a concern with family decline is through exposure to denomination-specific influences. Thus the content of religious communication may vary across denominational families, leading to exposure to different types of re- ligious influence. Qualitative differences be- tween denominations in the content of com- munication are central to U.S. religious in- stitutions (Regnerus, Sikkink, and Smith 1999; Woodberry and Smith 1998). Guided by recent findings regarding the social cohe- siveness and moral conservatism of evan- gelical Protestants (Wald, Owen, and Hill 1990; Wilcox forthcoming), I evaluate the hypothesis that regular participation in evan- gelical churches leads to higher levels of concern with family decline. This scenario would be empirically manifested in a signifi- cant interaction of church attendance x evangelical Protestant, suggesting that evangeli- cal churches facilitate higher levels of con- cern with family decline.

The theory of religious influence is consis- tent with findings that any of the preceding three causal mechanisms explain public con- cern with family decline. However, a finding that none of these mechanisms affects levels of public concern with family decline would make religious influence theory irrelevant. The results of the analyses provide evidence for the causal relevance of specific religious factors, thus advancing our understanding of religious influence with respect to issues and conflicts relating to family change.


A general expectation of the theoretical work on religious influence is that by shaping po- litically relevant attitudes and identities, re- ligion indirectly influences political behav- ior (Regnerus et al. 1999; Wald 1996; Woodberry and Smith 1998). Does public concern with family decline affect U.S. vot- ing behavior? Such a relationship could af- fect patterns of political party conflict and national policymaking. If this relationship persists over time, and if level of concern with family decline is shaped by (religious) group identities stemming from different in- terests or ideological orientations (Franklin, Mackie, and Valen 1992; Lipset and Rokkan 1967; Manza and Brooks 1999), this pro- vides evidence of a cleavage that separates voters into distinct partisan camps. As sug- gested by the case of race (Carmines and Stimson 1989), cleavages also can routinize social conflicts by identifying clear stakes for individuals and groups, and by provid- ing regular opportunities for the pursuit of group interests.

Has such a cleavage emerged in the con- temporary United States? Several studies suggest that voters perceive differences be- tween the Democratic and Republican par- ties on issues relating to family change (Freeman 1993; Hammond, Shibley, and Solow 1994; Stacey 1994). While no study has analyzed the effects of attitudes toward family decline on voting behavior, these studies provide grounds for considering the possibility of such effects over the relatively long time period from 1972 to 1996. A pattern of increasing effects would provide evi- dence of a cleavage that adds to our under- standing of new sources of voter alignments and party conflict in American politics. Fur- ther details are discussed below.


Guided by the theory of religious influence, my analyses address three questions: (1) Have levels of public concern with family decline changed over time? (2) What factors, relating especially to religion, lead to high levels of concern with family decline? (3) Do these attitudes affect voting behavior, leading to a new political cleavage? By an- swering these questions, I develop a compre- hensive analysis of the nature, sources, and political consequences of public concern with family decline in the United States.

I analyze data from the National Election Studies' (NES) surveys of voting behavior in

U.S. national elections (Center for Political Studies 1972-1996). The NES surveys rep- resent very high-quality data on public atti- tudes, sociodemographic attributes, and vot- ing behavior. Items measuring attitudes to- ward family decline over time are very scarce, and the NES surveys represent the

only data of which I am aware that permit measurement of both voting behavior and public concern with family decline.

The analyses span the years from 1972 through 1996, encompassing the period dur- ing which concern with family decline has been hypothesized to increase. Because spe- cific control variables are available in select surveys, the analyses are presented in three stages. Each stage corresponds to one of the three research questions and involves a data set extracted from the NES surveys. In all three data sets, year itself is a variable, en- abling hypotheses about time trends to be directly tested.



In six of the seven presidential election years between 1972 and 1996, the NES asked re- spondents the following open-ended ques- tion: "What do you think are the most im- portant problems facing this country? Of those you've mentioned, what would you say is the single most important problem the country faces?" The 1980 election survey used a modified version of this q~estion.~

I use the conceptualization discussed in the preceding section to identify responses that indicate a concern with family decline, assigning a score of "1" to respondents who mention one of the following problems: (1) family decline; (2) divorce rates or the in- stability of families; (3) single-parenthood or nonmarital births; (4) the inadequate so- cialization of children; and (5) references to child poverty that link it to one of the four preceding processes. All other respondents who were asked the NES question are as- signed a score of "0"(see Table 1).

Because of its explicit reference to the "most important problem," the NES item en- ables measurement of the theoretical concept of public concern with family de~line.~


The question in 1980 was: "What do you per- sonally feel are the most important problems the government in Washington should try to take care of?" This difference in question wording has no significant impact on the analyses.

Past research on social problems (Best 1995; Gusfield 1989), including experimental survey research (Iyengar and Kinder 1987; see Kinder 1998 for review), provides evidence for the as-

requirement that individuals rank family de- cline as the most important problem discour- ages casual mentions of family decline, re- ducing the possibility of Type I errors. Type I1 errors (i.e., underestimating the true ex- tent of concern with family decline) cannot be ruled out. However, given this study's fo- cus on over-time processes, the latter sce- nario is of limited consequence as there is no reason to assume that such errors are cor- related with time.

The first stage of the analyses (Table 1) ana- lyzes the over-time pattern and sources of public concern with family de~line,~

focusing on the three religion factors associated with my theoretical approach to understand- ing religious influence. Building from recent work on denominations (Steensland et al. 2000), I measure religious group member- ship using dummy variables for evangelical Protestants, mainline Protestants, black Prot- estants, Catholics, Jews, those with no reli- gion, and respondents in other religions. (A table listing the denominations included in each of these categories is available from the author; also see Steensland et al. 2000, a~p.)~

The second religion variable mea- sures frequency of church attendance, and it has six ordered categories ranging from don't attend (1) to attend every week (6). The third type of religious factor is mea- sured using a series of religion x attendance interactions (additional details are discussed in the results section).

sumption that mentions of social problems are valid measures of the level of salience/public concern with a specific issue.

Because the economic control variables and the family decline item were not fielded in the same ballot of the 1972 survey, the first stage of the analyses spans the 1976 to 1996 surveys.

While all measures of religious group mem- berships have some limitations because of ran- dom measurement error and the complexity of denominational families in the United States (T. Smith 1990; Woodberry and Smith 1998), the current scheme identifies the main boundaries between religious traditions, providing the most systematic approach to measuring U.S. religious denominations. A different 7-category typology derived from work by T. Smith (1990) yields similar findings.

