power, influence and social circles a new methodology for studying opinion makers

16
8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 1/16 Power, Influence and Social Circles: A New Methodology for Studying Opinion Makers Author(s): Charles Kadushin Source: American Sociological Review, Vol. 33, No. 5 (Oct., 1968), pp. 685-699 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2092880 Accessed: 22/05/2010 07:37 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=asa . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].  American Sociological Association  is collaborating with JSTOR to digitize, preserve and extend access to  American Sociological Review. http://www.jstor.org

Upload: lessya

Post on 03-Jun-2018

242 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 1/16

Power, Influence and Social Circles: A New Methodology for Studying Opinion MakersAuthor(s): Charles KadushinSource: American Sociological Review, Vol. 33, No. 5 (Oct., 1968), pp. 685-699Published by: American Sociological AssociationStable URL: http://www.jstor.org/stable/2092880

Accessed: 22/05/2010 07:37

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=asa.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of 

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

 American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to

 American Sociological Review.

http://www.jstor.org

Page 2: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 2/16

AMERICAN SOCIOLOGICAL REVIEW

October,1968Volume33, No. 5

POWER,INFLUENCEAND SOCIALCIRCLES:A NEW METHODOLOGYFOR STUDYING

OPINION MAKERS*

CHARLES KADUSHIN

ColumbiaUniversity

There is a poor mesh betweenfacts and theoriesin studies of elites and power. Presentdefinitionsof power and influenceare essentially ncomparablebecausepower is a disposi-tion conceptrequiringa seriesof reduction entences or more exact specification.The directmeasurementof power thereforerequiresmultiplemeasurement, nd power may, in givencircumstances, e multi-dimensional.Elites are not carefullydefined,except, frequently,intermsof power.Much of the argument n the field of powerand elite studies s over whetheror not the powerfuland the elite are informally nterconnected.Lack of work on the con-cept of socialcirclehas hampered he developmentof both specificpropositionsand adequatemethods to study informalconnections n largesocial systems. There are differenttypes ofcircles,and they bear differentrelationsto formalsocialstructures.Most propositionsaboutpower structurecan be located in these terms.Methodsfor measuringsocial circlesrequireopen-endedsociometricchains,such as snowball amples,although there are also usefulpseudo-interactionmethods.Both reputationaland decisionalmethods for studying powerelites are variationsof snowballtechniques.Whenpositionaltechniquesare used to developthe startingpointsfor snowballstudiesof powerinfluentials,and the conceptof social circleis invoked to develop variousdimensionsupon which to snowball,and multiplemeasure-mentsare used to define the power output of individualsso located,many of the issues inanalyzingpower elites are resolved.An examplefrom the designof a comparative tudy ofnationalopinionmakerss offered.

IN this paperwe shall reviewsomeof theconcepts and methods for obtainingfacts in the field of studies of elites--

a field in which theories with little factualsupport . . . have been replaced by massesof facts with little theoretical structure(Rustow, 1966:716). We shall suggest thatthere is basically nothing wrong with theconcept and methods; all the parts for build-ing good elite studies are, in fact, lyingabout. They simply have not been properlyfitted together in a coherent structure, for

one important linkage has been missing.This missing link is a concept which hasbeen fruitfully but unsystematicallyused fora number of years-the concept of social

* This paper has benefitedfrom the commentsof Terry N. Clark,Lewis Edinger,GeorgeFischerandJuanLinz. This is BASR publicationA-551.

circle (Kadushin, 1966). This concepthelps to resolvesome of the conceptualprob-lems in elite studies and also points the wayto gathering the right facts.

CONCEPTS AND METHODS

The major concepts in studies of elitescan be divided into those which pertain towhat elites do and those which pertain towho they are and what their place is in thesocial structure.The term elite itself refers

to who they are, as does the term rulingclass or governing class. But elitesmanipulate power, influence and opin-ion. Sometimes, therefore, elites are saidto have power or influence so that theset of persons comprising the elite deter-mines the characteristics of power and itsuse.

