You are on page 1of 16

Religious Research Association, Inc.

In Search of Denominational Subcultures: Religious Affiliation and "Pro-Family" Issues


Revisited
Author(s): David A. Gay, Christopher G. Ellison and Daniel A. Powers
Source: Review of Religious Research, Vol. 38, No. 1 (Sep., 1996), pp. 3-17
Published by: Religious Research Association, Inc.
Stable URL: http://www.jstor.org/stable/3512537 .
Accessed: 22/07/2014 12:38

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 support@jstor.org.

Religious Research Association, Inc. is collaborating with JSTOR to digitize, preserve and extend access to
Review of Religious Research.

http://www.jstor.org

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
3

IN SEARCHOF DENOMINATIONAL SUBCULTURES:


RELIGIOUSAFFILIATIONAND
"PRO-FAMILY"ISSUESREVISITED*
David A. Gay

University of Central Florida

Christopher G. Ellison and Daniel A. Powers

University of Texas at Austin

Review of Religious Research,Vol. 38, No. 1 (September,1996)

While researchers have long contended that religious denominations


promote and sustain "subcultural" differences within the American public,
this claim has rarely been subjected to rigorous empirical examination.
We argue that an adequate investigation requires attention to (1) group
differences in central tendencies and (2) group differences in homogeneity.
Further, comparisons of both types of group differences should be adjusted
to account for denominational variations in sociodemographic character-
istics. Focusing on attitudes toward "pro-family" issues (e.g., attitudes
toward gender roles, abortion, sexuality), we develop such an analysis
using data from the 1982-1991 General Social Surveys. Results suggest the
existence of both "conservative" and "liberal" subcultures regarding cer-
tain of these issues. Of particular interest are several intriguing patterns of
attitudinal heterogeneity within putatively conservative denominations.
Several promising directions for future research on religious variations in
social values and attitudes are discussed.

INTRODUCTION

For a number of years, sociologists of religion have debated the role of reli-
gious collectivities, particularly denominations, in shaping individual attitudes
and behaviors. Although some have argued that most denominational variations
of this type have now diminished to the point of triviality (e.g., Wuthnow,
1988), a large and growing body of evidence indicates that the members of vari-
ous religious denominations differ significantly in a wide range of attitudes and
behaviors (Hand and Van Liere, 1984; Hertel and Hughes, 1987; Homola,
Knudsen, and Marshall, 1987; Roof and McKinney, 1987; Clarke, Beeghley,
and Cochran, 1990; Ellison, 1991a).
With the rise of the Christian Right since the 1970s, research on denomina-
tional differences focused increasingly on what Hertel and Hughes (1987) have
termed "pro-family" issues: gender role attitudes (e.g., Peek, Lowe, and
Williams, 1991); attitudes toward legal abortion (e.g., Rhodes, 1985); and atti-
tudes and practices regarding human sexuality (e.g., Bock, Beeghley, and
Mixon, 1983; Beck, Cole, and Hammond, 1991). The weight of the evidence

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
4

suggests that, with regardto "pro-family"attitudes,religious groups can be


arrayedon a continuumfrom conservative(e.g., SouthernBaptist,otherevan-
gelical or fundamentalist), to moderate(e.g., Catholic,Lutheran,Methodist),to
liberal(e.g., Episcopal,Jewish). On average,personswith no religious affilia-
tion tendto reportthe most liberalattitudesof all.
Overthe yearsresearchon denominationaldifferencesin "pro-family" orien-
tationsand other social attitudesand behaviorshas tendedto assumethat reli-
gious groups are the "locus of subculturaldifferences"(Jeffries and Tygart,
1974: 318). Althoughthe term "subculture"is often used loosely by sociolo-
gists, it generallyimplies (1) the socializationof distinctivevalues, and (2) the
transmissionof these distinctiveorientationsto new members,andacrossgener-
ations (e.g., Wolfgang and Ferracuti,1967). To date, attemptsto identify
denominational"subcultures" have relied almostexclusivelyon comparisonsof
central tendencies- i.e., mean differencesin measuresof attitudesor behav-
iors. While such comparisonsare important,the "subculture"concept also
implies thatgroupsmay differin theirrespectivelevels of homo/heterogeneity,
or the degreeof dispersionaroundthose meanvalues.
Despite its importance,however, the issue of denominationalhomogeneity
has receivedminimalempiricalattention.In one recentexceptionto this general
patternof neglect, Gay and Ellison (1993) investigateddenominationaldiffer-
ences in homogeneityof attitudestowardpoliticaltolerance,or the willingness
to extend civil liberties to unpopularelements (e.g., homosexuals,atheists).
Consistent with the received wisdom, they found that members of liberal
denominations(e.g., Episcopalians),Jews, and unaffiliatedrespondentsprofess
greatertolerancethan others, while membersof conservativedenominations
(e.g., SouthernBaptists)reportlower levels of tolerance.On the otherhand,lib-
eral denominationsand unaffiliatedpersons exhibit greaterhomogeneity(i.e.,
less dispersionaroundthe unadjustedmean values of tolerance)thantheircon-
servative counterparts.However, these variationsin attitudinalhomogeneity
may not necessarilyresultfromprocessesinternalto the religiousgroupsunder
consideration.It remainsconceivable, for instance, that Episcopalians,Jews,
and unaffiliatedpersons may be disproportionatelyand uniformly tolerant
becausethey also tend to be well-educatedand non-southern.Further,because
thatstudyfocuses only on politicaltolerance,the findingsmay not be generaliz-
ableto othersocial attitudes.
The remainderof this paperaugmentsthe substantialliteratureon denomina-
tional differencesin social attitudesin severalways. We focus on several"pro-
family"issues which have received considerableattentionin previousstudies,
includingattitudestowardgenderroles, abortion,and humansexuality.After
brieflyreportingdenominationalvariationsin centraltendencieson these issues,
we examinegroupdifferencesin the dispersionaroundthese meansusing inter-
groupvarianceratios.We proposea simple but effective strategyfor adjusting
these intergroupvarianceratiosto take into accountdenominationaldifferences
in various sociodemographicand backgroundfactors,thus surmountinga key
limitationof past researchin this area.Severalimplicationsfor furtherresearch
arediscussedin the concludingsectionof the paper.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
5