Table 1. Coding and Sample Means for Variables'used in the Three Stages of the Analyses: National Election Study, 1972 to 1996

Stage of Analysis

Variable Coding (1) 1976-1996 (2) 1972-1996 (3) 1992-1996
Concern with family decline a Coded 1 [Reference = else] .04 .04 .08
Voted Democratic Coded 1 [Reference= Republic an] - .47 .58
Religion [Reference = Catholic]      
Evangelical Protestant     .20 .I8 .24
Mainline Protestant     .34 .37 .27
Black Protestant     .07 .06 .07
Jewish     .02 .02 .02
Other religion     .03 .03 .03
No religion     .10 .07 .ll
Church attendance Scale score: 1 = don't attend   3.76 4.01 3.76
  to 6 = attend every week      
Education Number of years 12.68 12.94 13.74
Household income Coded in 1992 U.S. dollars 36,422 39,543 40,210
Age Years 44.66 46.51 47.77
Residence in the South Coded 1 .24 .23 .29
African American Coded 1 .10 .10 .ll
Female Coded 1 .55 .55 .54

Class [Reference = not in the labor force] Professional .14 .16 .21 Manager .08 .09 .09

Routine white-collar .13 .13 .12

Self-employed .07 .07 .07

Skilled worker .07 .07 .06

Unskilledlsemiskilled worker .14 .12 .I0

Retrospective Economic Evaluations [Reference = better than year ago] Same as a year ago .3 1 -.35

Worse than a year ago .30 -.30

Prospective Economic Evaluations [Reference = better off than now] Same as now .55 -.58

Worse than now .12 -.09

Abortion Attitudes [Reference = abortion should Permit abortion only in case of never be permitted] --.27 rape, incest or when the woman's life is in danger

Permit abortion for reasons other --.15 than rape, incest or danger to the woman's life, but only after the need for the abortion has been clearly established

A woman should always be able --.48 to obtain an abortion as a matter of personal choice

(Continued on next page)

(Table I continued)

Stage of Analysis

Variable Coding
Gender attitudes A 7-item Likert scale coded - -
  from 0 to 6 with higher scores    
  designating more egalitarian    
Welfare state attitudes A 7-item Likert scale coded - - 2.52
  from 0 to 6 with higher scores      
  designating more egalitarian      
Number of cases   7,440 6,022 1,316

a Question wording for the dependent variable was, "What do you think are the most important problems facing this country? . . . Of those you've mentioned, what would you say is the single most important?" Question wording differed for the 1980 and 1982 election years: Alternative wording for these elections was, "What do you personally feel are the most important problems the government in Washington should

try to take care of?"

Given the importance of economic factors to perceptions of social problems (Kiewiet 1983), I include in the first (and also third) stage of the analyses controls for economic performance evaluations. Retrospective evaluations are measured using two dummy variables for responses comparing respon- dents' current economic and past economic situations; prospective evaluations are mea- sured using two dummy variables for re- sponses comparing current and future eco- nomic situations. All stages of the analyses include the following sociodemographic con- trol variables: education (in years); house- hold income (measured in 1992 dollars); age (in years); region (coded 1 for South); race (coded 1 for African American respondents); gender (coded 1for female respondents); and class (dummy variables for professionals, managers, routine white-collar employees, self-employed nonprofessionals, skilled workers, unskilled/semiskilled workers, and non-full time labor force participants).



In the second and third stages of the analy- sis (see Table I), the dependent variable is vote choice in presidential elections (coded 0 if voted for the Republican candidate and 1 if voted for the Democratic candidate).'

There is no clear theoretical basis for antici-

In the second stage I analyze the magnitude and over-time pattern of change in the po- litical-behavioral effects of concern with family decline over the entire 1972 through 1996 period.

In the third stage of the analysis, I take advantage of a larger array of items avail- able in the 1992 and 1996 NES surveys to obtain corroborating evidence that the politi- cal effects of concern with family decline are independent of other, better-known sources of the presidential vote (e.g., Abramson, Aldrich, and Rohde 1999).8 These items in-

pating that concern with family decline has a sig- nificant relationship to third-party candidacies or to voter turnout in presidential elections, and ad- ditional analysis provides no evidence of such relationships. Using a logistic regression model of third-party support that includes terms only for the main effects of election, the improvement in -2 log-likelihood obtained by estimating the main effect of a concern with family decline is a non-significant 2.47 (1 d.f. ); the corresponding change in -2 log-likelihood when using voter turnout as the dependent variable is 3.27 (1 d.f.), also not significant at the p < .05 level.

The inclusion of such measures (and also in- dependent variables related to vote choice and public opinion in the second stage of the analy- sis) are useful in controlling for significant sources of political change such as evaluations of the performance of the economy and ongoing trends in the effects of social cleavages (Abramson, Aldrich, and Rohde 1999; Dalton and Wattenberg 1993).

Table 2. Unstandardized Coefficients from the Logistic Regression of the Sources of Public Concern with Family Decline on Religion and Other Independent Variables: National Election

Study, 1976 to 1996

Model 1

Variable Coef. (S.E.)

Constant -6.42* (.58)

Election Year [Reference = 19761

1980 -l.O8* (.50)
1984 .38 (.29)
1988 1.11* (.28)
1992 1.40' (.25)
1996 2.09* (.26)
Religious Group  
Evangelical 1.38* (.19)
Mainline Protestant .20 (.22)
Black Protestant .44 (.53)
Jewish -3.77*(4.64)

Other religion
No religion

Church attendance Education Household income

(X 10,000) Age Residence in the South African American Female

l.Ol* (.33)

1.12* (.38)

.45* (.05) -.Ol (.03) .03 (<.Ol)

-.01*(< .01) -.30 (.16) -.52 (.38) -.I9 (.15)

Model / Independent
Coef. (S.E.) Variable


Professional Manager Routine white-

collar Self-employed Skilled worker Unskilledlsemi

skilled worker

Model 1 Coef. (S.E.) Model 2 Coef. (S.E.)
.16 .21 -.Ol (.22) (.26) (.23)  
-.28 .09 -.I9 (.31) (.29) (.24)  

Retrospective Economic Evaluations

Same as a year ago .34* (.16)

Worse than a year .Ol (.18) ago

Prospective Economic Evaluations

Same as now

Worse than now

Evangelical Protestant x Church attendance

-2 log-likelihood Degrees of freedom BIC

-.05 (.15) .21 (.24)


1,883.64 7,411 -64,183

Note: Numbers in parentheses are standard errors; N = 7,440.

*p < .05 (two-tailed tests)


What causal factors lead to high levels of concern with family decline? I address this question using the analyses presented in Table 2. The entries in this table are coeffi- cients and standard errors for two logistic regression models in which concern with

the 1980 and 1982 surveys. Although the esti- mates of family decline concern for 1980 and 1982 are significantly lower than those in the ad- jacent years, they are consistent with the implied picture of a declining trend through the late 1970s followed by an increasing trend starting in the mid-1980s.

family decline is predicted as a function of religion factors and other independent vari- ables.

The effects of education, household in- come, region, class, gender, and race are not significant. Only one of the four coefficients for the effects of the economic controls is statistically significant. The effect of age is significant, but the impact is small-a 20year difference in age changes the log-odds of a concern with family decline by only .2. While the sign of the coefficient for the ef- fect of race is in keeping with past research documenting enduring differences in policy preferences among African Americans versus others (Tate 1993), the coefficient is not significant.

