Regional Subcultures of the United States

Home , Cultural geography, Individualistic culture, Upland South

Cleveland State University EngagedScholarship@CSU

Political Science Faculty Publications Political Science Department

11-1993

Joel A. Lieske Cleveland State University, [email protected]

Follow this and additional works at: https://engagedscholarship.csuohio.edu/clpolsci_facpub

Part of the Political Science Commons How does access to this work benefit ou?y Let us know! Publisher's Statement Copyright 1993 Cambridge University Press. Available on publisher's site at http://journals.cambridge.org/abstract_S0022381600076234.

Original Citation Lieske, Joel. 1993. "Regional Subcultures of the United States." Journal of Politics 55:888-913.

Repository Citation Lieske, Joel A., "Regional Subcultures of the United States" (1993). Political Science Faculty Publications. 11. https://engagedscholarship.csuohio.edu/clpolsci_facpub/11

This Article is brought to you for free and open access by the Political Science Department at EngagedScholarship@CSU. It has been accepted for inclusion in Political Science Faculty Publications by an authorized administrator of EngagedScholarship@CSU. For more information, please contact [email protected]. RegionalSubcultures of the UnitedStates

Joel Lieske ClevelandState University

This study develops the case, theory, and statisticalmethodology for a new measureof American regional subcultures.Using principalcomponent and cluster analysis on some 45 measuresof racial origin, ethnic ancestry,religious affiliation,and social structure,I show how the entire populationof U.S. counties can be partitionedinto 10 distinctive, regionalsubcultures that are relativelyhomoge- neous and contiguous. Next, I identify the cultural characteristicsof each subcultureand relate my new classificationscheme to the work of Elazar,Gastil, Garreau,and Fischer. Finally, I comparethe relative utility of this new measurewith Elazar'stypology in explainingthe variationin a number of social, political,and policy indicators.

More work, in our view, needs to be done, to ascertain whether the cultural clusters that Elazar designates do indeed exist (either at the mass or elite level). This step, it seems to us, should precede attempts to link public policy outputs with political cultures. Unfortu- nately most work has startedwith the second step and assumed the first. (Thompson, Ellis, and Wildavsky1990, 245)

]wenty-seven years ago, Elazar(1984) advanceda revolutionarynew theory of American regional subcultures. This new theory was designed to help explain variationsin the politicalprocesses, institutional structures, political behavior, and policies and programsof state and local government.In short, Elazarargued that subculturaldifferences could help us understand:(1) what state and local governments do, (2) how they are organized,(3) what political rules they observe, and (4) who participatesin the politicalprocess (Press and VerBurg1983). To expedite this task, Elazardeveloped a geopoliticalmapping of the United States that classifies states and local regions into one or a combinationof three political subcultures:individualistic, moralistic, and traditionalistic.In Elazar's typology, there are a total of nine different subculturalclassifications that represent dominantand subordinatecombinations of his three "core"subcultures. His mappingincludes some 50 state and 224 substatedesignations. Since Elazaradvanced his theory, other scholarshave developedtheir own cultural classificationschemes and produced their own cultural mappings of the

This study owes a heavy intellectualdebt to Dan Elazar,Aaron Wildavsky, and my departedfriend and colleague,Tom Flinn. None of these individuals,of course, bearsany responsibilityfor any errors of analysis or interpretation.I also wish to thank Elazar, the Center for the Study of Federalismat Temple University,and ClevelandState Universityfor their financialsupport of my research.

THE JOURNAL OF POLITICS, Vol. 55, No. 4, November 1993, Pp. 888-913 C 1993 by the Universityof Texas Press RegionalSubcultures of the United States 889

United States (Luttbeg 1971; Zelinsky 1973; Gastil 1975; Garreau1981; Morgan and England 1987). However, Luttbeg's (1971) and Morganand England's(1987) classificationschemes are the only mappingsbased on replicable,statistical procedures. And Elazar's(1984) typology is the only one that has been subjected to extensiveempirical tests. In reviewingthis literature,Kincaid and I (1991)identify approximately 100 studies that have subjectedElazar's thesis or a subculturalvariation to empiricaltests (e.g., Sharkansky1969; Weber and Shaffer 1972; Johnson 1976; Joslyn 1980; Lowery and Sigelman 1982; Herzik 1985; Fitzpatrickand Hero 1988; Nardulli 1990;Morgan and Watson 1991; Dran, Albritton,and Wyckoff 1991). But as we note, one of the most frequentcriticisms made of this developingliterature con- cerns the difficultiesof measuringpolitical culture. It is well-knownthat Elazar's derivationof the three politicalsubcultures is not based on any rigorousstatistical procedures.Neither are his geopoliticaldesignations based on any empiricaldata, other than historical migrationpatterns, personal field observations,interviews, and scholarly studies of America'sregions2_sections, and ethnoreligiousgroups. Moreover, Elazarhas not adjustedhis mapping of the three subculturesfor the pervasive cultural changes that have occurred in American society. Indeed, his conceptof "culturalgeology" allows for overlayand change.But his state and substate classificationsare the same today as they were 27 yearsago. Yet another criticism of Elazar'sclassification scheme is the crudeness of his state and regionaldesignations. According to Elazar(1970, 236), political culture can be definedas the persistent,generational patterns of politicalattitudes, values, beliefs, and behaviorthat characterizea group of people who live in geographical proximity.But what constitutes distinctive subculturesis often a very subjective judgment.And when the geographicalboundaries of a subcultureinclude entire states or large substate regions, there is not a great deal of empiricalprecision. Since the county is the basic unit of state government,it would seem desirableto develop measures that differentiateAmerican subcultures at least down to the county level. A final criticism of many culturalclassification schemes, including Elazar's,is their circularity.Thus, it is often alleged that Elazar'stypology relies, in part, on informationabout past politicalbehavior to predict currentor future politicalbehavior. To get around this difficulty, Kincaid and I (1991) have proposed the developmentof new subculturalmeasures exclusively based on what Elazarhas called the "sources"of political culture such as racial origin, ethnic ancestry, religious affiliation, and social structure. This focus would exclude from the derivationof any new measuresthe confounding"manifestations" and "effects"of culture,i.e., politicalbehavior and institutionalarrangements. The purposeof this study is to develop and analyzea new measureof American subculturethat has these properties: 1. It is derivablefrom an explicit and replicableset of mathematicaland statistical algorithms. 890 Joel Lieske

2. It reflectscurrent cultural conditions. 3. It distinguishessubcultural differences down to the county level. 4. It is based entirelyon "nonpolitical"measures of Americanculture. To achievethese objectives,I will first discuss the theoreticallogic that undergirds my new measure. Next I will present the statisticalmethodology on which it is based. After this, I will discuss the distinguishing characteristicsof each subculture and relate my classificationscheme to the work of Elazar (1984), Gastil (1975), Garreau(1981), and Fischer (1989). Finally, since Elazar'stypology is the only one that has been subjectedto extensive empiricalanalysis, I will test the empirical utility of my new measurerelative to his. This will be done by comparing their relative ability to explain the variationin a number of social, political, and policy indicators.