685

Page 3: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 3/16

686 AMERICANSOCIOLOGICALEVIEW

Power and Influence. The concept ofpower seems to be responsible for many ofthe difficultiesin the study of elites, in partbecause the concept lies at the very rootof political theory and so each theorist tends

to develop his own definition of power. Fora comparison of some recent definitions ofpower see Riker (1964:341-349). Some-what different approaches are taken by so-ciologists. For general reviews, see Rose(1967:43-53), Parsons (1966:240-265),Danzger (1964:707-717), Kornhauser(1966:210-218), Spinrad (1965:335-356)and Clark (1968).

The basic reason for the multiplicity ofdefinitions, however, is that power, as it is

used by social scientists, is a dispositioncon-cept. According to Hempel (1952:24),

The property erm magnetic s an exampleof a disposition erm: it designates,not a di-rectly observable haracteristic, ut ratheradisposition,on the part of somephysicalob-jects, to display specificreactions(such asattractingsmall iron objects) under certainspecifiablecircumstances such as the pres-ence of small iron objects in the vicinity).The vocabularyof empirical cienceabounds

in dispositionerms....

The property of being magnetic is clearenough if there happens to be a small ironobject in the vicinity of, say, a bar magnet.But suppose the magnet is at the bottomof a lake, how do we now definemagnetism?Similarly, how can one tell if a person, role,or group is powerful? One can tell only ifspecified test conditions are met. For ex-ample: If A and B are in a committee-

voting situation, then A is more powerfulthan B if, and only if, A more frequentlycasts the deciding vote. ' This sort of par-tial, conditional definition is called a re-ductionsentence (Hempel, 1952:25-29). Ingeneral, the term power cannot be elimi-nated in favor of the terms in the definition.Rather, the indeterminacyin the meaningof a termintroducedby a reductionsentencemay be decreasedby laying down additional

reduction sentences for it which refer to dif-

ferent test conditions (Hempel, 1952:29).The many definitions of power and the factthat one cannot directly infer any one ofthem [five leading formal definitions] fromany other one both directly follow from the

dispositional nature of the concept power(Riker, 1964:348).

Reduction sentences also imply an em-pirical generalization (Hempel, 1952:28).To continue with the example of power ina committee-voting situation, the partici-pant's chance to be the last added memberof a minimal coalition . . . is highly attrac-tive presumably because the last added win-ner can control the form of distribution ofthe winnings (Riker, 1964:341-342). Ob-

viously, this presumptioncan be empiricallytested. If it is never true, then the reductionstatement which defines power as being thelast added member of a coalition is clearlynot reasonable; if it is true only under cer-tain specifiable conditions, then this reduc-tion sentence is even more limited than isat firstapparent.2

The various definitionsof power now usedby social scientists have the characteristics

of reduction sentences-they are incompletesentences.No less than six elementsof powerare invoked by the current literature. (Fora somewhat different set of dimensions seeClark, (1967b). The first three have to dowith the act of power itself-who does whatto whom. The second three specify the socialsetting of the act-when, where, and underwhat conditions it occurs. The social settingof power may be considered as equivalentto the motivation for exercising power.

A. The Act of Power

1. Who is said to have power-individualpersons,roles, or statuses, or collectivities.3

2. What is beingmanipulated-the acts ofparticularpeople (Dahl, as explicated inRiker,1964:342-343); the utilities of people(Karlsson, n Riker, 1964:343); or the gen-eral course of events (Shapely-Shubik,nRiker,1964:344).Although ny eventor any

1This is a paraphrase of the Shapely-Shubikdefinitionof power, cited by Riker (1964:341). Tosimplify our exposition, he statement s confined odyadic relations. Unlike some other definitions ofpower,however,the Shapely-Shubik efinitionwasoriginallystated so as to apply to an n-adic situa-tion.