The key to understanding the social sources of these attitudes lies with the causal factors associated with the theory of religious influence. With regard to religious group memberships, evangelical Protestants have the highest level of concern with family de- cline. With regard to religious participation, the .45 coefficient for church attendance sug- gests that attendance is an important addi- tional source of concern with family de- cline.1° Comparing across the full range of religious participation, moving from nonat- tendance to weekly attendance raises the odds of viewing family decline as the most important problem by a factor of 9.5 (corresponding to an approximately .28 increase in probability from an initial baseline of .05).

Model 2 explores these results in greater detail. Model 2 has a single, additional coef- ficient for the interaction of church atten- dance and membership in evangelical Prot- estant denominations, and it improves over Model 1's fit according to both the -2 log-likelihood statistic and BIC index. This in- dicates that the impact of church attendance on concern with family decline is signifi- cantly different for evangelical Protestants compared with all other religious groups.

Model 2's coefficients reveal two impor- tant findings. First, the difference between the coefficients for evangelical Protestants and those of other religious groups shrinks considerably, suggesting that it is the differ- ential and larger effects of church atten- dance among evangelicals that explain their higher levels of concern with family decline. Second, although participation in any religious institution has a significant effect in both models, this effect is smaller in Model 2 than in Model 1.

Because the pattern of interaction between religion and church attendance is central to understanding the sources of concern with

lo As a check on interval-level measurement of the church attendance covariate, I evaluated two alternative specifications to Model 1. The first of these, a logarithmetic transformation of attendance, consumes the same degrees of free- dom as Model 1, but provides a far worse fit to data (-2 log-likelihood = 1899.15; BIC = -64,167). The second specification uses dummy variables to measure attendance as a nominal covariate, and it also yields a worse fit to data (-2 log-likelihood = 1877.20; BIC = -64,153).

family decline, I also evaluate the fit of sev- eral supplementary models in Table 3 to test additional hypotheses. These analyses find no evidence for any additional interaction effects involving Protestant or Catholic de- nominational memberships and church at- tendance." The additional analyses also provide no evidence of change in the effects of attendance over time (the fit of Model 1 is superior to the fits of Models 7 through 10). The superiority of the preferred model from Table 2 (Model 1 in Table 3) implies that interaction effects involving religion and church attendance are limited to the two- way interaction of church attendance x evangelical Protestants.


Figure 2 summarizes the effects of religion and church attendance on concern with fam- ily decline. Given the interaction between evangelical Protestants and church atten- dance, it is appropriate to simultaneously consider the effects of all religion factors. Figure 2 presents the predicted probability of ranking family decline as the most impor- tant problem by varying religious group membership and church attendance rate, holding all other independent variables at fixed levels. l2

In Figure 2, the generally small curvature of the lines for mainline and black Protes- tants, Catholics, Jewish respondents, and those with no religion13 indicate that church

l1 The small reduction in the -2 log-likelihood statistic produced by allowing the effects of church attendance to differ across all religious groups (see Model 2) is not close to achieving significance using either the chi-square test or BIC. Comparison of Model 1 with Models 3 and 4 provides no evidence for interactions involving race or gender and religion; comparison of Model 1 with Models 5 and 6 provides no evidence for interactions involving race or gender and church attendance.

l2 The fixed values for these variables are: nonblack males living outside the South in 1984 who are not working full-time, who are at the sample means of age, education, and household income, and who endorse the "same" response to the economic evaluation items.

l3 While the fact that church attendance has any impact on respondents in the "no religion"

Table 3. Fit Statistics for Additional Logistic Regression Models Predicting the Sources of Public Concern with Family Decline: National Election Study, 1976 to 1996

-2 Log- Degrees of
Model Likelihood Freedom BIC
(1) Preferred model (Model 2 from Table 2) 1,873.62 7,410 -64,184
(2) Model 1 + church attendance x religion (6 dummy variables) 1,867.77 7,404 -64,136
(3) Model 1 + black x religion (5 dummy variables) 1,866.43 7,405 -64,146
(4) Model 1 + female x religion (6 dummy variables) 1,860.85 7,404 -64,143
(5) Model 1 + black x church attendance 1,873.05 7,409 -64,175
(6) Model 1 + female x church attendance 1,872.93 7,409 -64,176
(7) Model 1 + church attendance x evangelical Protestant 1,873.57 7,409 -64,175
x year (continuousa)      
(8) Model 1 + church attendance x religion x year (continuousa) 1,869.98 7,404 -64,134
(9) Model 1 + church attendance x year (continuousa) 1,869.97 7,409 -64,178
(10) Model 1 + church attendance x year (5 dummy variables) 1,867.37 7,405 -64,145
Note: N = 7,440.      
aYear is coded 1 for 1976,2 for 1980, . . . ,6 for 1996      

attendance has a small impact on level of concern with family decline. By contrast, church attendance has a slightly larger im- pact on family decline concern among re- spondents in other religions.14 The largest effects of church attendance are found, how- ever, among evangelical Protestants, with regular church attenders being over 24 times more likely than nonattenders to consider family decline as the most important prob- lem. The magnitudes of the church atten- dance effect among evangelical Protestants can be seen in the last panel's estimates; these estimates reveal that compared with

category may appear initially to represent a para- dox, it is readily explained by the heterogeneity of individuals within this category (~lenn 1987b), not all of whom are genuine atheists.

l4 Model 2's coefficients (see Table 2) assume that the impact of church attendance on the log- odds of concern with family decline is identical across all religious groups (with the single excep- tion of evangelical Protestants). Note that be- cause the level of these attitudes is higher among respondents in the "other religion" category com- pared with the remaining groups, the predicted effect of church attendance on the probability of family decline concern among other religions is higher. This divergence in logit versus probabil- ity metrics reflects the insensitivity of the logit estimates to the distribution of the dependent variable.

the 24-fold effect for evangelicals, the cor- responding effect of church attendance among all other respondents (i.e., all non- evangelicals) is approximately 5.

Although these results suggest that church attendance (and thus exposure to church- based communication and influence) is the underlying factor explaining evangelical Protestants' comparatively high levels of concern with family decline, an alternative hypothesis is that regular church attendance is instead the result of members' preexisting attitudes. Two considerations cast doubt on this alternative hypothesis. First, research on religious participation finds that church at- tendance is shaped strongly by socialization and life course transitions (Sherkat 1998; Stolzenberg, Blair-Loy, and Waite 1995), factors not likely to be influenced by a con- cern with family decline. Second, if family decline concern influenced church atten- dance, the over-time increase in concern would by itself have caused a subsequent rise in church attendance rates. Although contemporary research on church attendance is marked by a significant debate over whether U.S. church attendance remains high and generally stable (Hout and Greeley 1987; 1998) or instead involves over-reporting and possibly also over-time de- cline (Hadaway, Marler, and Chaves 1993,

Evangelical Protestant Mainline Protestant Black Protestant





Other Religion All Non-Evangelical

..................... .....................