A MODELOF AMERICANREGIONAL SUBCULTURES As Kincaid and I (1991) argue elsewhere,1the peoples who settled the United States came with distinctive ethnoreligiousidentities, cultural preferences,and ways of life (Fischer 1989). In most cases, these traditions shaped both their choices of and responsesto the environment.Generally, groups settled in clusters, not as randomlydispersed individuals. And as they pushed inland, as in the westward migrationof British-Americansacross the entire continent, they took with them and transmitted to future generations the values, beliefs, attitudes, and behavior of their forebears(Holbrook 1950). Moreover, given the pervasiveness of local self-goverment that constituted the American democraticexperiment, groups could give significantsocial and politicalexpression to their culturalpreferences within geographicallydefined political jurisdictions, namely, towns, town- ships, cities, and counties (Kincaid 1982). This gave an advantageto the early settlers who founded the first local and state governments(Zelinsky 1973; Gastil 1975). Unless they could numericallyoverwhelm these early settlers, laterarrivals had to contend with and adapt themselves to existing social and political power structures. Based on this historicalrecord, American subcultures should be dispersed,but not balkanized,across the countryin explicablepatterns or mosaicsthat are rooted in political jurisdictions.In addition, single subculturesor distinctive multicul- tural blends should be found in most local jurisdictionsbecause most states were settled and developedby diverse groups that clusteredgeographically within substate regions. Finally, the continuity and persistenceof these local subculturesis predicatedon two characteristicfeatures of Americansocial and politicalculture: ethnoreligioussocialization and federaldemocracy. For most Americans,socialization occurs within the ethnoreligiouskinship and neighborhoodties of small and medium-size jurisdictions.Even in large cities,

' This section drawson the ideas and contributionsof John Kincaidto this paper. RegionalSubcultures of the United States 891 ethnoreligiousgroups tend to cluster in neighborhoodsand ethnic enclaves, the so-called "turfs"of urban society, if not also in precincts, wards, and legislative districts. Thus, unless the mix of ethnoreligiousgroups changes or a dominant group is displaced,the prevailingsubculture is containedwithin a bounded set of economic and social relationshipsthat can be communicatedacross generations (Kincaid 1980;Lieske 1990). Federaldemocracy, in turn, has long entailedsubstantial local self-government and therefore has served, until recently, to bolster local autonomy and control. This has enabled jurisdictionallybased subculturesto institutionalizetheir cultural preferences,as reflected in local control of law enforcementpractices and state control of social welfareservices. Currenttrends in the federalsystem, however, both constrainand expand the ability of state and local governmentsto give expressionto their culturalpreferences in public institutionsand policies. On the one hand, there has been an unprecedentedincrease since the 1960s in the federal preemptionof state and local powersand federalmandating of state and local policies (Kincaid 1990). On the other hand, the federal governmenthas restored, or continued to allow, state and local discretionin a number of areas, such as "the new judicialfederalism." Under this doctrine state courts may rely on their state constitutionsto grantbroader protections in the culturallysensitive areaof rights than the U.S. Supreme Court is willing to grant under the federal Constitution (Collins, Galie, and Kincaid 1986). In sum, patterns of settlement and local self-governancein the United States have nurtured the evolution and development of jurisdictionallybased subcultures. Thus, differentethnoreligious groups, interactingwith their environments, have been able to translatetheir cultural preferencesinto relativelystable social relationships within substantially self-governing jurisdictions. These relationships, in turn, reinforcethe dominantcultural preferences in an areaand provide bounded nests for enduringpatterns of socialization,communication, and action. At the same time, migration, pluralism, Americanization,and diverse frontier conditions created different cultural mixes and responses, thereby giving rise to subcultures rather than to the very distinct, territorial-basedcultures found in countriessuch as India and Yugoslavia.Regional subcultures, then, are the prod- ucts of historicalinteractions between the culturalpreferences of differentethnoreligious settler groups and the nationallycentripetal and regionallycentrifugal demands of their environments.Therefore, they represent the historical extensions of earliersettlement patternsand the continuing advantagesof the first ef- fective settler groups (i.e., the founding groups) over later arrivals(i.e., newer groups)who are placed in the culturallysubordinate position of challengingsocial conventionsand tradition. Following Wildavsky's (1987) concept of "general"culture, as opposed to Elazar's(1970) concept of "political"culture, I view a subcultureas a "wayof life" and system of sharedvalues that legitimatea preferredset of social relationships. Conceptually, cultural preferences are derived from the learning that occurs 892 Joel Lieske within the web of social relationshipsin which people are embeddedand on which they are especiallydependent during their formativeyears. Accordingto Dreitzel (1977) and Wildavsky (1987), a culture shapes preferencesby answering three fundamentalquestions: (1) Who am I? (2) How should I behave?and (3) What is legitimate?Thus, culture provides individuals with their basic social identities (the identity function), norms for socially acceptablebehavior (the boundary- maintenancefunction), and standardsfor judgingsocial institutions (the legitimat- ing function).Presumably, these functionsare realizedprimarily through people's racialand ethnic kinshipties, religiousor secularvalue systems, and social ways of life and life-styles. Conceptually,these socializingagents define the culturalcon- text, or subculture,that shapesindividual preferences and behavior. In figure 1, I present a causal frameworkdepicting these assumptions.In this model, racial-ethnicancestries and religiousaffiliations are assumedto be the pri- mary carriersof subculturaldifferences (Elazar 1970). Throughouthistory, social differenceshave been defined largelyby differencesin racial-ethnicand religious background.Race and ethnicity representthe accumulatedlearning or history of a people who share a common ancestralidentity and usually a languageor dialect.

FIGURE 1

A SUBCULTURAL MODEL OF SOCIAL AND POLITICAL BEHAVIOR

Social Structure

Race-Ethnic Ancestry\

Social and Political

Behavior

Religious Affiliation

Regional Subculture RegionalSubcultures of the United States 893

Religion is a formal belief system that endeavorsto organizesocial relationships accordingto its tenets. Social structures,in turn, serve to define alternativeways of life or social life- styles. They representthe institutionalizedsocial preferencesof different racial- ethnic and religious groups. However, they also reflect the influence of certain environmentalconstraints, such as geography,climate, and the local economy. Thus, the work of many cultural theorists (Elazar 1970; Kleppner 1970; Gastil 1975; Garreau1981; Fischer 1989;Leege, Lieske, and Wald 1991) and behavioral researchers(Lazarsfeld, Berelson, and Gaudet 1944; Laswell and Kaplan 1950; Campbell,Converse, Miller, and Stokes 1960) suggests that many differencesin social life-style result from a mix of ethnoreligious,environmental, and economic conditions. These include: (1) the advanceof urbanizationand industrialization, (2) the persistenceof agrariantraditions, (3) populationmobility, (4) social status, (5) socialinequality, (6) familystructure, (7) life cycle, and (8) culturalhomogeneity. To the extent social relationshipsembody the institutionalizedsocial preferences of differentracial-ethnic and religiousgroups, they constitutea culture'spreferred way of life. Finally, a regionalsubculture can be conceptualizedas a weightedsum of the racial-ethnicidentifications, religious attachments,and social relationships thatcharacterize a groupof individualswho live in geographicalproximity.

Data andMethod The sample consists of all 3,164 U.S. counties. The data were drawnfrom the 1980 census and the 1980 Glenmary(1982) surveyof Americanchurch bodies. To representthe variablesin the culturalmodel, I selected 5 indicatorsof racialori- gin, 11 indicatorsof ethnic ancestry,14 indicatorsof religiousaffiliation, and 15 indicators of social structure. The racial and ethnic indicators contain the major groups tabulatedin the 1980 census; while the religious indicatorsembrace the 13 largestChristian denominations in the United States plus conservativeand re- form Judaism. About 99 million Americans,or 43.7% of the total populationin 1980, are affiliatedwith these church bodies. Accordingto estimatesdeveloped by Wald (1987), this figure represents85.8% of all church members in the United States. Finally, the structuralindicators were selected to measure differencesin the factors hypothesizedto have the greatestinfluence in defining social ways of life and life-styles. The five racialindicators include: (1) percent white, (2) percent black, (3) percent Asian, (4) percent Native American,and (5) percent Spanish-speaking.2The 11 ethnic indicatorsinclude: (1) percent British, (2) percent German,(3) percent Irish, (4) percent French, (5) percent Italian,(6) percent Slavic, (7) percent Scan- dinavian, (8) percent Russian, (9) percent Hungarian,(10) percent Portuguese, and (11) percentGreek. The 14 religiousindicators include: (1) percent Catholic,

2Strictly speaking, Spanish-speakingis not a racial but a linguistic designation, since Spanish- speakingcan be of any group. However, most Spanish-speakingAmericans are thought to be mestizo, i.e., of mixed Europeanand Native Americanancestry. 894 Joel Lieske