2The empirical implications of many reductionsentences defining power are not as clear as in thisexample. But this simply means that those defini-tions are less elegant.

3 In Marsh's definition (stated in Riker, 1964:342)the unit is a role; in Parson's definition (1966:242)the unit could be a collectivity. In most otherdefinitions, the unit is assumed to be an individualperson.

Page 4: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 4/16

POWER. INFLUENCE AND SOCIALCIRCLES 687

utilitiescan be expressedas the consequenceof the acts of individualpeople, there doesseem to be a practicaldifferencef one thinksaboutmanipulating eople, their rewards,orthe courseof events.If focus s exclusively nmanipulating eople,then there is a greatertendency to view powerin any system as afixed sum, in which if A has a lot of power,B must have less. If rewardsor events aremanipulated,however,it is possible to in-creaseor decrease he total amountof powerwithina systemso that a change n A'spowermay ormaynot signala change n B'srelativepower.

3. To whomdoespowerhaveconsequences-the self, other people, or other roles, orother collectivities.4Again,systems are com-posed of individualunits-either people orcollectivities-but there are both practical

and theoreticaldifferences f networks andsystems rather than individualsare said tobe affectedby power.

B. The Social Setting

4. Whenor whetherpower s an abilityorpotentialability to have an effect or repre-sentsanactualeffect.5

5. Where-the sectors,arenas and institu-tional areas for which particularunits canhave or do have certainconsequences.Thispointis systematically evelopedby Danzger,

1964:715-717). Definitions of poweras theexerciseof legitimateauthorityare especiallyclearas to the boundaries f power(Parsons,1966).

6. Underwhatconditions-theinstitutional,organizational nd moral constraintson theuse of power_(Kornhauser,966).5

Given the present state of theory and data

about power, the concept is specifiedwhen a

set of empirically true reduction statements

is developed which has something to sayabout each one of these aspects of powerand

thus specifies both necessary and sufficientconditions for its existence. Whether or notthese reduction sentences or indicators aresuch that they yield a unidimensional con-

cept of power, or group themselves into two

or more dimensions,is a matter which must

be settled empirically, and may depend onthe nature of the social system in whichpower is exercised. For example, in many

systems it is empirically true that potentialpower and actual effects cluster together. A

unit which, in the past, has affected eventsmay be better able to do so in the future,unless it has used up its credits. The im-

plication for the study of power is now ob-vious. A good study must have multiple in-dicators of power, as well as some way totest whetheror not the indicators fit together(Clark, 1967b, 1968). The six elements ofpower just identified are not themselves re-duction statements but, rather, represent a

formal classification of reduction statementsthat have been proposed in the literature.

Much of what has been said about powercan also be said about influence. (For ageneral review, see Cartwright, 1965). Thisconcept is also a disposition concept, andthe six formal elements of power alsoclassify reduction statements dealing withinfluence. If influence is said to be a gen-eralizedmechanismof persuasion (Parsons,1966:251), then in practice it is difficult todistinguishbetween influence and power, un-less power is taken to refer to legitimate

authority only. For example, Rose (1967:300) states, We recognize a theoretical dif-ference between 'influence' and 'power,' butfeel obliged to blur the distinction for pur-poses of our study of images and perceptionsof these social phenomena. The blending ofthe concepts is especially noticeable if powerrelations are defined as a probability that B

does A's bidding. For the notion of influence

4Although all definitions n the field of powerimply that the unit said to have power does ulti-mately derive some benefit from the exercise ofpower, the immediategoal of the use of power

differs. Riker (1964:344) notes that the Shapely-Shubikdefinitiondeals with ego'sability to increasehis own utility, whereasKarlsson'sdefinitiondealswith ego's ability to decreasealter's utilities. Theimplications of other definitions are not quite asclear. On the other hand, Parsons (1966:244) ismoreconcernedwith the consequences f powerforentiresocialsystems.