Low High Low High Low High Low High

Church Attendance Level Church Attendance Level Church Attendance Level Church Attendance Level

Figure 2. Predicted Probability of Concern with Family Decline among Religious Groups, by Level of Church Attendance: National Election Study, 1976 to 1996

Note: Probabilities are calculated using fixed levels of covariates (varying only religion and church attendance)

Table 4. Fit Statistics and Selected Unstandardized Coefficients from Logistic Regression Models Predicting the Effects of Concern with Family Decline on Presidential Vote Choice: National Election Study, 1972 to 1996

Fit Statistics

Models -2 Log- Degrees of Likelihood Freedom BIC Coefficients (S.E.) Pconcern Pconcern x year
(1) Election years only 8,177.50 6,015 44,172     -
(2) Model 1 + concern with family decline 8,062.83 6,014 -44,278 -1.66* (.18) -
(3) Model 2 + concern with family decline x year (continuousa) 8,048.58 6,013 44,284 -.21 (.38) -.30* (.08)
(4) Model 2 + concern with family decline x year(dummy variables) 8,045.34 6,008 -44,243 - -
(5) Model 3 + religion, church attendance, and all other covariatesb 6,952.14 5,993 45,206 -.01 (.42) -.31* (.08)

Note: Numbers in parentheses are standard errors; N = 6,022. a Year is coded 1 for 1972, 2 for 1976, . . . ,7 for 1996. Includes religion, church attendance, education, household income, age, region, race, gender, class, and

other religion x church attendance.

1998; Presser and Stinson 1998; see Chaves and Gorski 2001 for review of religious par- ticipation in comparative-historical perspec- tive), there is general agreement that church attendance rates did not rise during the past two decades. It is, in principle, possible that self-selection is operating while a third fac- tor that is uncorrelated with level of concern with family decline simultaneously acts so as to depress church attendance. Although this scenario cannot be ruled out, it is con- siderably more complicated than the infer- ence that church attendance influences con- cern with family decline. The absence of any evidence for the more complicated scenario suggests that the simpler proposition is pref- erable.

Given that public concern with family de- cline is concentrated among evangelical Protestants who attend church regularly, we can reestimate the increase in concerns over family decline from 1980 to 1996 to better observe the importance of evangelical atti- tudes to this trend: Whereas the overall in- crease in level of family decline concern is 9 percent, the corresponding increase without evangelicals' attitudes shrinks to 5 percent. Accordingly, without the dispropor- tionate concentration and growth of concern with family decline among evangelical Prot- estants, the overall increase in such attitudes would have been considerably smaller.


Does concern with family decline affect the behavior of voters, perhaps leading to the emergence of a new political cleavage? I an- swer this question using the results of the second stage of the analysis. In Table 4, I evaluate the fit of a series of models that pre- dict the log-odds of Democratic versus Re- publican Party vote choice in the seven presi- dential elections between 1972 and 1996.

The comparison between the first two models easily favors Model 2, providing evi- dence that concern with family decline does affect presidential vote choice. Model 3 is preferred over Model 2, establishing that the political effects of concern with family de- cline have changed over time, following a linear pattern. Allowing the effects of fam- ily decline concern to differ in each election (Model 4) does not improve over Model 3, and the small difference in -2 log-likelihood between Models 3 and 4 (3.24; 5 d.f.) im- plies that all the interaction between family decline concern and year is captured by Model 3. In the final model (Model 5), I add a series of controls to estimate the direct ef- fects of concern with family decline on vote choice.

The coefficients from Models 2, 3, and 5 provide more detailed information about the effect of public concern with family decline on vote choice. Given that Model 3 im- proves over Model 2's fit, Model 3's coeffi- cients reveal the ways in which Model 2 misspecifies the political effects of concern with family decline: Instead of a large, stable effect (as in Model 2), Model 3's coeffi- cients suggest that the effects of concern with family decline increased steadily over time. More specifically, the predicted effect of concern with family decline is modest in the 1970s, but by 1984 the effect is substan- tial (-1.4 I), and by 1996, the corresponding effect is predictably larger (-2.31). All else being equal, concern with family decline in- creasingly disposes voters to favor Republi- can over Democratic presidential candidates. The over-time growth in the effect on vote choice of a concern with family decline also suggests that high levels of such concerns prior to the 1970s were of limited political consequence (given the much smaller coef- ficient for family decline concern early in the 1972 to 1996 series).

Comparing the coefficients for Models 3 versus Model 5 reveals whether the effect of concern with family decline on vote choice is independent of the corresponding politi- cal effects of the sociodemographic co- variates. The coefficient for the family de- cline concern x year effect is virtually un- changed across Models 3 versus 5, suggest- ing the independence of the political effects of family decline concem.15

The magnitude of the effects of concern with family decline on vote choice over a relatively lengthy period of time suggests the emergence of a growing cleavage in presidential elections during the late 1980s and early 1990s. Using Model 5's coeffi- cients (and the relevant sample means), the predicted difference in vote choice for re- spondents scoring 0 versus 1 on the family decline item is .30 in 1988, .41 in 1992, and .48 in 1996. These predicted differences are

l5 I also compared Model 5 with a competing model that allows the religion and family decline concern variables to interact with one another. The resulting lack of improvement in fit (7.25; 6

d.f.1 implies that the effect of concern with fam- ily decline on vote choice is the same for the seven religious groups in the analysis (see Ap- pendix A for all coefficients and standard errors for Model 5 in Table 4).

large, especially given that they reflect the influence of a single variable. For instance, the preceding estimates indicate that view- ing family decline as the most important problem in the 1996 election is predicted as substantially lowering (by .48) the probabil- ity of a typical voter supporting the Demo- cratic candidate (Bill Clinton).

To provide additional evidence that the ef- fect of family decline concern on vote choice is independent of other, better-known sources of the presidential vote, Table 5 presents the third stage of the analysis, which takes ad- vantage of more extensive data available in the 1992 and 1996 NES surveys. Model 1 of Table 5 estimates the total effect of concern with family decline on presidential vote choice, and Model 2 estimates the direct ef- fect by controlling for attitudes toward abor- tion, gender, the welfare state, economic evaluations (and sociodemographic factors). The key comparison in Table 5 is between the coefficient for concern with family de- cline in Model 1 versus Model 2. The total effect of a concern with family decline is larger than the direct effect, but the coeffi- cient in Model 2 retains over 73 percent of its original magnitude. This result provides evidence that the association between con- cern with family decline and presidential vote choice represents a new source of voter alignments in the United States (rather than the political effect of a preexisting cleavage). It should be emphasized that because the actual percentage of voters exhibiting high levels of concern with family decline has been relatively modest (10 percent at its highest level in 1996), the effects of this cleavage on the outcome of elections have also been modest. For instance, the 6-percent increase in concern with family decline from 1988 to 1996 is predicted as lowering the Democratic share of the major party vote in 1996 by approximately 3 percent, resulting in a slightly lower margin of victory for Democratic incumbent Bill Clinton without necessarily changing the outcome of that election. For the family decline cleavage to have a larger effect on the outcome of elec- tions, either the level of family decline con- cern or its association with vote choice would have to increase. In the absence of these de- velopments, the family decline cleavage is most notable as a new dimension of partisan

Table 5. Unstandardized Coefficients from the Logistic Regression of the Effects of Concern with Family Decline on Presidential Vote Choice and Other Independent Variables: National Election Study, 1992 to 1996

Model 1 Model 1 Model 2

Independent Model Independent
Variable Coef.(S.E.) Coef.(S.E.) I Variable Coef. (S.E.) Coef.(S.E.)