DATA AND MEASURES

Data

Like manypreviousstudiesof denominationaldifferencesin social attitudes,


our investigationuses data from the General Social Survey (hereafterGSS;
Davis and Smith, 1991), a nationalcross-sectionalsample replicatedannually
(except for 1979 and 1981) since 1972. The GSS is an appropriatedataset
becauseit containssurveyitems tappingattitudestoward"pro-family" issues, as
well as informationon the religiousaffiliation,participation,and variousback-
groundcharacteristicsof respondents.Due to the small numbersof certainreli-
gious groups(e.g., Jews, Episcopalians)in any given year of the GSS, we pool
datafromthe 1982-1991 surveys.Further,given the distinctivepatternsof reli-
gious affiliationand the distinctivehistoriesof predominantlyBlack denomina-
tions (EllisonandSherkat,1990), we have followedthe lead of a numberof pre-
vious studies(e.g., Peek et al., 1991) in restrictingour analysesto white respon-
dentsonly.

Dependent Variables: Attitudes Toward "Pro-Family" Issues

Ouranalysisfocuses on five "pro-family" issues whichhave receivedconsid-


erableattentionin previousstudies:attitudesregardinggenderroles, legal abor-
tion, and homosexual,premarital,and extramaritalsexual relations.We use the
following fouritems to measuregenderrole attitudes:(1) "Do you agreeor dis-
agreewith this statement:'Womenshouldtake care of runningtheirhomes and
leave runningthe countryto men'?" (2) "Do you approveor disapproveof a
marriedwoman earningmoney in business or industryis she has a husband
capableof supportingher?"(3) "Ifyourpartynominateda womanfor President,
would you vote for her if she were qualifiedfor the job?" (4) Tell me if you
agree or disagreewith this statement:'Most men are bettersuited emotionally
for politics thanare most women.'" Responsesto each item are dichotomously
coded. The resulting index ranges from 0 to 4 (alpha=.65;mean=3.15, st.
dev.=1.13),with higherscoresreflectingmoreegalitariangenderrole attitudes.
The GSS asks six questionsconcerningattitudestowardlegal abortionunder
the following circumstances:(1) if thereis strongrisk of seriousbirthdefects;
(2) if the woman's own healthis endangered;(3) if she became pregnantas a
resultof rape;(4) if she is marriedand does not want any more children;(5) if
she lacks the financialresourcesto raise the child; and (6) if she is unmarried
and does not wish to marrythe fatherof the child. Responsesto each item are
coded dichotomously.Althoughthese items have been used by some to form a
varietyof abortionattitudescales (e.g., HertelandHughes, 1987), it has become
commonfor researchersto constructan index by simply summingresponsesto
these six items (Wilcox, 1992; Lynxwilerand Gay, 1994). This strategyyields
an index that ranges from 0 to 6 (alpha=.84;mean=3.93,st. dev.=1.91), with
higherscoresreflectingsupportfor greateraccess to legal abortion.
Single-item indicatorsare used to gauge attitudestowardpremarital sex
(mean=2.78, st. dev.=1.23), homosexuality(mean=1.63, st. dev.=1.12), and

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
6

extramaritalsex (mean=1.37,st. dev.=.71),respectively.In each case, responses


are coded 1 ("always wrong"),2 ("almostalways wrong"), 3 ("wrongonly
sometimes"),and4 ("notwrongat all").

IndependentVariable:DenominationalAffiliation

Following Hertel and Hughes (1987) and many others,our denominational


classificationscheme includes SouthernBaptists, Other Baptists, Methodists,
Lutherans,Presbyterians,Episcopalians,Catholics,Jews, and unaffiliatedper-
sons.' In addition,we follow Roof and McKinney(1987: 253-56) in identifying
members of Other ConservativeProtestantgroups, including the following:
Assemblies of God; Church of the Brethren;Plymouth Brethren, United
Brethren,and United Brethrenin Christ;Churchof Christ;Churchesof God;
various Pentecostal and Holiness churches; Nazarene churches; Adventist
churches;ChristianAlliance and relatedgroups;IndependentBible and Open
Bible churches,andassortedsmallerevangelicalandfundamentalist groups.
For statisticalreasonsit was necessaryto exclude the membersof moderate,
liberal, and non-traditionalProtestantdenominationswith small numbersof
respondentsin the GSS (e.g., United Churchof Christ,Unitarian,Disciples of
Christ,Quaker,Mennonite,Mormonand Jehovah'sWitness)from these analy-
ses. The handfulof persons reportingaffiliationwith non-Christianreligions
besides Judaism(e.g., Islam, Buddhism,Hinduism,Baha'i) were also dropped
fromthe comparisonsfor similarreasons.

ControlVariables

Recentstudiesshow thatviews concerningourfive "pro-family" issues varies


accordingto a numberof background factors(e.g., LynxwilerandGay, 1994),and
a long traditionof researchalso indicatesthatreligiousdenominations differstrik-
ingly in the sociodemographic profile of theirmembers (see Roof and McKinney,
1987).Therefore,ouranalysesincludecontrolsfor the followingfactors:southern
residence(South= 1, U.S. Censusdesignation;33%);gender(female= 1; 57%);
age (in years;mean=45.81,st. dev.=18.05);householdincome(rescaledto per-
centiles across years; mean=56.02,st. dev.=29.36);educationalattainment(in
years;mean=12.58,st. dev.=3.04);andcommunity size (six-pointordinalmeasure,
withmoreurbanareasreceivinghigherscores;mean=2.80,st. dev.=1.43).