5This difference ies at the cruxof the argumentbetween reputational and decisional modes ofmeasuringpower.Spinrad(1965:223-228)discussesthis mainly in terms of the motivationof powerfulindividualsor groupsto intervene n any particular

decision. See also Danzger (1964:712-714).6 When translated nto our terms,this is one of

the major points made by Peter Bachrach andMortonS. Baratz (1962:947-952). Their notion of

politicalbias is also salientto our notion of themotivation to exercisepower, though it refers sogenerallyto the social system in which power isexercised hat power per se becomes ost in the gen-eral study of social systems. Amitai Etzioni

(1961:5) suggests that the most important thingabout power is the means employed to make thesubjects comply. But this is a consideration whichdeserves separate study. We also have not consid-ered why units have power. Lasswell and Kaplan(1963:87) have a multidimensional discussion ofthis issue.

Page 5: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 5/16

688 AMERICANSOCIOLOGICALEVIEW

implies hatA suggests omethingo B, whothen may or may not act uponit-an ideathat is virtuallysynonymouswith the prob-ability thatB will take action.Nevertheless,it is poor strategy n the study of elites to

abandon he conceptof influence.From thestandpoint f the theoryof action (Parsons,et al., 1954), it is extremely mportant hata person feels anotherhas influenceoverhim, that in a given actionhe takesintoaccount anotherperson'sopinionsor feel-ings,even if the finalaction s only partiallydeterminedby another person'swishes oropinions. Influenceconcernsthe extent towhicha personprovides ome of the frame-work withinwhich outcomesoccur, and it

involvescommunicationboutvalues. ThusLasswelland Kaplan (1963:58, 75) definedpoweras participationn makingdecisionsandinfluence s havingvaluepositionpoten-tial. Sincemanygroups n societyhave valuepotential, a study of influence tends tobroaden the set of people who constitutethe elite-a considerationwhich leads ussquarely ntothe topic of elites,powerelites,opinion eadersand rulingclasses.

Elites and Influentials. Unlike the conceptof power,neither he conceptnor the defini-tions of elite have been given especiallyformal treatmentby contemporary ocialscientists (Lasswell, 1964:4; Bottomore,1964:Chapter1). The set of indicatorsofelite are important ather han the conceptitself. Despiteor becauseof this focus,thereis muchdisagreementbout the appropriateset of indicators.As onecriticof elite theorycomments, We may state what an elite is,but it is muchmoredifficulto find out whoare the elites (Sereno,1962:79). [For areview of this conflictsee Walton (1966a:684-689; 1966b:430-438)andRose (1967:255-297). SeeEdinger 1967) for a generalreviewof the elite literature ndan excellentannotated ibliography.]

There appear to be three types of defini-

tion of elite : statistical, functional, andstructural.7 Frequently the types are com-

bined, and often the definitions ontainexplicitor implicitpropositions.Definitionsof the statistical type are exemplifiedby

Lasswell's1934 definition: The few whoget the most of any value are the elite

(Lasswell,1950:3). Rosenaun a functionaldefinitionpecificallyschews he conceptofelite, preferringa more specificnotion ofopinionmaker: Those who occupy posi-tions whichenablethem to transmit,withsome regularity, opinions about foreignissues to unknown persons (Rosenau,1963:6). Keller's unctionaldefinitionmoreobviouslycontains a fundmentalproposi-tion:

The erm litesrefersirstof alltoa minorityof individualsesignatedo servea collec-tivity in a sociallyvaluedway. . . . Sociallysignificantelites are ultimately responsiblefor the realizationof majorsocialgoals andfor the continuityof the socialorder(Keller,1963:4).

Dahl (1958) defines elites as those who

more frequently make certain types of deci-

sions, thus combininga statistical and func-

tional definition.