(2) percent Southern Baptist, (3) percent United Methodist, (4) percent United Presbyterian,(5) percent Lutheran Church of America, (6) percent Episcopal, (7) percent Mormon (The Church of Jesus Christ of the Latter-Day Saints), (8) percent Missouri-SynodLutheran, (9) percent AmericanLutheran, (10) percent United Churchof Christ, (11) percentAmerican Baptist, (12) percent Chris- tian (Disciples of Christ), (13) percent Southern Presbyterian,and (14) percent conservativeor reformJudaism. The 15 structuralindicators include: three measuresof urbanizationand industrialization(total population,percent urban population,and percent of the labor force in manufacturingjobs); two indicatorsof agrarianism(percent of the labor force in agrarianoccupations and percentof the laborforce in naturalresource oc- cupations);one indicatorof populationmobility (percentliving in the same house in 1975);three indicatorsof social status (percentof all people 25 and older with a college degree,percent of the laborforce in managerialoccupations, and percentof the laborforce in professionaloccupations); one indicatorof income inequality(the index of dissimilarity,on a scale from 0 to 1, in the distributionof income);two indicators of family structure(the percent of all families that are traditionaltwo- parentand the percentof all femalesin the laborforce); two indicatorsof life cycle (the percentof the populationaged 18-29 and the percentof the populationaged 65 and over);and one indicator(Lieske 1988)of culturalhomogeneity (an index of homogeneity,on a scale from 0 to 1, in the distributionof the majorracial groups tabulatedin the 1980 census-whites, blacks,Asians, and Native Americans).3 To classify the universeof counties on the basis of these 45 selected indicators, I consideredseveral statisticaltechniques. Perhaps the two most commonly used approacheshave been the applicationof Q-factorand R-factoranalysis to selected cultural measures (Luttbeg 1971; Savage 1973; Monroe 1977; Morgan and En- gland 1987). In Q-factoranalysis, the datahave to be transposedbefore calculating the correlationmatrix. The correlationsmeasure the extent to which each pair of units share common culturaltraits. The usual algorithmis to assign a unit to the factor on which it has the highest absolute factor loading. This technique, however, is indeterminant(i.e., not workable)when the numberof geographicalunits that need to be classified (3,164) exceeds the number of indicatorsavailable for analysis(45). Similarly, R-factor analysis is good for analyzing variable relationshipsand for purposes of data reduction. But by itself, it does not provide an objective algorithmfor classifyingindividual units. In R-factor analysis,it is customaryto classifyindividual units on the basis of factorscores. However,this often turns out to be a very subjectiveexercise when counties have high absolutescores on more than one factor. So how does one proceed? Since the concept of regional subculture connotes the idea of subculturalhomogeneity (based on sharedvalues and

3The indicatorof racialhomogeneity is included to measurethe relativeracial homogeneity versus heterogeneityof the population;while the five racialindicators are intendedto measurethe relativesize or numericaldominance of each racialgroup. RegionalSubcultures of the United States 895 ways of life), one might try to group counties on the basis of common cultural characteristics. One well-knowntechnique for accomplishingthis objectiveis cluster analysis. While there are a large numberof clusteringalgorithms, perhaps the most useful for largedata sets (from 100 to 100,000units) is a variationof a method developed by Anderberg(1973). This method, called FASTCLUS, employs a technique of classifyingor sorting units on the basis of their spatialor Euclideandistance to the nearestcluster centroid.4 First, a set of points called "clusterseeds" is selectedas a first approximationto the means of the clusters.Next, each unit is assignedto the nearest seed to form temporaryclusters. The means of these temporaryclusters are then used to replacethe cluster seeds, and the process is repeatediteratively. The iterationsstop when the conditionsset by the researcherare met and no further changes occur in the clusters. These conditions concern such mattersas the maximumnumber of clustersthat are allowed,the minimumnumber of units that are permitted to define a separatecluster, and the convergencecriteria that stop the iterativeprocess (SAS 1985, 377-79). Thus, FASTCLUS greatlyfacilitates the-execution of a disjointcluster analysis on the basis of Euclideandistances computed from one or more quantitativevari- ables. Unfortunately,the clustersare often difficultto interpret.And if there are a largenumber of groupingvariables, the clusteringprocess does not alwaysproduce parsimonioussolutions. To circumventthese difficulties,I first employedprincipal componentanalysis with varimaxrotation. Like other factor analytictechniques, principalcomponent analysis can transformthe variationin a largenumber of indi- catorsto a reducedset of factorswhose meaningscan be easilyinferred from the in- dicatorswith the largestloading coefficients. An addedadvantage of this technique is that the reduced set of factorsare unique and orthogonalto each other. Then I used the derived factors (actuallyfactor scores) as input variablesto the cluster analysis.The output from the clusteranalysis consists of county groupingsor clusters (generallyfewer in numberthan the input factors)that sharecommon cultural characteristics.To interpretand labelthese clusters(i.e., relatethem to the derived factors),I first createddummy variablesfor each cluster. Thus, a county was as- signed a valueof one if it belongedto a given cluster,zero otherwise.Then I corre- lated each of the cluster dummieswith the input factors.In this way, it is possible to matchor associateeach clusterwith its definingfactor(s).

RESULTS The results of first performinga principalcomponent analysis on the 45 cultural indicators discussed earlier and then rotating them using the vari- max method are presentedin table 1.5 The tabulateddata show that the selected

40ne majoradvantage of the FASTCLUS procedureis that its initializationprocess makes it sensi- tive to outliers. 5Inspectionof a scree plot of the eigenvaluesand their associatedcumulative explained variances suggestedsetting a minimumeigenvalue of 0.98. t en 00 eq e1 eq r- 00 C,, c" r- m - go

...... _ 0 O . O . 0 . O . O .O .

ul r- ul en UM"o en U: 0s 4 S 0 ~~o-1 om O- z mO

...... z ~ ~~~~~S O. O. O. O. O. O. O. O. g, O. O...... ~ ~~.

H~~~~~~ _o o0 oen 00 NO (- ,og"o Ln t. x>Y

g ~ ~~~~ 1omONs,!e+ F, o~~~~~~~~c p! ~~ ~ ~~L. _o (:oo g. 00e o olol UM 9t^< Ln _

...... N = o. o. oo. . o. o.. o. ]g s

. .'1 ...... 1- . . . . . 1- 1- 1-

X,q4*0

8^ i ?z & WO c B2 Y a|2r;:'~~~~~~~~~~~~~~4- = sW - eq Ln v r- tn 00 C.4 en (4 (.4 en eq CD c Rl

CD 00 e4 (1-4 LAI -4," CD ...... CD. . . . (= . = . (= . (= CD . . . .

r- Ln r- en en m Ln (,4 qt 00

en U) LA') 00 m tAn'd, c,4 m

tn C14C14 en r- en C4 ei CD 4=

Ln 8 Ln en S CNIen Ln en 00

en nD C4 ,0,0 Ln

00 00 00 00 Ul) en Ln en Ln 00 r- en U113 10 (= = - e"I en e"I CD CD = C4 - = (= 0, ...... Lne4

(11.4Ln (Z (= 'IC en (14 en Ln Ul) 00 Ln ...... CD - ......

t," m Ln e'j a- Ln en

U" 4, ul r- = 00 CK en U un en r- "o . . . . . = . --I. -q. -i --! . = . = . C14 . .