ANALYTICALSTRATEGY

Ourinvestigationof denominationalsubculturesproceedsin two phases.We


begin by examiningcentraltendencies,or unadjustedand adjustedmean scores
on these attitudinalindices,using analysesof varianceandcovariance.Instances
in whichthe centraltendencyof a given denominationdifferssignificantlyfrom
the centraltendencyin the remainderof the sampleare interpretedas evidence
of subculturaldistinctiveness.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
7

In the second phase of this study, we examinedifferencesin the denomina-


tionalheterogeneityof attitudestowardvarious"pro-family"issues. In brief,the
variancewithin each group is comparedwith the variance of the remaining
denominationalcategories.If the within-groupvarianceon a given "pro-family"
issue is significantlyless than the variance in the remainderof the sample
(excludingthe particularreligious group),then we have additionalevidence of
denominationalsubcultures.
To test the null hypothesis(i.e., that the varianceon a particularitem for a
given denominationis the same as the overallvarianceon thatitem), we use the
standardF-test for the equalityof variances:
F= 2YR (1)
2(YG)
where G2(YR)is the varianceof all otherremaininggroupscombinedbased on a
sampleof size nR,and 2(yG) is the varianceon a particularitem for a particular
religiousgroupbased on a sampleof size nG.In large samples,this statisticfol-
lows an F distribution with (nR- 1) and (nG - 1) degrees of freedom (Taro, 1973;
NeterandWasserman,1974).
In addition,we conduct a net (conditional)variancecomparison.Here we
apply the Goldfeld-Quandt(1965) test for homoscedasticity,which is robust
whenwe areableto identifythe criterionvariableon whichthe sampleis divided
(denominationalaffiliationin this case). Separateregressionequationsare esti-
matedfor each denominationandfor the overallsampleexcludingthatparticular
denomination.All modelsincorporatethe same set of controlvariablesdiscussed
above.The errorvariancefromthe regressionfor a given denominationis com-
paredwith the errorvariancefromthe equationfor the remainderof the sample.
In gross (unconditional)variancecomparisons,the variancearoundthe mean is
the best estimateof populationheterogeneity.However,when controlvariables
areintroduced,the variancearoundthe predictedvalueor conditionalmean(y) is
the best estimateof the net (conditional)populationheterogeneity.The Goldfeld-
Quandttest is accomplishedby use of the followingF-test:
F- - (2(yR Ix) (2
(G I ))
where 2(YRIx) is the conditionalvarianceon a particularitem froma sampleof
size nR for all remaininggroupscombined,and 2(YGIx) is the conditionalvari-
ance on an item for a particulardenominationbased on a sample of size nG.
Underthe null hypothesisof homoscedasticityand normallydistributeddistur-
bances, this ratio follows an F distribution with nR - k - 1 and nG - k - 1 degrees
of freedom,wherek is the numberof controlvariables.
In both the unconditionaland conditionalvariancecomparisontests, an F
ratio greaterthan one indicates that the varianceon an item for a particular
group is less than it is for the overall population(excluding that particular
group).Thus,F-ratiosgreaterthanone provideevidenceof denominationalsub-
cultures,while varianceratiosless thanone provideevidenceof dissensuswith-
in a particulargroup,or greaterdiversitythanthe overallpopulation.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
8

RESULTS

Mean Comparisons

Table 1 presentsdenominationalmeanson the five dependentvariables,gen-


eratedvia analysesof varianceand covariance.To conservespace, our discus-
sion focuses primarilyon the net or adjusteddenominationaldifferencesin atti-
tudes on these "pro-family"issues, althoughthe raw or unadjustedmeans are
also reported.The significancetests displayedhere are designedto gauge the
differencebetween the mean score of a given denominationalgroup and the
mean score for the remainderof the sample(less the given denomination).2 The
unique contribution of denominationalvariables to the of
explanatorypower the
overallmodels of social attitudesis generallymodest,rangingfromroughly 1-
6%.Denominational"effects"areespeciallyweak in modelsof attitudestoward
extramaritaland homosexualrelations;most Americans- religious or other-
wise - tendto disapproveof such activities.
Severalpatternsof denominationalvariationpresentedin Table 1 are espe-
Of the groupingsconsideredhere,Jewishrespondentsexpress
cially interesting.3
the most liberalattitudeson four of the five dependentvariables,and they rank
second in their toleranceof extramaritalsexual activity. Unaffiliatedpersons
also reportstrikinglyliberalviews on these "pro-family"issues. They express
greatertoleranceof extramaritalsexual relationsthan othergroupsundercon-
sideration.They ranksecondin theirtoleranceof homosexualityandpremarital
sexuality and in supportfor legal abortion,and they rank thirdin supportfor
egalitariangenderroles. While Jewish and unaffiliatedrespondentsare dispro-
portionatelywell-educated,affluent,and non-southern,theirliberalismon "pro-
family"issues persistsdespitecontrolsfor these covariates.Episcopaliansalso
tend to embracerelativelyliberal views, particularlyregardingabortionrights
andpremaritalsexual activity.However,theircomparativeliberalismon homo-
sexualityis attenuatedsomewhatby the inclusionof controlsfor socioeconomic
statusvariablesandplace of residence,as is theirrelativelyegalitarianstanceon
genderroles, althoughthis differenceremainsmarginallysignificant(p=.002)
accordingto ourrelativelystringenttest.
At the otherend of the ideologicalspectrum,SouthernBaptistsandmembers
of other fundamentalist,evangelical, and charismaticgroups reportthe most
conservativeattitudeson all "pro-family"issues. Althoughmembersof these
denominationalcategoriestendto have relativelylow levels of income andedu-
cationalattainment,and to reside in less-populatedareasand in the South,evi-
dence of a distinctivelyconservative"pro-family"subculturepersists despite
statisticaladjustmentsfor these factors.4Membersof otherBaptistgroups(e.g.,
independentBaptists) also express relatively conservativeviews on average,
althoughonce controlsareincludedthey differsignificantlyfromthe remainder
of the sampleonly in theirviews regardingpremaritalsexualrelationsandabor-
tion rights.
Between these ideological poles we find a sizeable center comprisedof
Catholics,Methodists,Presbyterians,and Lutherans.In only a few instancesdo