Both statistical and functional definitionsof elite are derived from classificatory con-

cepts-i.e., the concept elite is defined ex-

plicitly or implicitly in terms of some other

primary concept. The vagueness and variety

of definitions of elite are thus caused, in

part, by relianceon some primaryconcept-

usually power or influence. The problem of

finding indicators appears to be passed on

to the primary concept. In statistical defini-

tions the measurement, n principle, is quitestraightforward:one lists all personsin rank

order of the possessionof an attribute-say,

power, religiousness, or what have you.

Those who are the top N percent (N is any

arbitrary number) are then called the elite.

(As Bottomore (1964:1, 2) points out, this

was Pareto's original notion.) In functional

definitions the primary concept concerns

some outputs of the elite rather than at-

tributes of the elite themselves. To find a

functional elite, all outputs of a relevantcategory or system are listed. The label

elite is assigned to those persons who are

said to have caused more outputs, or more

important outputs, where importance is

defined in some way other than frequency.

Structural efinitions f elitesuseconcepts

7For another and somewhatdifferentclassifica-tion of definitionsof elites, see Mills (1956:366).He gives personalityas a fourth definition.I in-clude it under statistical-a more abstract notionthan positional.Also, Bell, et al. (1965:Chapter2)offerstill anotherclassification.Ours s more formalthan the otherswhose categorieswill be discussedunder he topic of methodology.

Page 6: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 6/16

POWER,INFLUENCEAND SOCIALCIRCLES 689

such as group, role, or status. Such defini-tions tend to be complex. Since structuralconcepts are felt to be theoretically insuffi-cient to define an elite, other concepts suchas power or influence are usually added tothe definition. And

because these definitionscontain more than one concept, they eithercontain propositionswhich link the concepts,or at least the concepts imply a complexproperty space. Friedrich (1963:316) offersan example of a structural definition.

A politicalrulingor governingelite may bedefinedas follows: it is a group of personswho are distinguishedby exceptionalper-formancein politics, who effectively unite(monopolize) he ruleof a particular ommu-

nity in their hands,who possessa sense ofgroupcohesionand a correspondingsprit decorps, usually expressed n cooptation. . .a politicalelite excels in the abilityto securepowerandrule.

Two key concepts are invoked-power andgroup. Each is defined by a series of sub-conceptswhich serve as chapterheadings fora further series of implied reduction state-ments. Thus a group is defined as a unit

which has a high probability of possessingcohesion, an esprit de corps,and cooptation.Power is defined functionally as an outputof rules or decisions, and statistically aspossession of the attributes of potentialpower and exceptionalpolitical performance.In the attribute space, formed by cross-classifying power and group, only the con-joint cell is said to be an elite, though sometheorists would reserve the concept rulingclass for this cell. Sets of personswho have

powerbut who are not membersof a cohesivegroup are specifically excluded by Friedrichfrom the category of elite. In his terms, un-less such an assortment of persons, a classin the mathematicalor statistical sense, alsopossesses the characteristics of a group withinternal cohesion and a consequentcapacityfor acting jointly, it is hardly justifiable tospeak of them as a political elite.

On the other hand, Mills (1959:3, 4) spe-

cifically excludes the notion of a self-con-scious group, a matter which he prefers totest separately.8To him, the key structuralconceptsare positions or statuses.

The powerelite is composedof men [who]. . v are in positions o makedecisionshavingmajor consequences.Whether hey do or donot make such decisions is less importantthanthe fact thatthey dooccupysuchpivotalpositions. . . They are in commandof themajorhierarchies nd

organizationsf modernsociety. . . . They occupythe strategiccom-mandposts of the social structure, n whichare now centeredthe effectivemeans of thepowerandthe wealthandthe celebritywhichtheyenjoy.

Power, though never specifically defined, isapparently the potential to make importantdecisions, regardlessof whether such poten-tial is actually translated into action. Thisdefinition of elite is also statistical, for it

implies the proposition that in addition topower, the elite have more of whatever itis peoplewant. Domhoff (1967) bypasses thenotion of self-consciousness. His structuraldefinition of a power elite emphasizes inter-action between members of a structuralunit(the upperclass) fromwhich the powereliteis drawn.The power elite itself may or maynot be cohesive or exhibit high interactionrates.