un C4 FiZZ-=7 Ln 1 en 1m*l "7nN

un so C7, z ull

'IR oq Ln

4w f4. u =1 0-4 898 Joel Lieske indicatorsreduce to 13 orthogonalfactors, which explain 71% of the cumulative variance.The cell entries consist of the computed factor loadings on each indicator. The enclosed boxes, which indicate all loadings more than .5 in absolute value, identify those indicatorsthat have the greatesteffect on each factor. The first rotatedfactor is labeledas a "Rurban"factor since it loads stronglyon those characteristicsthat Elazar(1984) has associatedwith the emergent"rurban- cybernetics"frontier, namely, rural-urbanhabitats, high levels of education,pro- fessional and managerialoccupations, working women, populationmobility, and youngerpopulations. This culturalstream is generallyfound in pastoralacademic settings and in the less populatedstates west of the Mississippi. The second factor, labeled "Ethnic," represents a diverse polyglot of ethnic groups that are concentratedin the largestand most urbanizedmetropolitan areas of the United States. Includedin this streamare groups who claim Italian,Slavic, Russian, Hungarian,and Greek ancestry as well as religious affiliationwith the reformand conservativetraditions of Judaism. For reasons which will become more evident, the third factor is labeled as a "Border"stream. This factor loads strongly on counties that are predominantly white; include significantconcentrations of residents whose ancestorscame from the "border"regions of Great Britain, such as the (Scotch)-Irish(Fischer 1989); are egalitarianin social structure;and favor traditionalfamily-oriented life-styles. The fourth factoris labeled "Agrarian"because of its high positive loadingson the relative sizes of the work force engaged in agricultureand the extractionof naturalresources. Since these economic sectors are generallycharacterized by a high degree of economicconcentration and a low division of labor,this factoralso loads highly on the indicatorof income inequality. The fifth factor, a "Nordic" stream, is typified by high densities of residents who claim Scandinavianancestry and who belong to the American Lutheran Church, the church body that united a number of small Lutheran synods with former ties to the old state churches of Scandinavia.Similarly, the "Germanic" stream is associated with residents of German ancestry, many of whom claim membershipin the Missouri-SynodLutheran Church and the United Churchof Christ. The formerchurch body used to be called the GermanLutheran Church; while the latter representsthe merger of the New England-basedCongregational Church with the largely midwestern-basedEvangelical and Reformed Church, a Calvinistbody formed by Germanand Dutch immigrantsto Pennsylvania(Cas- sels 1965, 114-15). In like manner,the sixth factormay be interpretedas a "Hispanic"stream since it loads stronglyon the relativesize of Spanish-speakingresidents, many of whom are affiliatedwith the CatholicChurch. On the other hand, the "Mormon"stream can be identifiedon the basis of a single tag, namely,membership in The Church of Jesus Christ of the Latter-Day Saints. Similarly,French ancestryserves as an identifying markerfor the "Anglo-French"factor, but this stream is also associated with Americanswho claim Britishancestry. RegionalSubcultures of the United States 899

TABLE2

GOODNESSOF FIT DIAGNOSTICS

ClusterSummary

Max. Seed Nearest Centroid Cluster Freq. RMS Std. Dev. Distance Cluster Distance

Germanic 340 1.02 23.1 Border 2.8 Ethnic 101 1.00 9.2 Border 4.8 Heartland 438 0.69 12.6 Border 2.4 Hispanic 128 0.75 6.7 Border 4.2 Nordic 133 0.76 6.5 Border 4.2 Mormon 52 0.70 6.0 Rurban 6.2 Border 855 0.53 4.8 Rurban 2.4 Blackbelt 479 0.68 10.9 Border 2.5 Rurban 464 0.89 16.0 Border 2.4 Anglo-French 145 1.11 21.0 Border 3.8

R2=.41 F9,3125=Z46.1 p <.001

Because of their historicalconfinement to reservations,Native Americansap- pear as a separate"Native" stream. Again for reasonsthat will become more evident, the "Heartland"factor is largelydefined by the followersof severalmainline Protestant denominations,most notably the American Baptist Church, whose forebearssettled the central heartlandof the United States and sided with the North during the Civil War. Although the twelfth factorloads on both the Asian and Portugueseindicators, it has been labeledas an "Asian"stream because of the numericaldominance of this group. Finally, the "Blackbelt"factor is distinguishablenot only by the significant numbers of residents who claim membershipin the Southern Presbyterianand Episcopalchurches, but also as the only one that has a significant,though small, positiveloading on the blackracial indicator. The results of using these rotatedfactors as input variablesto the clusteranalysis are presented in table 2.6 Table 3 presents the intercorrelationsbetween dummy measures of the derived clusters and the rotated factors. These inter- correlationswere used, along with a color-coded map of the continentalUnited States pinpointingthe geographicallocation of each county, to interpretand label the clusters(see figure2).7 Table 2 shows that some 3,000 U.S. counties can be classifiedinto 10 relatively homogeneousclusters or subcultures.Moreover, two goodness-of-fitdiagnostics,

6In this analysis,the maximumnumber of clusterswas set at 20; the minimumcluster size was set at 15; the convergenceparameter was set at .01; and the maximumnumber of iterationswas set at 100. 'I wish to thankLida Allen for setting up the computerprograms that made the constructionof this map possible. . o .o . .; ......

. L I L c

Lo%go " oC4 ooo

D 610 t Cho o_1 oO I ot~ _a ot I I RI 1- 1' 1- 1- 1- 1- 1 . 1- 1- 1- 1-

8 ew 1 ?i >~i ? ? i ?i

N O N o o zogo~ -00

.II...... HX ~ ~~o. 0.

U: N00 N~z0 x w0 : 00<0w RegionalSubcultures of the United States 901 the overall explained varianceand the pseudo F-statistic, show that the clusters are relativelywell-defined (SAS 1985, 386). Figure 2 shows that, with the exception of the "Rurban"subculture, the derived clusters of counties are fairly contiguous and regionally concentrated.Based on the highest intercorrelationsin table 3 (enclosed in boxes), I have labeled the clusters respectivelyas: (1) Ger- manic, (2) Ethnic, (3) Heartland,(4) Hispanic, (5) Nordic, (6) Mormon, (7) Bor- der, (8) Blackbelt,(9) Rurban,and (10) Anglo-French.8 The cluster with the largest number of counties, 855, is the "Border"subculture (see table 2). With the lowest maximumseed distance (4.8) and the smallest root mean squarestandard deviation (0.53), it is also the most culturallyuniform. In addition,it is also the closest, i.e., most similar,to eight of the nine other clusters in my typology. The only exception is the "Mormon"cluster which is closest to the "Rurban"cluster. Finally, as can be seen from figure 2, the "Border"cluster, along with the "Rurban"cluster, is one of the most regionallydispersed.

TheGenealogy and Morphology of the Subcultures One of the most insightfulhistorical studies on the genealogyof Americancul- ture is Fischer'sAlbion's Seed (1989). Adopting what is basicallya "germ"theory, Fischer contends that American culture was largely established by four great migrationsof English-speakingimmigrants from the British Isles. These immi- grants, he argues, carriedacross the Atlantic four different ways of life that be- came the basis of regionalcultures in the New World. The first wave was the exodus of Puritansfrom East Anglia to Massachusetts from 1629 to 1640. The second was the transplantationof a small Royalist elite and large numbersof indenturedservants from the South of Englandto Virginia between 1642 and 1675. The third was a migrationfrom the North Midlands of England and Wales to the Delaware River Valley from 1675 to 1725. And the fourth was a numericallylarge stream of English-speakingpeople from the bor- ders of England and northern Ireland to the Appalachianbackcountry between 1718 and 1775. Generallyspeaking, the first three waves correspondrespectively to Elazar's moralistic, traditionalistic,and individualistic "core" subcultures. However, the fourth representsa new and distinctive subculturethat Elazarat- tempts to representas an Appalachianblend of his moralisticand traditionalistic streams.9

81t might be noted that three of the rotated factors do not appear as distinctive clusters. The "Native"stream, which has a weakcorrelation with the "Rurban"cluster, sharesseveral cultural com- monalitieswith the "Rurban"stream including the representationof groups who tend to live in insti- tutionalsettings, spatiallydecentralized habitats, or west of the MississippiRiver. The "Agrarian"and "Asian"streams are too small to be separatelyidentifiable. 9The culturalsimilarities between Fischer'sfirst three wavesand Elazar's"core" cultures are clearly evident from Fischer's(1989, 785-86) summarydescriptions. He describesthe first,a Puritanwave, as "a very special culture with unique patternsof speech and architecture,distinctive ideas about mar- riageand the family,nucleated settlements, congregational churches, town meetings,and a traditionof 902 Joel Lieske