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
Table 1
Denominational Differences in "Pro-Family" Attitudes: Adjusted a

Gender Abortion Premarital


Roles Sex

Jewish 3.46* Jewish 5.16* Jewish 3.40* Jewish


(n=186) (3.66*) (n=223) (5.56*) (n=212) (3.46*) (n=171)
Episcopal 3.39* No Affiliation 4.85* No Affiliation 3.19* No Affili
(n=196) (3.48*) (n=715) (5.05*) (n=630) (3.49*) (n=581)
No Affiliation 3.31* Episcopal 4.69* Episcopal 3.12* Episcopa
(n=533) (3.50*) (n=258) (4.86*) (n=230) (3.07*) (n=203)
Methodist 3.29* Presbyterian 4.49* Lutheran 2.87 Catholic
(n=695) (3.24) (n=479) (4.54) (n=711) (2.81) (n=2157)

Presbyterian 3.21 Lutheran 4.24* Methodist 2.90* Luthera


(n=330) (3.26) (n=792) (4.22*) (n=827) (2.77) (n=612)
Catholic 3.16 Methodist 4.32* Presbyterian 2.89 Presbyte
(n=2007) (3.21) (n=1023) (4.25*) (n=389) (2.83) (n=423)
Lutheran 3.12 SouthernBaptist 3.64* Catholic 2.86 Methodi
(n=612) (3.12) (n=1141) (3.47*) (n=2327) (2.94*) (n=859)
OtherBaptist 3.08 OtherBaptist 3.64* OtherBaptist 2.55* OtherBa
(n=247) (3.04) (n=369) (3.58*) (n=536) (2.59*) (n=309)
SouthernBaptist 2.98* Catholic 3.55* SouthernBaptist 2.42* Souther
(n=837) (2.76*) (n=2693) (3.60*) (n=987) (2.25*) (n=958)
Other 2.83* Other 2.98* Other 2.09* Other
Conservative (2.67*) Conservative (2.80*) Conservative (1.99*) Conserv
(n=434) (n=623) (n=536) (n=530)
Overall 3.15 3.93 2.79
Totaln 6077 8334 7385
Notes:Adjustedmean,all controls(unadjustedmean).
* Meandifferentfromthatof remainderof sample,p < .001.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
10

the attitudesof theserespondentsdiffersignificantlyfromthose of the remainder


of the sample.On average,Catholicsreportconservativeviews on abortion-
indeed,theirmean score is slightlylower thanthatof SouthernBaptists- but
comparativelytolerantattitudesregardingpremaritalsexual activity.Methodists
tend to hold relativelyliberalopinionson genderroles, while Presbyteriansare
somewhatinclinedtowardthe pro-choicepositionon abortion.Suchslightdevia-
tions from the center notwithstanding,these denominationsappearto sustain
moderateviews on most of the "pro-family" issues consideredhere.

VarianceComparisons:A Testof Homogeneity

Turningto the tests of denominationalheterogeneity,Table2 displaysthe F-


ratios for the gross and net variancecomparisons.As we noted above, the F-
ratio is the varianceof all others divided by the varianceof each group.For
example,the ratioin the first column(2.49) means thatthe varianceamongall
othergroupscombinedis 2.49 times greaterthanthe variancewithinthe Jewish
category.Thus,a ratioof less thanone indicatesthatthe varianceof a particular
group is greaterthan the varianceof the remaininggroupscombined.Groups
thatarecharacterizedby substantiallyless internalvariancein "pro-family"atti-
tudes than the remainderof the sample, after adjustmentsfor differencesin
backgroundfactors,can be said to exhibit evidence of subculturalorientations
on these "pro-family"issues. In all, 27 of the 50 gross ratiosand 25 of the 50
net ratiosare statisticallysignificantat the .001 level. Again, to conservespace
our discussion focuses primarilyon the net variance comparisons,in which
denominationalvariationsin a rangeof backgroundfactorshave been takeninto
account.
Severalempiricalpatternsareespeciallyinteresting.First,the threemost lib-
eral categories- Jews, Episcopalians,and unaffiliatedrespondents- exhibit
considerablehomogeneityin their views on three "pro-family"issues: gender
roles, abortion, and premaritalsexuality. In particular,Jewish respondents
exhibit relativeuniformityin their views regardinggenderroles and access to
abortion.Given our inabilityto distinguishmembersof Reform,Conservative,
and Orthodox Jewish communities, this uniformity is especially striking.
Althoughour main interestis in relative homogeneityas an indicatorof reli-
gious subcultures,it is intriguingthatsome of the same liberalelements- par-
ticularlyunaffiliatedpersonsand Jews - exhibit substantialdissensusin their
attitudestowardhomosexualandextramaritalsexualrelations.
Second, the two most conservativesegmentsof the population- Southern
Baptists and members of other ConservativeProtestantgroups - are quite
homogeneous in their views regardinghomosexualityand extramaritalsex.
However, contraryto some popularstereotypes,and to the public pronounce-
ments of some ConservativeProtestantspokespersons,the membersof these
groups appearstrikinglydiverse in their orientationstowardgender roles. In
their attitudestowardabortionrights and premaritalsexuality,these conserva-
tives do not exhibitsignificantlymoreconsensusthanthe remainderof the sam-
ple. OtherBaptistsare notablyhomogeneousonly in theirattitudestowardpre-
maritalsexualrelations.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
Table 2
DenominationalHomo/Heterogeneityin "Pro-Family"Attitudes: Gros