Constant .44* (.07) -.34 (.73) 1 Retrospective Economic Evaluations

Same as a year ago -.62* (.20)

Election Year [Reference = 19921 1996 .14* (.12) 1.46' (.29) Worse than a year ago -.99* (.23)

Concern with family -2.00' (.26) -1.47* (.33) decline

Prospective Economic Evaluations

Religious Group [Reference = Catholic] Same as now --.21 (.19) Evangelical Protestant --.97* (.21)

Worse than now --.I6 (.32) Mainline Protestant --.72* (.20)

Black Protestant --.04 (.87)

Retrospective economic --1.27* (.20) Jewish -1.77* (.78) evaluations x year1996 Other religion --.45 (.39)

prospective economic -.05 (.27) No religion --.06 (.30) evaluations x year1996

Church attendance --.I 1* (.05)

Abortion Attitudes

Education --. 12' (.04) Only in case of rape, -.21 (.28)

Household income --1.10*(<.01) (X 10,000) incest, or when woman's life is in danger -.01*(<.01)

Age Only after need is -.38 (.31)

Residence in the South -.10 (.17)

established African American -2.75* (.64)

Always a matter of -1.26* (.30)

Female -.07 (.IS)

personal choice


Professionals -.47* (.24)

Gender attitudes -.16* (.05) Managers -'I9 ('29)

Welfare state attitudes -
Routine white-collar -.26 (.27)

Self-employed --.15 (.31) -2 log-likelihood 1,707.07 1,213.51

Skilled workers --. I5 (.34) Degrees of freedom 1,313 1,283

Unskilledlsemiskilled -.39 (.28) worker BIC -7,723 -8,001

Note: Numbers in parentheses are standard errors; N = 1,3 16
*p< .05 (two-tailed tests)

political conflict since the 1970s, providing nomic consequences of divorce. As a result, evidence for the significance of the family neither proponents nor critics of the family decline issue for political parties and elec- decline thesis have systematically investi- toral competition. gated attitudes toward family change. The absence of analysis makes it impossible to evaluate important but untested hypotheses


about rising public concern with family de- Scholarly debates over the family decline cline (Popenoe 1993; Scammon and Watten- thesis have focused primarily on demo-berg [I9701 1992; Wilson 1993). Moreover, graphic or economic processes such as such investigations can yield an informative trends in nonmarital birth rates or the eco- portrait of the meaning of family processes, especially because attitudes, beliefs, and preferences do not necessarily correspond to the precise contours of social structural change. By analyzing the phenomenon of family decline in terms of the level of public concern, I have sought to extend research on family change to address new and fruitful questions of relevance to political sociology and the sociology of religion as well.

My approach has been guided by a theory of religious influence that emerges from re- search on the substantial effects of religion factors on political participation and family behaviors (Sherkat and Ellison 1999; Verba et al. 1995, chap. 15). This study's results reveal the utility of this theoretical approach, providing new evidence for the relevance of specific religious processes to understanding conflicts, attitudes toward family change, and voters' political alignment based on these attitudes. All three factors considered here affect individuals' level of concern with fam- ily decline, showing the importance of con- ceptualizing and testing hypotheses about what specific religion factor is behind a par- ticular phenomenon. In doing so, these re- sults suggest the fruitfulness of synthesizing institutional and individual-level factors to better understand the influence of religion within other empirical contexts (Chaves and Gorski 2001; Wuthnow and Lawson 1994).


Regarding controversies over subjective as- pects of family change, my analyses provide evidence of a growing trend between 1980 and 1996 in public concern with family de- cline as the most important social problem in the United States. Whereas in 1980 less than 1 percent of Americans considered fam- ily decline as the most important problem, this figure had grown to 10 percent by 1996. This trend is consistent with the hypothesis discussed above, and it can be further appre- ciated when juxtaposed with results from my analyses of its consequences for political be- havior. Indeed, in the absence of such evi- dence, skeptical commentators might rea- sonably call into question the sociological significance of a phenomenon that has no ef- fect on the actual behavior of individuals. My analyses address this question by provid- ing evidence that voters' level of concern with family decline has had a growing im- pact on their voting behavior in presidential elections, with concern with family decline by itself disposing voters to support Repub- lican candidates.

The growing association between level of family decline concern and presidential vote choice suggests the emergence of a new cleavage in U.S. politics. We can appreciate the justification for this inference by consid- ering the implications of results that instead showed a pattern of short-term political-be- havioral effects (or a pattern in which the ef- fects of family decline concern varied widely in magnitude over time). Results of this sort would suggest the emergence of a policy is- sue whose partisan relevance was specific to one or more particular campaigns or elec- tions, but not the emergence of a new politi- cal cleavage. However, because it is charac- terized by a growing trend in partisan impact, and because it is shaped by the divergent ex- periences of groups with long-standing dif- ferences in identity and ideological orienta- tion, the issue of family decline appears to qualify as a novel cleavage.




The results of this study advance political- sociological research on electoral politics in two additional ways. First, evidence for the emergence of a new and relatively enduring source of U.S. voter alignments extends con- temporary research and scholarly debates over partisan change in developed democra- cies (Dalton and Wattenberg 1993). Indeed, voters' alignments with political parties since the 1960s, once assumed to be "frozen" in historical time (Lipset and Rokkan 1967), have increasingly been found to evolve in novel ways that become explicable when linked to ongoing social conflicts and opin- ion trends (Brooks 2000; Carmines and Lay- man 1997).

A second connection to political-socio- logical research concerns the revitalization of network-based theories of U.S. politics. Recent advances have provided an analyti- cal basis for understanding how social struc- tural factors influence individuals' political attitudes and behavior by transmitting infor- mation and influence through social network ties (Huckfeldt, Plutzer, Sprague 1993; Huckfeldt and Sprague 1991). My findings bear on social network theorizing, suggest- ing not only that religious group member- ships influence politically-relevant attitudes, but also that their influence operates simul- taneously through congregation-level pro- cesses, attesting to the importance of reli- gious institutions and the communicative ca- pacities of religious leaders.