The extensions of Fischer's four "core" cultures are evident respectively in my Anglo-French, Blackbelt, Heartland,and Border subcultures(see figure 2). The Anglo-Frenchsubculture, of course, representsthe extension of the Puritan stream into greaterNew England and the Upper Midwest as well as the trans- planted French-Acadianstream in Louisianaand the influx of French-Canadians into the Northeast. To obtain more detailed informationon these four subcultures, as well as the six other subculturesidentified in this study, I computed the respective cluster means for each of the 45 cultural indicators.As expected, the tabulatedresults (not reportedhere for reasonsof space) reveala number of distinctive culturaltraits for each subculture. Thus, the Anglo-Frenchsubculture is a predominantlywhite, ethnicallydiverse regionthat representsa uniqueblend of the two largestethnic groups-British and French-which are roughly the same size. The high concentrationof non-Anglo Europeans, especially the French ethnics, also makes it one of the strongest Catholic regions in the country. On average,more than 35% of the population claim membershipin the Catholic Church, placing it second after the Hispanic subculturein catholicity.Although the New Englandregion is renown for its fine colleges,this subculture,overall, is not particularlydistinctive in the proportionsof college graduates,professionals, and managersin its workforce. The Blackbeltsubculture includes most of what Gastil (1975) has labeledas the Lowland and Upland South. The Lowland South is predominantlyan agricul- tural region of dispersed settlements that was originally based on a plantation economy and slavery. Known for its rich black soil and high concentrationsof blackresidents, this original"Blackbelt" stretches from the tip of the Del-Mar-Va peninsula south through the eastern shore of Maryland,tidewater Virginia, and the eastern Carolinas,then west through the Gulf region and up the Mississippi Delta past Memphis. Gastil's Upland South representsthe rolling Piedmont sections of the Old Confederacyand containsmost of the manufacturingcenters that were establishedat the fall lines of majorrivers. The blending of the two regions into a single consolidated Blackbelt subculture may be the result of cultural

ordered liberty."He argues that the second, a Cavalierwave, "was characterizedby scatteredsettle- ments, extreme hierarchiesof rank,strong oligarchies,Anglican churches, a highly developedsense of honor and an idea of hegemonicliberty." He contends that the third, a North Midlands wave, "was founded on a Christianidea of spiritualequality, a workethic of unusualintensity, a suspicionof social hierarchy,and an austeritywhich Max Weber called 'worldlyasceticism.'s' Moreover, he argues that the third wave "preservedmany elements of North Midland speech, architecture,dress, and food ways" and that "it deliberatelycreated a pluralisticsystem of reciprocalliberty in the DelawareVal- ley." Finally, while the emigrantsfrom his fourth, or "border"wave, were of differentethnic stocks, Fischer maintainsthat they "shareda common borderculture which was unique in its speech, architecture, family ways and child-rearingcustoms." He also argues that the materialculture of the "border" wave "was markedby extreme inequalitiesof condition, and its public life was dominatedby a distinctiveideal of naturalliberty." RegionalSubcultures of the United States 903 accretion,brought on by the increasingmechanization of southernagriculture, the forces of industrializationand urbanization,and the consequent movement of blacksfrom the rurallowland to the urbanupland. The most distinctivefeature of the Blackbelt,of course, is its racialcomposition (Key 1949). Blackresidents average more than 36% of the population.The largest Europeangroup consists of people who claim British ancestry,with an averageof 27% of the population.Consistent with its Bible Belt reputation,this subculture also contains an unusually high concentrationof Baptists. An averageof almost 25% of the populationare membersof the Southern Baptist Convention.As expected, there is also a high proportionof residents employed in manufacturing jobs (many of them at low wages), the highest concentrationof any region, including the more industrializedEthnic subculture (Luttbeg 1971; Pierce and Hagstrom1984). Because there is so much povertyin the region, especiallyamong blacks,this subculturealso has the highest levels of income inequality. The Heartlandsubculture, by comparison,largely coincides with what Elazar includes in his individualisticculture and_Gastillabels as the CentralMidwest. It extends from the upper reachesof the MohawkRiver Valleyand the northernAp- palachianswestward in a broad swath across Ohio, Indiana, and Illinois into southern Iowa, Nebraska,Kansas, and eastern Colorado.Its midwesternsections tend to follow the westwardmovement of the U.S. populationcentroid and en- compasssome of the richest and most productiveagricultural land in the nation's breadbasket.10Although a few isolated counties can be found further west, its western terminus appearsto be the beginning of the Rockies, i.e., the boundary between Garreau's(1981) Breadbasketand Empty Quarter.Culturally, this is the most raciallywhite and homogeneousof all the regions. Moreover, relativelyfew "white ethnics" are to be found, since on averagemore than 84% of the population claim either British, German,or (Scotch)-Irishancestry. Finally, the Border subculture represents the geographic extension of the largest, by far, of the four British migratorystreams. Beginning in the Appa- lachianMountains, Gastil's (1975) Mountain South, it extends in a southwesterly direction across the Border South states-Kentucky, Tennessee, Missouri, and Oklahoma-and includes substantialsections of Arkansasand Texas. This stream is not only the largestin areaof the four streamsbut also the most geographically mobile and restless, as evidenced by its extensions into Florida, Michigan, and more recentlythe far West.11These extensionsappear to reflecta continuinglove of the British borderpeople for the backcountryand out-of-doors(Fischer 1989). In terms of cultural characteristics,this region is very similar to the Heartland

'?This midcontinentregion is typified by states such as Ohio, whose motto is: "The heart of it all." Other remindersof this heartlandpsychology can be seen in the naming of business establishments such as the HeartlandMotel chain in centralIowa. "Its recent inroadsinto the far West can be seen in the growingnumber of counties with sizeable concentrationsof residentswho claim membershipin the SouthernBaptist Convention. 904 Joel Lieske region. The majordifferences that seem to distinguish the two are the somewhat greater representationof blacks in the Border region, which is reflected in a greaterheterogeneity of the population,and the pronouncedrole of the Southern BaptistConvention in the Borderway of life. In terms of historicalsettlement, the Hispanic subculturerepresents the oldest, and now perhaps the most territoriallyand culturally aggressive non-British stream. With a "core"in northernNew Mexico establishedby the Spanish con- quistadors,it extends into southern Colorado to the north; southeast along the Texas border to the Houston metropolitanarea; westward across the southern partsof New Mexico, Arizona,and California;and then finallyup into the Central Valley. Geographically,this subculturebears a strikingresemblance to Garreau's (1981) MexAmericaregion. Not surprisingly,an averageof more than 45% of the population in this region consider themselves Spanish-speaking.Notwithstand- ing, approximately90%, on average, consider themselves white rather than a memberof some other racialgroup such as mestizo. The still culturallydominant "Anglo"groups in this region are those of British, German,and (Scotch)-Irishan- cestry. Comparedto other subculturesthis is a religiouslydevout region, with an averageof 60% of the populationclaiming membership in just three dominantre- ligious bodies-Catholic, SouthernBaptist, and United Methodist. Closely borderingand often interspersedthroughout the Heartlandsubculture is the Germanic subculture. This stream originated in Philadelphiaduring the 1630s but extended its influence westward during subsequent waves emanating out of New York into Ohio, northern Indiana and Illinois, much of Wisconsin, into the upper Mississippi and Minnesota River Valleys, across northern Iowa, and then into the Dakotasand Nebraska.While one small streamout of New Or- leans headed for the hill country of Texas (Jordan 1966), most German immi- grants who embarkedfrom this port headed north to the St. Louis area and then up the Missouri River Valley. Like the settlement patternsof the Bordersubcul- ture, the Germanicsubculture exemplifies an unusuallove of country,and like the Heartlandsubculture it encompassessome of the richest and most fertile land in the United States today. Similar to the Heartlandregion, the Germanic subculture is overwhelmingly white and culturallyhomogeneous. It is also only one of two subcultureswhere the dominantethnic group, Germans,also constitute a clearmajority of the population. Although an average of more than one in every five residents claim membershipin the Catholic Church, they are still significantlyoutnumbered by adherentsof mainlineProtestant denominations, particularly in the more agricul- tural subregions. Economically,this region has perhaps the healthiest and most diversifiedeconomic base, with almost 60% of the population, on average,en- gagedin manufacturing,agriculture, or naturalresources. The Nordic subcultureconstitutes another border culture that is closely related to the Germanic by language, religion, social customs, and life-style. But there RegionalSubcultures of the United States 905 are enough cultural differences to distinguish it from the latter even though on averageGerman ethnics outnumberScandinavian ethnics by more than a two-to- one ratio. Also overwhelminglywhite and racially homogeneous, it is perhaps dominantin only two states-Minnesota and North Dakota-but its progressive influence also extends into the states of Wisconsin, South Dakota, and Montana (Nye 1951). Insulatedby the Germanicsubculture and situatedoff the majoreast- west migratoryroutes, the Nordic culture has enjoyed a high degree of cultural autonomythat is reinforcedby an economicbase of agriculture,natural resources, and high-tech industries. Another relativelyinsulated subculture is the Mormon region. But its isolation is more one of design than accident.Beset by religiouspersecution and economic resentment, the Mormons finally found, after an exodus that criss-crossed the Heartland subculture, a sanctuaryin the semi-arid region of the Great Basin, where it thrives today (O'Dea 1957; Mullen 1966; Lieske 1984). Since Mor- monism is a uniquely AmericanHeartland religion that grew out of the Second Great Awakeningof the 1820s, its converts for the most part were drawn from people of British ancestry.12Their descendantshelp make this subculturethe only other subculture(besides the Germanic)in which the dominantethnic group and religion still constitute a majorityof the population.Despite the lack of natural resourcesand industrialwealth, this subculturehas produceda very civilized way of life that is distinguishedby the stability of family life and an unusually well- educatedand highly skilledwork force (Lieske 1984). The final two subculturesare the largest,with some 67 million residentsapiece, and also the most inclusive, pluralistic,and spatiallydispersed. The Ethnic subculture is an ethnic polyglot of metropolitanareas located primarilyin the north- easternand midwesternindustrial corridors. With the exceptionof the Miami and Los Angeles areas, this subculture is being eclipsed in growth by metropolitan and "rurban"areas somewhatbetter equipped to compete in an emerging post- industrialand cyberneticseconomy (Elazar 1984). While both subcultureshave highly educatedand skilled work forces, the Rurbansubculture is less dependent on the manufacturingsector. The emergenceof this new Rurbansubculture can be seen not only in the nucleatedmetropolitan areas spawned by large state uni- versitiesbut also in the more prosperousand growingregions of the sunbelt.