Gender Abortion Premarital


Roles Sex

Jewish 2.49* Jewish 3.27* No Affiliation 1.88* Other


(2.75*) (2.65*) (2.00*) Conserv
No Affiliation 1.57* No Affiliation 1.68* Jewish 1.64* Souther
(1.63*) (1.56*) (1.63*)
Episcopal 1.43* Episcopal 1.50* Episcopal 1.53* Method
(1.43*) (1.53*) (1.46*)
Catholic 1.16* Presbyterian 1.35* Catholic 1.13 OtherB
(1.24*) (1.42*) (1.19*)
Methodist 1.13 Methodist 1.31* Lutheran 1.04 Luthera
(1.19*) (1.20*) (1.07)
Presbyterian 1.10 Lutheran 1.20 Methodist 1.09 Presbyt
(1.16*) (1.25) (1.05)
Lutheran 1.02 SouthernBaptist 1.14 Presbyterian 1.05 Catholi
(0.97) (1.16*) (1.12)
OtherBaptist 0.87 Other 1.00 SouthernBaptist 0.90 Episcop
(0.92) Conservative (1.07) (0.90)
SouthernBaptist 0.81* OtherBaptist 0.94 Other 0.83 Jewish
(0.70*) (0.99) Conservative (0.99)
Other 0.66* Catholic 0.80* OtherBaptist 0.80* No Affi
Conservative (0.69*) (0.90) (0.96)

Notes:Net varianceratio,all controls(grossvarianceratio).n's aredisplayedin Table 1.


*Variancedifferentfromthatof remainderof sample,p < .001.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
12

Together the liberal and conservative religious "blocs" account for 19 of the
25 significant net variance ratios reported in Table 2. For the most part, the vari-
ance ratios of other groups - Catholics, Methodists, Lutherans, and
Presbyterians - rarely differ substantially from those of the remainder of the
sample. Three of the six exceptions involve Catholics, who exhibit slightly
greater homogeneity than average on gender role ideology and somewhat more
internal diversity than the norm in their attitudes toward abortion and homosex-
ual relations. Methodists exhibit greater consensus than most other groups in
their views regarding abortion and homosexuality, and Presbyterians are some-
what more uniform than most groups in their views on abortion rights.

SUMMARY AND DISCUSSION

This study has revisited the popular research topic of denominational subcul-
tures, focusing on what Hertel and Hughes (1987) have termed "pro-family"
issues, such as gender roles, abortion, and sexuality. We began by analyzing dif-
ferences in central tendencies, and our results reveal substantial denominational
differences in "pro-family" attitudes, even with an array of statistical controls.
Our work confirms previous findings (e.g., Hertel and Hughes, 1987; Roof and
McKinney, 1987) that religious denominations are aligned along a liberal-mod-
erate-conservative continuum, and that these rankings are broadly consistent
across a range of "pro-family" issues, with only a handful of exceptions.
We then compared gross and net variance ratios as a way of assessing attitu-
dinal homogeneity within denominations. Here the results yield an intriguing
and sometimes counterintuitive picture. On the one hand, Jews, unaffiliated per-
sons, and Episcopalians are uniformly liberal on three sets of "pro-family"
issues: gender roles, abortion, and premarital sexuality. It is important to empha-
size that this liberal homogeneity cannot be attributedto the comparatively high
education levels, affluence, and disproportionately urban and non-southern resi-
dential concentrations that characterize the members of these groups. Thus,
although Liberal Protestant groups have sometimes been criticized as "weak"
churches (e.g., Kelley 1972) which fail to project clear values or consistent stan-
dards, such characterizations seem to be inaccurate. On the contrary, apparently
some religious groups (e.g., Episcopalians) have sustained viable subcultures of
liberalism despite membership losses and internal conflicts, particularly over
such issues as the ordination of women and liberal social activism.
Some of the more conservativedenominations(e.g., SouthernBaptists)exhibit no
less internaldiversitythanothergroupson such issues as abortionandpremaritalsexu-
ality, and they exhibit significantlygreaterinternalheterogeneityon gender role atti-
tudes. These findings may surprisemany readers.It has been widely assunredthat the
institutionalcharacteristicsand belief systems of some ConservativeProtestantgroups
demand an exceptional degree of ideological and behavioralconformity from their
members,and that these groups are more willing to impose - and are better able to
enforce - sanctionsagainstdeviantsthanotherreligious groups(Kelley, 1972; Wald,
Owen, and Hill, 1989; Wald, Hill, and Owen, 1990; Iannaccone,1994). Our findings,
like those of Gay and Ellison (1993) and others (e.g., Pargamentet al., 1987), suggest
thatsuch characterizationsmay be exaggerated.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
13