My research thus corroborates the general expectation that the political and communi- cative influence of social structure is often mediated by the quantity of networks link- ing individuals (Zuckerman, Valentino, and Zuckerman 1994). Although recent network research on U.S. politics has found evidence for the effects of weak ties that bridge dif

ferent social groups (Huckfeldt and Sprague 1991), my findings regarding the effects of denomination and church attendance attest to the ongoing importance of social ties that reinforce cohesive group boundaries. Reli- gious institutions provide a particularly in- structive example of such factors because churches facilitate types of communication that activate members' preexisting beliefs and also the authority of clergy (Wald et al. 1988, 1990). The network position of reli- gious leaders appears particularly signifi- cant, suggesting the disproportionate influ- ence of evangelical clergy and also the ways in which network and institutional flows of communication are not historically invari- ant, but depend instead upon the activities of strategically positioned groups or actors.


My findings also contribute to long-standing debates in the sociology of religion over conservative political trends affecting evan- gelical Protestants. Prompted by journalistic commentary on the apparent influence of Christian Right leaders in the 1980 presiden- tial election, a subsequent body of research casts doubt on these early hypotheses. Not only were the Moral Majority and its policy positions found to be unpopular (Buell and Sigelman 1985; Sigelman and Presser 1988), but the proportion of evangelicals actively involved in conservative interest groups ap- pears to have been small (Wuthnow 1999).

Furthermore, conservative Protestants as a whole have comparatively low levels of po- litical participation, remaining in a Republi- can political alignment that appears to have been established prior to the 1980s (Bruce 1994; Manza and Brooks 1997).

More recently, however, an emerging body of research has begun to reexamine conser- vative trends among evangelicals and pos- sible avenues of Christian Right influence, suggesting some modifications to earlier findings. In addition to the success of a small number of Christian Right candidates in state and local elections in the 1990s (Green 1995; Rozell and Wilcox 1997), a 1996 sur- vey found that 20 percent of respondents re- ported using information provided by Chris- tian Right organizations in deciding whom to support in elections (Regnerus et al. 1999). Moreover, although earlier claims about religion-based polarization in attitudes find little support (Davis and Robinson 1996; DiMaggio, Evans, and Bryson 1996), evangelicals (or conservative Protestants) are more likely than others to support con- servative policy positions on such specific issues as gender roles and sexuality (Sherkat and Ellison 1999; Woodberry and Smith 1998).

The current study's results contribute to the reconsideration of political trends among evangelicals by providing evidence that growing levels of concern with family de- cline between 1980 and 1996 were dispro- portionately concentrated among evangelical Protestants. I find evidence that high levels of concern with family decline are facilitated by exposure to church-based religious com- munication. The effect of regular church at- tendance is larger for members of evangeli- cal Protestant churches, attesting to differ- ences in the effects of religious communica- tion across denominations, especially be- tween evangelical versus nonevangelical churches. These results imply that claims about family decline found their most recep- tive audience among active members of evangelical Protestant denominations.16

l6 The possibility that concern with family de- cline influences church attendance is less likely than the reverse, given that it implies that growth in family decline concern would by itself have raised church attendance rates-and research has

My results bear further on questions con- cerning the effects of religious movements. During the 1990s, the Christian Coalition offered a problem-oriented characterization of family decline, using voter guides that were widely distributed in churches and through public addresses often conducted by (then) executive director, Ralph Reed (Reed 1994, 1996). When viewed through the lens of the theory of religious influence applied here, the preceding results raise fruitful questions about the influence of Christian Right organizations on clergy and members of evangelical Protestant churches. In gen- eral, my findings are consistent with this scenario and also with recent research on the political influence of the Christian Right on evangelical voters in the 1990s (Regnerus et al. 1999). Although it seems likely that well- positioned evangelical leaders communi- cated messages resulting in higher levels of concern with family decline among their congregations, the possibility that communi- cative processes among church attenders (or some combination of these two factors) were also important cannot be conclusively ruled out. Using the results of the current study as a baseline, further research may be able to shed additional light on this line of interpre- tation.

The causal importance of religion factors also suggests some limits on the growth of public concern with family decline. First, the changing social and economic context of the past decade-stabilization in divorce rates and nonmarital births coupled with rising household income and low unemployment- may direct some religious leaders' attention to other issues while lowering the receptiv- ity of individuals to family decline claims that emphasize the negative economic con- sequences of divorce or single-parenthood. A second factor relates to denominational differences and the concentration of con- cerns with family decline among evangeli-

found no evidence of such trends (Hadaway et al. 1993, 1998; Presser and Stinson 1998; Woodberry 1998)-or the operation of a consid- erably more complicated causal scenario for which there is no evidence. Evidence that church attendance is shaped by socialization and life- course transitions (Sherkat 1998)is also relevant, given that it is improbable that concern with fam- ily decline could influence the latter factors.

cal Protestants. Contemporary research on perceptions of religious groups finds that in- dividuals who are not members of evangeli- cal Protestant denominations have generally negative attitudes toward the views they at- tribute to evangelicals (C. Smith 1998). In the absence of social networks that would extend their communicative influence across denominational boundaries, it seems likely that claims associated with evangelical Prot- estant clergy or laity will have a more lim- ited impact on members of other religious traditions. The disproportionate concentra- tion of concern with family decline among the most observant evangelicals reveals not only the most receptive constituency for this phenomenon but also suggests some pos- sible limits on its further growth.

Clem Brooks is Associate Professor of Sociology at Indiana University, Bloomington. His research interests are elections and voting behavior, po- litical culture theory, social change, religion, and quantitative methods. Current projects include a study of the dynamics of partisanship change, social cleavage voting in comparative perspective, and a manuscript investigating the changing role of ideology in U.S. national elec- tions. He is working with Jeff Manza on a review of theory and research in the study of voting be- havior, and is beginning work on a comparative study of racism and policy attitudes in the United States and Western Europe.

Unstandarized Logistic Regression Coefficients from the Preferred Model Predicting the Effect of Concern with Family Decline on Presidential Vote Choice: National Election Study, 1972 to

Independent Variable Coef. (S.E.)
Constant Concern with family decline Concern with family decline x Year (continuous) .95* .O1 -.3 I' ~25) (.42) (.08)
Election Year [Reference = 19 1976 1980 1984 1988 1992 721 .78* .35* .32* .68* 1.08* (.I 1) (.12) (.I 1) (.I31
1996 1.39' (.I41

(Continued on next page)

(Appendix A continued)

Independent Variable Coef. (S.E.)

Religious Group [Reference = Catholic]

Evangelical Protestant -.80* (.09)
Mainline Protestant -.70* (.07)
Black Protestant -.47 (.33)
Jewish 1.14' (.21)
Other religion .96* (.49)
No religion -.07 (.I41
Church attendance (scale score) -. 14* (.02)
Education (in years) -.04* (.01)
Household income (in 1992 -1.30' (<.01)
dollars) x 10,000    
Age (in years) <.01 (<.01)
Residence in the South .07 (.08)
African American 2.77' (.26)
Female .24* (.07)
Class [Reference = Non-Labor Force]  
Professional .38* (.lo)
Manager -.26* (.I21
Routine white-collar .09 (.lo)
Self-employed -.30* (.I31
Skilled worker .18 (.I31
Unskilled/Semiskilled worker .20* (.lo)
Other religion x church -.32* (.11)


-2 log-likelihood 6,952.14

Degrees of freedom 5,993

BIC -45,206

Note:Estimates are for Model 5 from Table 4. Num- bers in parentheses are standard errors; N = 6,022.