TheSocial andPolitical Manifestations of theSubcultures The value of any theory, of course, lies in its ability to explain social and political behavior. A comprehensivetest to assess the utility of my new cultural typology is beyond the scope of this paper.However, a more restrictedtest can be obtainedby comparingthe predictivepower of my new measurewith Elazar'son a

2This religion originatedin Palmyra,NY, which lies just within the northeasternboundary of my Heartlandsubculture. 906 Joel Lieske limited number of social, political, and policy indicators."3Since Elazar'scultural map providesdesignations for only 224 substateareas, I extendedhis schemato all U.S. counties by painstakinglyinterpolating his classifications,county by county, in one of three ways. First, I attemptedto classify each county on the basis of its geographicalproximity to one of his designatedareas. If this was not possible, I tried to classify it on the basis of the westwardstreams he delineates.Finally, if both of these methods failed to providea clearclassification, I used the subcultural divisions providedby other culturalscholars who have built on his research,par- ticularlyGastil (1975) and Garreau(1981), to determineborderline cases.14 While Elazar'stheory was not intended to explain differencesin socialbehavior per se, many other subculturaltheories, including the one developed here, have been advancedwith this objective in mind (Banfield 1968; Gastil 1975; Garreau 1981; Thompson, Ellis, and Wildavsky 1990). In addition, there is a growing empiricalliterature that is beginning to document the effects of culture and subculture, including Elazar'stypology, on a number of social pathologiesand life qualitymeasures (Gastil 1975;Bayley 1976;Kincaid 1980;Lieske 1990). So it is in this vein that the resultsin table4 are presented. This table comparesthe utility of using my typology and Elazar'sto explain the variationin a number of social problemsthat concern Americanvoters and that have become recurrentissues in recent presidentialcampaigns. The analysisof-variance(ANOVA) results show that while the two typologies are essentially equivalentin predictingthe incidenceof povertyand teen-agepregnancy, my new typology is clearly superiorin explainingother social pathologies.These include the incidence of family breakdown,welfare dependency, and violent as well as propertycrime.15 Similarly, table 5 and table 6 present the ANOVA results for 10 selected indicators of political behavior. They include two measures of turnout in presidential elections, three measuresof party registration,and five measuresof the presidential vote. The results in table 5 show that my cultural typology and Elazar'sdo about equally well in explainingturnout in the 1980 and 1984 presidential elections and Republicanregistration in 1982. However, Elazar'sclassification does significantlybetter in accounting for Democratic and independent registration.

13Unfortunately,space constraintspreclude a discussionof the hypothesizedrelationships between my new measure,as well as Elazar's,and these indicators.Data for most of the indicatorsthat are ana- lyzed here were drawn from the U.S. Bureauof the Census' County StatisticsFile 3. I wish to thank BarbaraGroethe for her help and assistance.Data on party registrationwere generouslyprovided by the RepublicanNational Committee.Presidential voting data were obtainedfrom the Interuniversity Consortiumfor Politicaland Social Research. 141wish to thankDaniel Prugarfor assistingme in this task. -'5As might be expected, the two most urbanizedsubcultures, the Ethnic and Rurban,are also most prone to the attendantsocial problemsof urbanization,including family breakdown, violent crime, and propertycrime. C,) (1 ~~~~~~~~~~~~~~~~~~~~~~~R 44~~~~~~~~~~~~

ri~~~~li to . ...I.I TABLE 4

VARIATION IN SOCIAL PROBLEMS BY REGIONAL SUBCULTURE

Social Problems Regional Broken Poor Teen-age AFDC Violent Property Subculture Homesa Familiesb Mothersc Recipientsd Crimee Crime' Germanic 4.5 10.7 12.3 2.1 91.8 2,135.7 Ethnic 8.9 7.7 12.5 4.7 519.0 4,816.4 Heartland 6.3 9.6 16.2 2.6 125.4 2,451.8 Hispanic 7.0 17.9 20.1 3.5 334.2 3,163.8 Nordic 3.9 11.7 11.8 2.4 55.3 2,221.0 Mormon 5.5 10.5 12.8 1.7 220.2 3,337.4 Border 7.0 14.0 20.7 3.0 190.3 2,314.7 Blackbelt 8.7 18.0 22.8 6.0 347.9 2,744.6 Rurban 9.0 9.2 14.3 3.0 400.0 5,184.9 Anglo-French 7.5 10.2 14.7 4.5 240.0 3,738.6 N 3,135 3,135 3,116 3,135 3,037 3,037 R2 40.6% 26.5% 40.9% 20.8% 20.3% 23.7% R2(Elazar) 14.1% 27.3% 42.8% 13.3% 4.6% 4.9% aPercentageof all women 15 yearsand older who are separatedor divorced. bPercentageof all familiesbelow the povertylevel. cPercentageof all births to mothersunder age 20. dPercentageof populationreceiving AFDC assistance. eNumberof homicides,forcible rapes, robberies, and aggravatedassaults per 10,000 population. fNumberof grandlarcenies, burglaries, and auto thefts per 10,000population.