Structural definitions seem to use morereductionstatements than do pure statisticalor functional definitions because they sub-sume such definitions. In addition to invok-ing some of the social conditions for power,structural definitions imply some notion ofpower as an effect or output, as well as somestatement about what is being manipulated.They tend to introduce another social unit-a governing or a ruling class (Bottomore,1964). In practice, the definitions are suffi-ciently complex so that the investigator istempted not to measure separately the im-plications of each reduction statement butrather to locate the particularstructuralunitnoted and simply announce it as powerful.Mills, for example, gives a list of commandposts assumed to be powerful; the occu-pants are the power elite. Thus the elite isdefinedas the powerful,but poweris definedas membership in the elite. This circularity

brings us to a brief mention of some of theproblemsin the methodologyof elite studies.

It would seem that finding an adequateset of reductionstatements for power (or in-fluence or literary quality, etc.) would alsosolve the problem of defining and locatingpowerelites, influential, or any other sort of

8Mills (1956:283) does find the power elite tobe socially connected and class conscious, butthat be reservesfor demonstration,not definition.

Page 7: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 7/16

690 AMERICANSOCIOLOGICALEVIEW

elite. Most of those working in the field ofelite studies, however, are interested not inthe refinement of the concept of elite butin the location of a set of individuals whoconstitute the elite. That is, interest has

centered on the specification of instancesrather than on the classificatory rule whichlocates these instances. The result has beenthat the validity of the various indicatorsof power or influence has frequently beenevaluated not on the basis of any specificset of reduction statements defining poweror influence but on whether or not the

right set of persons has been identified as

members of the elite. As we shall suggest,from the standpoint of a social circle theory

of elites, this is not unreasonable, thoughas Spinrad (1965) points out, it leads to aset of methods generally unrelated to classicnotions of power. The chief methods in use-the positional, reputational and decision-making methods (Bell, et al., 1965; Clark,1968)-will be analyzed in greater detailafter we discuss the theory of social circles.For the moment, we can observe that thedebate over which method is correct has

been quite acrimonious, and the differencein results may be more a reflection of thetype of community studied than anythingelse (Clark, et al., forthcoming). The fewstudies which have used more than onemethod in the same community report onlymoderate overlap in lists of elite obtainedfrom the various methods (Freeman, et al.,1963; Presthus, 1964).

The more fundamental problem in theliterature is the use to which definitionsof power and elite are put. While the studyof formal power or official authority andthose who wield it is interesting in itself,almost all sociologists and political scientistswish to go beyond the study of formal accessto facilities. The issue which most dividesstudents is the interconnection betweenholders of formal authority and the extentto which holders of formal authority are

connected to those who do not have such

positions. And it is in this area that currentdefinitionsof the elite and the methods used

to find it produce the most equivocal evi-

dence, and which suffers most from the

circularityof currentdefinitions. The matter

of linkages and connections will therefore

nowoccupy our attention.

SOCIALCIRCLESAND THE THEORYOF POWER

Most theorists agree that social power

can be studied effectively through the back-

ward approachof locating powerful persons.

For whatever their theory or method, ob-

servers agree that much social power isexercisedin the form of legitimate authority,

albeit not necessarily in political institu-

tions. Persons who hold such positions can,

in principle, easily be identified through

the use of informants, documentary ma-

terials, or both. Direct and detailed measure-

ment of power as a way of locating such

persons is generally rejected, except by the

decision-study school of power elites. Even

they tend to back up this direct measure-ment with several other techniques (Dahl,

1958).

The difficultyof measuring power directly

stems from its nature as a disposition con-

cept. Any disposition concept, such as at-

titude, requires the collection of a large

number of measurementswhich must then

be reduced with the aid of a mathematical

model. The model determines whether the

concept is, in fact, uni- or multidimensional.