While this observed conservativeattitudinalheterogeneityruns counterto


some popularimages of these groups,it dovetailsneatlywith severalstrandsof
recentresearch.In brief, growingattentionto the conflicts within the Southern
BaptistConvention,and amongotherConservativeProtestantgroupsand orga-
nizations,makesit clearthatcontemporaryevangelicalismis far frommonolith-
ic, socially or politically (e.g., Ammerman,1990). The work of Hunter(1983,
1987) and othersalso calls attentionto the potentiallytransformativeimpactof
two sets of processes within evangelicalcommunities:(1) "cognitivebargain-
ing," or selective accommodationto (and incorporationof) changes in the
social, economic,andpoliticalenvironment;and (2) generationalchange,or the
maturationof a youngercohort of evangelicalswith more flexible theological
andsocial valuesthanthose of theirparents.Moreover,recentethnographicevi-
dence indicatesthatConservativeProtestantattitudesand practicesin the areas
of genderandfamilylife (e.g., householddivisionof labor,femaleemployment)
aremorenuancedandnegotiatedthanpreviouslyrecognized(McNamara,1984;
Rose, 1987; Stacey, 1990;Demmitt,1992).
Althoughconservativedenominationsappearsomewhatdividedover gender
rolesandpremarital sexuality,they arestrikinglyhomogeneousin theiropposition
to homosexualandextramarital sexualrelations.Thesemaybe moresalientissues
for ConservativeProtestantsthan for others,because membersof these groups
havetraditionally viewed sexualimproprietyas gravesin, tangibleevidenceof the
spiritualand moralthreatthatSatan'stemptationposes to individualsand to the
broadersocialfabric.Further,ourfindingsmay also reflectthe impactof addition-
al developments:(1) contemporary constructionsof the conservativedoctrineof
biblicalinerrancy,whichhighlightspecificscriptural passagescondemninghomo-
sexuality;and (2) recenteffortsby ConservativeProtestantleadersandorganiza-
tions to mobilizereligiousconservativesaroundsocial and culturalforces which
(allegedly)threatenthe traditionalfamilyform,includinghomosexualityandlurid
mediadepictionsof sexuality,includingextramarital sexualrelations.
Whileouranalysisof "pro-family" issues locatesbothliberalandconservative
subcultures, it is moredifficultto identifyclearsubculturesof moderationon these
issues. MainlineProtestantbodiesmay trulybe culturallydiverse,for the reasons
notedat the outsetof this study.However,it is also possiblethatourresultsmay
underestimatethe extent of moderate "pro-family"subcultures,due to the
unavoidablycoarsedenominational classification.Priorto the mid-1980sthe GSS
codingschemedid not distinguishamongbranchesof the variousProtestanttradi-
tions,includingPresbyterian andLutherandivisions(see note 1).
Methodists,however,do appearto sustaina moderatesubcultureof sorts.As
a groupthey (1) tend to reportcentristviews on most issues, and (2) are charac-
terizedby greaterattitudinalhomogeneitythanthe remainderof the sampleon
two of the most emotional "pro-family"issues, abortionand homosexuality.
This patternis somewhatsurprisingbecause the United MethodistChurchhas
been characterizedby many of the dynamicsthat are often thoughtto promote
internaldiversity,includingprominentideologicalconflicts and vibrantspecial
purposegroups- particularlythe charismaticand theologicallyconservative
Good News movement (Keyson, 1980). On the other hand, contemporary
AmericanCatholicpublic opinion - while also centristin many respects
appearsnotablydividedon these sametwo issues.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
14

Our findings concerningdenominationalsubculturescould be clarifiedand


extendedby severaltypes of research.First,althoughwe have followedconven-
tion in restrictingour analysis to white respondents,futurework should also
focus on African-Americans. While attitudeson many issues (e.g., civil rights,
the role of religionin politics and serviceprovision)seem to cut acrossinstitu-
tional and theologicaldivides withinthe Black religiousestablishment(Lincoln
andMamiya,1990),on some otherissues thereis evidenceof non-trivialdenom-
inationaldifferencesin African-American public opinion(e.g., Ellison, 1991b).
To the best of our knowledge,thereis little systematicinformationon the con-
toursof denominational subcultureswithinthe African-American community.
Second,futureinvestigationsshouldseek to clarifyhow denominational sub-
culturesin the generalpopulationare sustainedand transmitted.For instance,
recentcase studieshaveimplicateda varietyof mechanisms,includingdenomina-
tional gatherings(e.g., conferences,regionalmeetings),church-affiliated educa-
tionalinstitutions,campusministries,and otherfactors(see the essays in Carroll
and Roof [1993]). Many of these factorsmay help to explainthe existenceof
some subcultural orientationsamongConservativeProtestantdenominations (e.g.,
SouthernBaptists),althoughthey do not explainwhy such groupsexhibithomo-
geneity on some "pro-family" issues but dissensuson others.Meanwhile,addi-
tionalworkis still neededto accountfor the attitudinal homogeneityamongliber-
al segmentsof thepopulation,suchas Episcopaliansandunaffiliatedpersons.
Third,a numberof scholars(e.g., Wuthnow,1988) have arguedthatdenomi-
nationaldistinctivenesshas declinedin recentdecades,due to the internaltrans-
formationof religious organizationsas well as various shifts in the broader
social environmentnotedat the outsetof this paper.While an empiricalinvesti-
gationof trendslies beyondthe scope of our study,such changesin the pattern-
ing of denominationalsubcultures- gaugedvia measuresof centraltendency
andhomogeneity- could and shouldbe examinedusing datafromthe General
Social Surveysand/orotherlarge-scalesurveydata.
Finally,althoughwe have focused on denominationalsubcultures,in recent
yearsprominentscholarshaverecommendedcloserattentionto the congregation
as a key unitof analysisin the sociologyof religion.Indeed,a fullerunderstand-
ing of how ideologicalandbehavioraluniformityis producedandsustainedwithin
denominationsshould ultimatelyfocus on fellowship ties and social network
dynamicswhichoperateprimarilyat the congregational level (e.g., Olson, 1993).
Forinstance,specificlifestylenormsmaybe relayedvia "moralmessages"encod-
ed in sermonsand otherformalcommunicationsfrom religiousauthorities,and
may be reinforcedthroughinformalsocial interactionwith coreligionists.The
prospectof rejectingor violatingthese deeply-heldvaluesmay inducefeelingsof
shameand guilt (Grasmick,Bursik,and Cochran,1991). Religiousgroupsmay
also constitutereference groups for their members (Clarke, Beeghley, and
Cochran,1990). Congregationalmembersmay voluntarilyadjusttheir conduct
and values to make them consistentwith those of key individualsthey consider
importantor worthyof emulation(e.g., clergy,lay elites),or withthe broadernor-
mativeclimatewithintheirchurch.On the otherhand,social coercionmay be at
workwithinsome religiouscongregations,as memberswho rejectpopularviews
on theologicalandsocialissuesriskformalor informalexpressionsof disapproval
or censurefromfellow membersor churchleaders.Selectionmay also shapepat-