TABLE 5

VARIATION IN TURNOUT AND PARTY REGISTRATION BY REGIONAL SUBCULTURE

Regional Turnouta PartyRegistrationb Subculture 1980 1984 1982 Dem. 1982 Ind. 1982 Rep. Germanic 65.6 64.2 34.3 18.1 47.6 Ethnic 54.8 58.5 45.7 25.4 28.9 Heartland 59.7 59.8 37.4 20.3 42.3 Hispanic 52.4 55.9 64.9 8.6 26.5 Nordic 72.2 69.4 40.3 11.8 47.9 Mormon 73.4 74.3 46.2 11.3 42.5 Border 55.5 57,9 67.4 5.5 27.1 Blackbelt 49.1 52.6 81.8 3.4 14.8 Rurban 51.9 56.3 48.2 16.8 35.0 Anglo-French 62.6 64.6 50.0 22.4 27.6 N 3,135 3,135 1,395 1,395 1,395 R2 28.6% 17.2% 38.9% 20.5% 26.8% R2 29.4% 19.1% 55.7% 33.0% 25.4% aPercentageof all adults 18 yearsand older who voted. bPercentageof all registeredvoters. 908 Joel Lieske

TABLE 6

VARIATION IN PRESIDENTIAL VOTE BY REGIONAL SUBCULTURE

Regional PresidentialVotea Subculture 1980 Dem. 1980 Ind. 1980 Rep. 1984 Dem. 1984 Rep. Germanic 30.6 5.5 62.5 32.0 67.2 Ethnic 42.0 7.6 48.8 44.3 55.2 Heartland 34.9 4.9 58.9 33.9 65.4 Hispanic 43.0 2.5 53.5 39.8 59.6 Nordic 36.3 6.4 55.7 41.3 57.8 Mormon 20.6 3.0 74.7 21.7 77.5 Border 44.8 2.3 51.8 36.6 62.8 Blackbelt 53.7 1.6 43.6 42.7 56.7 Rurban 35.4 7.1 55.4 35.1 64.0 Anglo-French 41.1 8.2 49.0 38.9 60.4 N 3,111 3,111 3,111 3,113 3,113 R2 36.6% 44.8% 28.7% 15.1% 28.9% R2 (Elazar) 37.3% 50.7% 20.8% 6.6% 6.2%

aPercentageof the total presidentialvote.

As the results of table 6 indicate,both typologiesdo quite well in explainingthe subculturalvariation in the 1980 presidentialvote. The two are essentiallyequiva- lent in accounting for the Cartervote. But Elazar'sis better able to explain the vote for independentcandidate John Anderson;while mine is better able to predict the Reagan vote. However, my typology fares uniformly much better in accountingfor the Mondaleand Reaganvotes in 1984. Table 7 presentsthe ANOVA results of using the two classificationsto explain the variationin seven selected measuresof public policy at the local level. These indicatorsinclude local governmentrevenues, the local tax burden, educational expenditures, the educationaltax burden, welfare expenditures,the welfare tax burden, and local AFDC payments circa 1980. The results show that Elazar's classificationdoes significantlybetter in accounting for AFDC payments, local governmentrevenues, welfareexpenditures, and the welfaretax burden. The two are about equal in explainingthe overalltax burdenand educationalexpenditures. But mine is significantlybetter in accountingfor local differencesin the educational tax burden. In sum, my measureperforms much better than Elazar'sin predictinga number of social pathologiesthat appearto be ethnoculturallyrelated. This is perhapsto be expected since my typology is much more precise than Elazar'sin measuring ethnoreligiousand social differences.On the other hand, the two typologies are about equal in predictivepower for politicalbehavior that is habitualin character (e.g., voting turnout)or reflectiveof ethnoculturaldivisions in the Americanelec- torate(e.g., candidatepreferences in the 1980 and 1984 elections).Finally, Elazar's RegionalSubcultures of the United States 909

TABLE 7

VARIATION IN PUBLIC POLICY BY REGIONAL SUBCULTURE

Public Policy Regional Local Tax Education Welfare Subculture Rev., Burdenb Expend.c Burdend Expend.' Burdenf AFDC9 Germanic 444 5.5 526 6.3 31 .36 253 Ethnic 507 5.0 522 4.9 79 .76 279 Heartland 411 4.7 505 6.0 30 .37 236 Hispanic 394 6.5 670 9.1 31 .39 141 Nordic 521 4.8 585 7.3 77 .97 271 Mormon 317 4.5 636 8.2 4 .05 271 Border 276 3.2 422 5.7 9 .12 169 Blackbelt 289 3.0 415 6.2 13 .17 137 Rurban 343 4.5 561 5.8 29 .30 233 Anglo-French 411 5.5 501 6.1 41 .51 273 N 3,130 3,098 3,130 3,098 3,130 3,098 3,135 R2 23.9% 7.1% 6.2% 9.1% 3.3% 12.4% 33.5% R2 (Elazar) 31.7% 9.8% 6.0% 1.0% 8.8% 17.4% 57.3%/o

"Percapita local governmentrevenues in dollars. bPercapita local governmenttaxes as a percentageof per capitapersonal income. cPercapita local governmentexpenditures on educationin dollars. dPercapita local governmentexpenditures as a percentageof per capitapersonal income. ePercapita local governmentexpenditures on welfarein dollars. fPer capitalocal governmentexpenditures on welfareas a percentageof per capitapersonal income. gAveragefamily monthly payment in dollars. typologyappears to be superiorin explainingcertain types of politicalbehavior that result from state-levelinfluences (e.g., policies that are primarilystate-funded) or that are patternedalong state-party lines (e.g., voting registration). Overall,my measureappears to do about as well as Elazar's.At the same time, it is a statisticallyrigorous measure that avoids the problemof politicalcircularity inherentin his measure.In addition,it remainsfaithful to a key assumptionof his theory, namely, that subculturaldifferences are transmittedprimarily by ethnic and religious groups. Last but not least, my new measureoffers to researchersa major refinement to his "core" British subcultures (four not three) that first settled the United States, as well as a contemporaryaccounting of other ethnocul- tural streams-French, Germanic,Nordic, Hispanic, Mormon, Ethnic, and Rur- ban-that have left their subculturalimprint.

DISCUSSION AND CONCLUSION The primaryobjective of this study was to develop the case, theory, and statistical methodology for a new measure of American regional subcultures. A 910 Joel Lieske secondaryobjective was to assess its utility in explainingdifferences in social and politicalbehavior. Overall, the results supporta numberof conclusions. First, the combined use of principalcomponent and cluster analysismay offer cultural scholars a new and promising methodology for deriving rigorous subculturalmeasures. In using this methodology,however, researchers must exercise care in how they conceptualizeculture and the indicatorsthey select for analysis. In short, the techniqueonly succeedsto the extent it is informedby theory. Second, the results suggest that the entire populationof U.S. counties can be partitionedinto 10 distinctive subculturesthat are relatively homogeneous and contiguous. In addition, these subculturesappear to make theoretical sense, as evidenced by their close relationshipto those identifiedby other culturalscholars such as Elazar(1984), Gastil (1975), Garreau(1981), and Fischer (1989). Third, it is evident that my new typology compares very favorably with Elazar'sin accounting for differences in a variety of social and political indicators. Although my new measureis no more parsimoniousthan Elazar'stypology (10 versus 9 categories),it is more precise and takes into account recent demo- graphic changes, especially the United States' greater cultural diversity. And while even better results might be obtained through the incorporationof some "effect" and "manifestation"indicators, their inclusion would defeat my effort to avoid the problem of circularity.For example, several measuresof past political behaviorcould be included in the factor-clusteranalysis that would enhance the predictive power of my typology. These might include, for instance, turnout in the 1976 presidentialelection, party registrationlevels in 1980, and local government tax and spending figures circa 1970. Their inclusion might increase significantlythe explanatorypower of my typology, but only at the cost of ob- scuring the necessarytheoretical distinction between cultural"causes" and political "effects." Fourth, it would seem desirableto subject my new measureto further testing, especially at the state level. This could be done, for instance, by aggregating across counties and computing the respective proportionsof the total state-wide populationthat are under the influence of each subculture.In effect the political culture of each state would be representedby a 10-dimensionalvector variable. Presumably,this variablewould measurethe relativesizes of the contendingsub- cultural constituencies within a state that vie for control of legislative districts and the governorship.And like Sharkansky's(1969) cultural index it could be used to predict differences among the states in political institutions and public policies. Finally, my new measuremay also prove useful in contextualresearch that con- ceptualizessubcultures as arenasof conflict (Schattschneider1960). For example, it can be demonstratedthat the factorswhich structurepresidential voting behavior differ in their effects from one subcultureor culturalregion to another(Lieske 1989, 1991). To conclude, this study should help lay the theoreticaland empirical RegionalSubcultures of the United States 911 bases for future studies that attempt to understandthe effects of regionalsubcultures on Americansocial and politicalbehavior.