*p < .05 (two-tailed tests)


Abramson, Paul, John Aldrich, and David Rohde. 1999. Change and Continuity in the 1996 and 1998 Elections. Washington, DC: Congres- sional Quarterly Press.

Bane, Mary Jo and Paul A. Jargowsky. 1988. "The Links between Government Policy and Family Structure: What Matters and What Doesn't." Pp. 219-61 in The Changing Ameri- can Family and Public Policy, edited by A. Cherlin. Washington, DC: Urban Institute Press.

Best, Joel. 1995. "Constructionism in Context." Pp. 337-54 in Images of Issues: Typifying Contemporary Social Problems, 2d ed, edited by J. Best. New York: Aldine de Gruyter.

Blankenhorn, David, Jr. 1995. Fatherless America: Confronting Our Most Urgent Social

I Problem. New York: Basic Books.

Brooks, Clem. 2000. "Civil Rights Liberalism and the Suppression of a Republican Political Realignment in the United States, 1972 to 1996." American Sociological Review 65:483


Bruce, Steve. 1994. "The Inevitable Failure of the New Christian Right." Sociology of Reli- gion 55:229-42.

Burns, Nancy, Kay Schlozman, and Sidney Verba. 2001. The Private Roots of Public Ac- tion: Gender, Equality, and Political Partici- pation. Cambridge, MA: Harvard University Press.

Buell, Emmett H., Jr. and Lee Sigelman. 1985. "An Army That Meets Every Sunday? Popular Support for the Moral Majority in 1980." Social Science Quarterly 66:426-34.

Carmines, Edward and Geoffrey Layman. 1997. "Value Priorities, Partisanship, and Electoral Choice: The Neglected Case of the United States." Political Behavior 19:283-316.

Carmines, Edward and James Stimson. 1989. Issue Evolution: Race and the Transformation of American Politics. Princeton, NJ: Princeton University Press.

Center for Political Studies. 1972-1996. American National Election Studies. (Pre and Post Election Surveys 1972, 1976, 1980, 1984, 1988, 1992, and 1996; MRDFs and code- books). Ann Arbor, MI: National Election Studies [producer], Center for Political Stud- ies [distributor].

Chaves, Mark and Philip Gorski. 2001. "Reli- gious Pluralism and Religious Participation." Annual Review of Sociology 27:261-81.

Cherlin, Andrew. 1988. "The Changing Ameri- can Family and Public Policy." Pp. 1-29 in

The Changing American Family and Public Policy, edited by A. Cherlin. Washington, DC: Urban Institute Press.

Christian Coalition. 1995. Contract with the American Family: A Bold Plan by Christian Coalition to Strengthen the Family and Re- store Common-Sense Values. Nashville, TN: Moorings.

Dalton, Russell and Martin Wattenberg. 1993. "The Not So Simple Act of Voting." Pp. 193- 218 in Political Science: The State of the Dis- cipline, edited by A. Finifter. Washington, DC: American Political Science Association.

Davis, Nancy and Robert Robinson. 1996. "Are the Rumors of War Exaggerated? Religious Orthodoxy and Moral Progressivism in America." American Journal of Sociology 102:756-87.

DiMaggio, Paul, John Evans, and Bethany Bryson. 1996. "Have Americans' Social Atti- tudes Become More Polarized?" American Journal of Sociology 102:690-755.

Dobson, James and Gary Bauer. 1990. Children at Risk: The Battle for the Hearts and Minds of Our Kids. Dallas, TX: Word Publishing.

Falwell, Jerry. 1992. The New American Family: The Rebirth of the American Dream. Dallas, TX: Word Publishing.

Franklin, Mark, Thomas Mackie, and Henry Valen. 1992. "Introduction." Pp. 3-32 in Electoral Change, edited by M. Franklin, T. Mackie, and H. Valen. New York: Cambridge University Press.

Freeman, Jo. 1993. "Feminism and Family Val- ues: Women at the 1992 Democratic and Re- publican Conventions. " PS: Political Science and Politics 26:21-8.

Furstenberg, Frank. 1998. "Relative Risks: What Is the Family Doing to Our Children?'Con- temporary Sociology 27:223-5.

Gill, Richard. 1997. Posterity Lost: Progress, Ideology, and the Decline of the American Family. Lanham, MD: Rowman and Little- field. Glenn, Norval. 1987a. "Continuity versus Change, Sanguineness versus Concern: Views of the American Family in the Late 1980s." Journal of Family Issues 8:348-54. . 1987b. "The Trend in 'No Religion' Re- spondents to U.S. National Surveys, Late 1950s to Early 1980s." Public Opinion Quar- terly 5 1 :293-3 14. . 1993. "A Plea for Objective Assessment of the Notion of Family Decline." Journal of Marriage and the Family 55:5424. Greeley, Andrew. 1991. Faithful Attraction: Dis- covering Intimacy, Love, and Fidelity in American Marriage. New York: Tor. Green, John. 1995. "The Christian Right and the 1994 Elections: A View from the States." PS: Political Science and Politics 28:5-8. Gusfield, Joseph R. 1989. "Constructing the Ownership of Social Problems: Fun and Profit in the Welfare State." Social Problems 36: 431-41. Hadaway, C. Kirk, Penny L. Marler, and Mark Chaves. 1993. "What the Polls Don't Show: A Closer Look at U.S. Church Attendance." American Sociological Review 58:741-52. . 1998. "Over-Reporting Church Atten- dance in America: Evidence that Demands the Same Verdict." American Sociological Review 64:122-30. Hammond, Phillip, Mark Shibley, and Peter Solow. 1994. "Religion and Family Values in Presidential Voting." Sociology of Religion 55: 277-90. Hout, Michael and Andrew Greeley. 1987. "The Center Doesn't Hold: Church Attendance in the United States, 1940-1984." American So- ciological Review 52:325-45. . 1998. "What Church Officials' Reports

Don't Show." American Sociological Review


Huckfeldt, Robert, Eric Plutzer, and John Sprague. 1993. "Alternative Contexts of Politi- cal Behavior: Churches, Neighborhoods, and Individuals." Journal of Politics 55:365-81.

Huckfeldt, Robert and John Sprague. 1991. "Dis- cussant Effects on Vote Choice: Intimacy, Structure, and Inderdependence." Journal of Politics 53: 122-58.

Iyengar, Shanto and Donald Kinder. 1987. News That Matters. Chicago, IL: University of Chi- cago Press.

Kiewiet, D. Roderick. 1983. Macroeconomics and Micropolitics: The Electoral Effects of Economic Issues. Chicago, IL: University of Chicago Press.

Kinder, Donald. 1998. "Communication and Opinion." Annual Review of Political Science 1:167-87.