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
15

terns of homogeneity, with disaffected church members switching their allegiance


to congregations or denominations where more sympathetic views prevail.
Although the interpersonalmechanisms which shape variations in ideological and
behavioral conformity within religious groups remain underresearchedand poorly
understood, they deserve more sustained attentionfrom researchersin the future.

NOTES

*An earlierversion of this paperwas presentedat the 1994 meetingof the Southern
Sociological Society, Raleigh, NC. Data were made available via the Interuniversity
Consortiumfor Politicaland Social Research(ICPSR),Ann Arbor.We wish to thankthe
anonymousreviewersfor helpful commentson an earlierdraft.The authorsalone are
presentedhere.
responsiblefor the analysesandinterpretations
1. Priorto themid-1980s,the GSS denominational codingschemeidentifiedonly gener-
al categories-"Baptist," "Lutheran," and so forth.Even with the use of a moredetailed
scheme, however, many respondentshave neglected to specify the precise "Baptist,"
"Lutheran," or otherProtestantbodiesto whichtheybelong.Consequently,like otherswho
have used GSS denominational data(e.g., HertelandHughes,1987),we areforcedto rely
on relativelybroad denominationalcategories.We do, however, attemptto distinguish
"SouthernBaptists"(coarselyidentifiedhereas Baptistswho also residein a southernstate,
accordingto U.S. Censusdesignation)from"OtherBaptists."Underthe circumstances, this
is a justifiablestrategyfor the following reasons:(1) the SouthernBaptistConvention
(SBC) is vastlylargerthanany otherBaptistbody;(2) most white Baptistsresidingin the
Southare associatedwith SBC-membercongregations;(3) althoughwe acknowledgethe
riskthatour strategymisclassifiessome membersof independentBaptistchurchesas well,
these are also likely to endorseconservativepositionson "pro-family" issues; (4) SBC-
membercongregationsare much less common outside the US Census South, although
recentevidence suggeststhat theirnumbersare increasing(e.g., Shibley, 1991), and we
acknowledgethatwe may misclassifya smallnumberof SBC memberswho residein non-
southernstates;(5) therearevirtuallyno American(Northern)Baptistcongregationsin the
South,save thosewhichhavejoint affiliations(withthe SBC);and(6) becausethe studyis
restrictedto whites,the risk of measurementerrorvia the inclusionof membersof Black
Baptistdenominations - whichalso havelargesouthernconstituencies- is virtuallynil.
2. Ouranalysesare designedto comparethe meanandvarianceof a given denomina-
tion with those of all othersin the sample(i.e., all otherrespondentsexcept for members
of the given denominationalgrouping).Althougha more detailedcomparisonof each
denominationwith each otherdenominationwould also be interesting,such an investiga-
tion lies beyondthe scope of ourstudy.
3. It bearsmentioningthatthesedenominational categoriesareconstructed fromitemstap-
pingreligiouspreferenceratherthanactualchurchmembership. Information on congregation-
al membership is unavailablein mostyearsof the GSS, andthe itemson (a) strengthof reli-
gious identity(i.e., the RELITENitem)and(b) membership in church-affiliated
groups(i.e.,
theMEMCHURH item)arenot adequatesubstitutes. However,one suspectsthatmanyof the
patternsidentifiedherewouldbe morepronounced to churchmembers.
in a samplerestricted
4. At the sametime, it wouldbe a mistaketo exaggeratethe views of the most conser-
vative groups.Even among the most conservativedenominations,for instance,there is
actuallywidespreaddiscontentwith "traditional" patriarchalgenderrole definitions(on
average,on nearlythreeof the four items), and thereis non-trivialopennessto abortion
undercertaincircumstances(on average,threeof the six conditions).

REFERENCES

Ammerman,NancyT.
1990 BaptistBattles:Social Changeand ReligiousConflictin the SouthernBaptist
Convention.New Brunswick,NJ:RutgersUniversityPress.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
16