Manuscriptsubmitted 18 October1991 Final manuscriptreceived 18 March1993

REFERENCES

Anderberg,M. R. 1973. ClusterAnalysisforApplications.New York:Academic Press. Banfield,Edward C. 1960. The UnheavenlyCity. Boston:Little, Brown. Bayley, David H. 1976. "Learning about Crime-the Japanese Experience." Public Interest (Summer):55-68. Campbell, Angus, Philip Converse, Warren Miller, and Donald Stokes. 1960. The AmericanVoter. New York:Wiley. Cassels,Louis. 1965. What'sthe Difference?AComparison of the FaithsMen Live By. GardenCity, NY: Doubleday. Collins, Ronald K. L., Peter J. Galie, and John Kiicaid. 1986. "State High Courts, State Constitu- tions, and IndividualRights LitigationSince 1980:A JudicialSurvey." Publius 16:141-61. Dran, Ellen M., Robert B. Albritton, and Mikel Wyckoff. 1991. "Surrogateversus Direct Mea- sures of Political Culture: Explaining Participation and Policy Attitudes in Illinois." Publius 21:15-30. Dreitzel, Hans Peter. 1977. "On the Political Meaning of Culture."In Beyondthe Crisis,ed. Norman Birnbaum.London: Oxford University Press. Elazar,DanielJ. 1970. Citiesof the Prairie.New York:Basic. Elazar,DanielJ. 1984.American Federalism. 3d ed. New York:Harper & Row. Fischer, David Hackett. 1989.Albion's Seed. Oxford:Oxford University Press. Fitzpatrick,Jody L., and Rodney E. Hero. 1988. "PoliticalCulture and PoliticalCharacteristics of the AmericanStates: A Considerationof Some Old and New Questions." WesternPolitical Quarterly 41: 145-53. Garreau,Joel. 1981. TheNine Nationsof NorthAmerica. Boston: Houghton Mifflin. Gastil, Raymond D. 1975. CulturalRegions of the UnitedStates. Seattle: University of Washington Press. GlenmaryResearch Center. 1982. Churchesand ChurchMembership in the UnitedStates. Atlanta:GRC. Herzik, Eric B. 1985. "The Legal-Formal Structuring of State Politics: A Cultural Explanation." WesternPoliticalQuarterly 38:413-23. Holbrook,Stewart H. 1950. The YankeeExodus. New York:Macmillan. Johnson, Charles A. 1976. "PoliticalCulture in American States: Elazar'sFormulation Examined." American ournalofPoliticalScience18:331-45. Jordan,Terry G. 1966. GermanSeed in TexasSoil: ImmigrantFarmers in Nineteenth-CenturyTexas. Austin, TX: Universityof Texas Press. Joslyn, RichardA. 1980. "Manifestationsof Elazar'sPolitical Subcultures:State Public Opinion and the Contentof PoliticalCampaign Advertising." Publius 10:37-58. Key, V. O.,Jr. 1949. SouthernPolitics in State andNation. New York:Knopf. Kincaid,John. 1980. "PoliticalCulture and the Qualityof Life." Publius10:89-110. Kincaid,John. 1982. "Introduction."In Political Culture,Public Policy, and the AmericanStates, ed. John Kincaid.Philadelphia: ISHI Press. I Kincaid,John. 1990. "From Cooperativeto Coercive Federalism."The Annals of the AmericanAcad- emyof Politicaland Social Science509: 139-52. 912 Joel Lieske

Kincaid,John, and Joel Lieske. 1991. "PoliticalSubcultures of the AmericanStates: State of the Art and Agendafor Research."Paper delivered at the annualmeeting of the AmericanPolitical Science Association. Kleppner, Paul. 1970. The Crossof Culture:A Social Analysisof MidwesternPolitics, 1850-1900. New York:Free Press. Lasswell, Harold D., and AbrahamKaplan. 1950. Power and Society: A Frameworkfor Political In- quiry.New Haven:Yale University Press. Lazarsfeld,Paul, BernardBerelson, and Helen Gaudet. 1944. ThePeopk's Choice.New York:Colum- bia UniversityPress. Leege, David C., Joel A. Lieske, and Kenneth D. Wald. 1991. "TowardCultural Theories of Ameri- can PoliticalBehavior: Religion, Ethnicityand Race, and Class Outlook."In PoliticalScience: Look- ing to the Future,ed. WilliamCrotty. Chicago: University of NorthwesternPress. Lieske, Joel A. 1984. "The Salvation of American Cities." In RebuildingAmerica's Cities: Roads to Recovery,eds. Paul R. Porter and David C. Sweet. New Brunswick,NJ: Rutgers University Press. Lieske, Joel A. 1988. "The Culturl Origins of Political Partisanship."Paper delivered at the annual meeting of the WesternPolitical Science Association. Lieske, Joel A. 1989. "SubculturalDifferences in the U.S. PresidentialVote: The Evidence from Re- cent Aggregateand IndividualLevel Data." Paperdelivered at the annualmeeting of the American PoliticalScience Association,Atlanta. Lieske, Joel A. 1990. "The Correlatesof Life Quality in U.S. MetropolitanAreas." Publius 20:43-54. Lieske,Joel A. 1991. "CulturalIssues and Imagesin the 1988 PresidentialCampaign: Why the Demo- cratsLost-Again!" PS: PoliticalScience and Politics 24:180-87. Lowery, David, and Lee Sigelman. 1982. "Political Culture and State Public Policy: The Missing Link." WesternPolitical Quarterly 35:376-84. Luttbeg, Norman R. 1971. "Classifyingthe AmericanStates: An EmpiricalAttempt to Identify Varia- tions." MidwestjournalofPolitical Science15:703-21. Monroe, Alan D. 1977. "OperationalizingPolitical Culture: The Illinois Case."Publius 7:107-20. Morgan,David R., and RobertE. England. 1987. "Classifyingthe AmericanStates: An Update." So- cial ScienceQuarterly 68:405-17. Morgan, David R., and Sheilah S. Watson. 1991. "PoliticalCulture, Political System Characteristics, and Public Policies amongthe AmericanStates." Publius 21:31-48. Mullen, Robert. 1966. TheLatter-Day Saints. New York:Doubleday. Nardulli, Peter F. 1990. "PoliticalSubcultures in the AmericanStates: An EmpiricalExamination of Elazar'sFormulation." American Politics Quarterly 18:287-315. Nye, Russel B. 1951. MidwesternProgressive Politics: A HistoricalStudy of Its Originsand Development, 1870-1950. Lansing,MI: MichiganState UniversityPress. Peirce, Neal R., and Jerry Hagstrom. 1984. The Book of America:Inside the Fifty States. New York: W. W. Norton. Press, Charles,and Kenneth VerBurg. 1983. State and CommunityGovernments in the FederalSystem. 2d ed. New York:Wiley. O'Dea, Thomas F. 1957. TheMormons. Chicago: University of ChicagoPress. SAS InstituteInc. 1985. SAS User'sGuide: Statistics. 5th ed. Cary,NC: SAS. Savage,Robert L. 1973. "Patternsof MultilinearEvolution in AmericanStates." Publius 3:75-108. Schattschneider,E. E. 1960. TheSemi-Sovereign People. New York:Holt, Rinehart,and Winston. Sharkansky,Ira. 1969. "The Utility of Elazar'sPolitical Culture." Polity 2:66-83. Thompson, Michael, Richard Ellis, and Aaron Wildavsky. 1990. Cultural Theory. Boulder, CO: Westview. Wald, Kenneth D. 1987. Religionand Politics in the UnitedStates. New York:St. Martin'sPress. Weber, Ronald E., and William R. Shaffer. 1972. "Public Opinion and American State Policy- making."Midwest journal of PoliticalScience 16:683-99. RegionalSubcultures of the United States 913

Wildavsky,Aaron. 1987. "ChoosingPreferences by ConstructingInstitutions: A CulturalTheory of PreferenceFormation." American Political Science Review 81: 3-21. Zelinsky, Wilbur. 1973. The Cultural Geographyof the United States. Englewood Cliffs, NJ: Prentice-Hall.

Joel Lieske is associateprofessor of political science, Cleveland State Univer- sity, Cleveland,OH 44115.