Ladd, Everett. 1999. The Ladd Report. New York: Free Press.

Lichter, Daniel. 1997. "Poverty and Inequality among Children." Annual Review of Sociology 23:121-45.

Lipset, Seymour Martin and Stein Rokkan. 1967. "Cleavage Structures, Party Systems, and Voter Alignments: An Introduction." Pp. 1-64 in Party Systems and Voter Alignments, edited by S. M. Lipset and S. Rokkan. New York: Free Press.

Luker, Kristin. 1996. Dubious Conceptions: The Politics of Teenage Pregnancy. Cambridge, MA: Harvard University Press. Manza, Jeff and Clem Brooks. 1997. "The Reli- gious Factor in U.S. Presidential Elections, 1960-1992." American Journal of Sociology 103:38-81. . 1999. Social Cleavages and Political Change: Voter Alignments and U.S. Party Coalitions. New York: Oxford University Press. Murray, Charles. 1993. "The Coming White Underclass." Wall Street Journal, October 29,

p. A14.

Popenoe, David.1993. "American Family De- cline, 1960-1990: A Review and Appraisal." Journal of Marriage and the Family 55527


. 1994. "The Family Condition of America: Cultural Change and Public Policy." Pp. 81- 112 in Values and Public Policy, edited by H.

J. Aaron, T. E. Mann, and T. Taylor. Washing- ton, DC: The Brookings Institution.

Presser, Stanley and Linda Stinson. 1998. "Data Collection Mode and Social Desirability Bias in Self-Reported Religious Attendance." American Sociological Review 63: 137-45.

Reed, Ralph. 1994. Politically Incorrect. Dallas, TX: Word Publishing.

. 1996. Active Faith: How Christians Are Changing the Soul of American Politics. New York: Free Press.

Regnerus, Mark, David Sikkink, and Christian Smith. 1999. "Voting with the Christian Right: Contextual and Individual Patterns of Electoral Influence." Social Forces 77: 1375401.

Rozell, Mark and Clyde Wilcox. 1997. "Conclu- sion: The Christian Right in Campaign '96." Pp. 255-69 in God at the Grass Roots, 1996: The Christian Right in the American Elections, edited by M. Rozell and C. Wilcox. Lanham, MD: Rowman and Littlefield.

Scammon, Richard M. and Ben J. Wattenberg. [I9701 1992. The Real Majority. New York: Donald I. Fine.

Sherkat, Darren. 1998. "Counterculture or Conti- nuity? Competing Influences on Baby Boomers' Religious Orientations and Partici- pation." Social Forces 76: 1087-1 1 15.

Sherkat, Darren and Christopher Ellison. 1997. "The Cognitive Structure of a Moral Crusade: Conservative Protestantism and Opposition to Pornography." Social Forces 75:957-82. . 1999. "Recent Developments and Cur- rent Controversies in the Sociology of Reli- gion." Annual Review of Sociology 25:363-94.

Sigelman, Lee and Stanley Presser. 1988. "Mea- suring Public Support for the New Christian Right: The Perils of Point Estimation." Public Opinion Quarterly 52:325-37.

Skolnick, Arlene. 1991. Embattled Paradise: The American Family in an Age of Uncertainty. New York: Basic Books.

Smith, Christian. 1998. American Evangelical- ism: Embattled and Thriving. Chicago, IL: University of Chicago Press.

Smith, Tom W. 1990. "Classifying Protestant Denominations." Review of Religious Re- search 31:22545.

Spain, Daphne and Suzanne Bianchi. 1996. Balancing Act: Motherhood, Marriage, and Em- ployment among American Women. New York: Russell Sage Foundation.

Stacey, Judith. 1994. "Scents, Scholars, and Stigma: The Revisionist Campaign for Family Values." Social Text 40:5 1-75.

Steensland, Brian, Jerry Park, Mark Regnerus, Lynn Robinson, Bradford Wilcox, and Robert Woodberry. 2000. "The Measure of American Religion: Toward Improving the State of the Art." Social Forces 79:291-318.

Stevens, Mitchell. 2001. Kingdom of Children: Culture and Controversy in the Home-Schooling Movement. Princeton, NJ: Princeton University Press. Stolzenberg, Ross, Mary Blair-Loy, and Linda Waite. 1995. "Religious Participation in Early Adulthood: Age and Family Life Cycle Effects on Church Membership." American Sociologi-

cal Review 60:84-103. Strober, Myra and Sanford Dornbusch. 1988. "Public Policy Alternatives." Pp. 327-55 in

Feminism, Children, and the New Families, edited by M. Strober and S. Dornbusch. New York: Guilford.

Tate, Katherine. 1993. From Protest to Politics. Cambridge, MA: Harvard University Press.

Thornton, Arland. 1985. "Reciprocal Influences of Family and Religion in a Changing World." Journal of Marriage and the Family 47:381-94.

Verba, Sidney, Kay Schlozman, and Henry Brady. 1995. Voice and Equality: Civic Volun- tarism in American Politics. Cambridge, MA: Harvard University Press.

Wald, Kenneth. 1996. Religion and Politics in the United States. 3d ed. Washington DC: Congressional Quarterly Press.

Wald, Kenneth, Dennis Owen, and Samuel Hill, Jr. 1988. "Churches as Political Communi- ties." American Political Science Review 82: 53148.

. 1990. "Political Cohesion in Churches." Journal of Politics 52:197-215. Whitehead, Barbara Dafoe. 1993. "Dan Quayle Was Right." Atlantic 271:47-84.

Wilcox, W. Bradford. Forthcoming. "For the Sake of the Children? Family-Related Dis- course and Practice in the Mainline." In Quietly Influential: The Public Role of Mainline Protestantism, edited by R. Wuthnow. Princeton, NJ: Princeton University Press.

Wilson, James Q. 1993. "The Family-Values De- bate." Commentary 95:24-31.

Woodberry, Robert. 1998. "When Surveys Lie and People Tell the Truth: How Surveys Over- sample Church Attenders." American Socio- logical Review 63: 119-22.

Woodberry, Robert and Christian Smith. 1998. "Fundamentalism et a].: Conservative Protes- tants in America." Annual Review of Sociology 24:25-56.

Wuthnow, Robert. 1999. "Mobilizing Civic En- gagement: The Changing Impact of Religious Involvement." Pp. 331-63 in Civic Engage- ment in American Democracy, edited by T. Skocpol and M. Fiorina. Washington, DC: Brookings Institution.

Wuthnow, Robert and Matthew Lawson. 1994. "Sources of Christian Fundamentalism in the United States." Pp. 18-56 in Accounting for Fundamentalisms: The Dynamic Character of Movements, edited by M. Marty and R. S. Appleby. Chicago, IL: University of Chicago Press. Zuckerman, Alan, Nicholas Valentino, and Ezra Zuckerman. 1994. "A Structural Theory of Vote Choice: Social and Political Networks and Electoral Flows in Britain and the United States." Journal of Politics 56:1008-33.

  • Recommend Us