Beck, Scott,BettieCole, andJudithHammond


1991 "ReligiousHeritageand PremaritalSex: EvidenceFroma National Sample
of YoungAdults."Journalforthe Scientificof Religion30:173-180.
Bock, E. Wilbur,LeonardBeeghley,andRonaldMixon
1983 "Religion,SocioeconomicStatus, and Sexual Morality:An Applicationof
ReferenceGroupTheory."SociologicalQuarterly24:545-559.
Carroll,JacksonandWadeClarkRoof (eds.)
1993 Beyond Establishment:Protestant Identity in a Post-Protestant Age.
Louisville:Westminster/John Knox.
Clarke,Leslie, LeonardBeeghley,andJohnCochran
1990 "Religiosity,Social Class, and Alcohol Use: An Applicationof Reference
GroupTheory."SociologicalPerspectives33:201-218.
Davis, JamesA. andTomW. Smith
1991 The General Social Surveys: CumulativeCodebook,1972-1991. Chicago:
NationalOpinionResearchCenter.
Demmitt,Kevin
1992 "Loosening the Ties That Bind: The Accommodationof Dual-Earner
Familiesin a ConservativeProtestantChurch."Reviewof ReligiousResearch
34:3-19.
Ellison,ChristopherG.
1991a "ReligiousInvolvementand SubjectiveWell-Being."Journalof Healthand
Social Behavior32:80-99.
1991b "Identification and Separatism: Religious Involvement and Racial
OrientationsAmongBlack Americans."SociologicalQuarterly32:477-494.
Ellison,ChristopherG. andDarrenE. Sherkat
1990 "Patternsof Religious Mobility Among Black Americans."Sociological
Quarterly31:551-568.
Gay,DavidA. andChristopher G. Ellison
1993 "Religious Subculturesand Political Tolerance:Do DenominationsStill
Matter?"Reviewof ReligiousResearch34:311-332.
Goldfeld,S. andR. Quandt
1965 "Some Tests for Homoscedasticity."Journal of the American Statistical
Association60:539-547.
Grasmick,HaroldG., RobertJ. Bursik,andJohnCochran
1991 "'Render Unto Caesar What is Caesar's': Religiosity and Taxpayers'
Inclinationsto Cheat."SociologicalQuarterly32:251-266.
Hand,CarlandKentD. VanLiere
1984 "Religion,Mastery-over-nature, andEnvironmental Concern."Social Forces
63:555-570.
Hertel,BradleyandMichaelHughes
1987 "ReligiousAffiliation,Attendance,and Supportfor 'Pro-Family'Issues in
the UnitedStates."Social Forces65:858-882.
Homola,Michael,Dean Knudsen,andHarveyMarshall
1987 "ReligionandSocioeconomicAchievement."Journalfor the ScientificStudy
of Religion26:201-217.
Hunter,JamesDavison
1983 American Evangelicalism: ConservativeReligion and the Quandary of
Modernity.New Brunswick,NJ:RutgersUniversityPress.
1987 Evangelicalism:The Coming Generation.Chicago:Universityof Chicago
Press.
Iannaccone,LaurenceR.
1994 "WhyStrictChurchesareStrong." AmericanJournalof Sociology99:1180-1211.
Jeffries,VincentandClarenceTygart
1974 "TheInfluenceof Theology,Denominationand Values upon the Positionsof
Clergyon SocialIssues."JournalfortheScientificStudyof Religion13:309-324.
Kelley, Dean
1972 WhyConservativeChurchesare Growing.New York:HarperandRow.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions
17

Keyson,Charles
1980 "TheStoryof GoodNews."GoodNews 14:9-73.
Lincoln,C. EricandLawrenceH. Mamiya
1990 The Black Churchin the AfricanAmericanExperience.Durham,NC: Duke
UniversityPress.
Lynxwiler,JohnandDavidA. Gay
1994 "ReconsideringRace Differences in Abortion Attitudes."Social Science
Quarterly75:67-84.
McNamara,PatrickH.
1984 "ConservativeChristianFamiliesand TheirMoralWorld:Some Reflections
for Sociologists."SociologicalAnalysis46:93-99.
Neter,JohnandWilliamWasserman
1974 Applied Linear StatisticalModels: Regression,Analysis of Variance,and
ExperimentalDesigns. Homewood,Illinois:RichardIrwin,Inc.
Olson,DanielV.A.
1993 "FellowshipTies and the Transmissionof Religious Identity."Pp. 32-53 in
Jackson Carroll and Wade Clark Roof (eds.), Beyond Establishment:
ProtestantIdentityin a Post-ProtestantAge. Louisville: Westminster/John
Knox.
Pargament,KennethI., RubenJ. Echemendia,StevenJohnson,PaulCook,Cheryl
McGrath,JenniferG. Myers,andMichaelBrannick
1987 "The ConservativeChurch:Psychosocial Advantagesand Disadvantages."
AmericanJournalof Community Psychology15:269-286.
Peek, CharlesW., GeorgeD. Lowe, andL. SusanWilliams
1991 "Genderand God's Word:AnotherLook at Religious Fundamentalism and
Sexism."Social Forces 69:1205-1221.
Rhodes,A. Lewis
1985 "Religionand Oppositionto AbortionReconsidered."Review of Religious
Research27:158-168.
Roof, WadeClarkandWilliamMcKinney
1987 AmericanMainlineReligion.New Brunswick,NJ:RutgersUniversityPress.
Rose, SusanD.
1987 "Women Warriors: The Negotiation of Gender in a Charismatic
Community."SociologicalAnalysis48:245-258.
Shibley,MarkA.
1991 "The Southernizationof American Religion: Testing a Hypothesis."
SociologicalAnalysis51:159-174.
Stacey,Judith
1990 BraveNew Families.New York:Basic Books.
Taro,Yamane
1973 Statistics:An Introductory Analysis,3rded. New York:HarperandRow.
Wald,Kenneth,SamuelHill, Jr.andDennisOwen
1990 "PoliticalCohesionin Churches."Journalof Politics 52:197-215.
Wald,Kenneth,DennisOwen,andSamuelHill, Jr.
1989 "Habitsof the Mind?The Problemof Authorityin the New ChristianRight."
Pp. 93-108 in Ted G. Jelen (ed.), Religion and Political Behavior in the
UnitedStates.New York:Praeger.
Wilcox, Clyde
1992 "Race, Religion, Region and Abortion Attitudes."Sociological Analysis
53:97-105.
Wolfgang,MarvinandFrancoFerracuti
1967 TheSubcultureof Violence.London:Tavistock.
Wuthnow,Robert
1988 TheRestructuringof AmericanReligion.Princeton,NJ: PrincetonUniversity
Press.

This content downloaded from 132.170.130.214 on Tue, 22 Jul 2014 12:38:31 PM


All use subject to JSTOR Terms and Conditions

You might also like