The proper measurementof power requires

the examination of a large number of situa-

tions in which power is said to be exercised.

When more than two units are said to have

power, then the data which must be ex-

amined are still more complex. The multi-

measurement,multidimensionalapproach to

the measurementof power which we suggest

has, therefore, generally not been taken. In

part, this approach has not been used be-

cause power has been viewed as a conceptwhich should have a single definition and

therefore should be measured with a simple

measurement in one dimension. For the

moment, we bow to the collected intuitive

wisdom of students of the field, for the pro-

cedures we suggest would be quite arduous

if applied to an entire population. Ideally,

measurement of power should be confined

to those likely to have some in the first

place. Even the most confirmed decision-

making theorists of national or community

power begin their study with persons in

the position to make the decision.This leads

us right back to the problem of locating

powerelites.As with the study of organizations, there

are two systems of power in society at large.

Page 8: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 8/16

POWER,INFLUENCEAND SOCIALCIRCLES 691

One is the formal system, and the other,the informal system. The informal linking

mechanisms within formal organizationshave been fairly well studied. The same isnot true of the informal system when society

is viewed as a whole.The mechanisms hroughwhich such exten-sions take place [from small groupsto largeunits] also applyin extending ohesion romone rank to the whole organizational nit,and, for that matter, to a larger social unitas well, even to society as a whole.The studyof these mechanisms eemsto be one majordirection n which sociologymust develop fit is to become less a study of small groupsand more a study of society (Etzioni,1961:193).

The major informal mechanism which links

power persons and powerful organizationsis the social circle: the exact counterparton the social system level of the informal

shop system at the organizational level.

The structure and function of informal

social circles in powersituations is the singlemost controversial issue in the study of

power and elites, though the concept has

generally remained implicit. The key issues

of the field, reconceptualized n social circle

terms, are: (1) whether or not persons in

formal positions of power are linked to-

gether through informal networks of poweror influence; (2) whether or not personswho do not hold formal political system

power positions are nonetheless linked tothe formal positions in such a way that

they can systematically influence decisions;and (3) whether or not influential persons

who do not hold formal political powerare themselves also interconnected. Alliedto these questions are furtherspecifications:

What is the precise nature of links, if any,and what is the total structureof the linkagesystem. Further, the answers to these ques-tions are likely to be different for the six

aspects of power previously identified.In community studies, for example, if the

elite are closely responsive one to another,then the community power structure is said

to resemblea pyramid; 9 if the elite form twoor more cohesive circles, then power is saidto be factional; if there are separate circlesof elite which come together for limited and

9This is obviously just one type of responsivenetwork, but the lack of social circle theory hasblindedanalysts to other possibilities.

specific purposes, then power is said to be

coalitional. Finally, if it is difficult to iden-tify any coherent power circles, then poweris said to be amorphous (Walton, 1966a,

1966b). In studies of national power circles

in the Western democracies, Mills (1956),Hunter (1959) and Domhoff (1967), who

wrote on the United States, and Guttsman(1963), who wrote on Britain, all found

strong linkages between members of the

elite. On the other hand, Aron (1966) andKeller (1963), in general reviews, and Par-

sons (1960), Dahl (1961), Kornhauser

(1965) and Rose (1967), all writing about

the United States, each for somewhat dif-ferent reasons, found that linkages were

non-existent or highly limited in scope andfunction. Keller (1963:149) even found thelack of linkages between the various stra-

tegic elites to be a social problem. Domhoff(1967:Chapter 7) was content to find thatthe governing class, which controls the

power elite, is mainly recruited from anational upper class. At the same time he

suggested that these groups might be splitinto cliques of unspecified power.

The lack of a clear model of connectednessmeans that the limited amount of data col-lected is subject to various interpretations.For example, in the community power field,Walton (1966a:431) classifies Hunter ashaving found, in his major study, a pyra-midal structure. Yet Hunter himself sum-marizes his work as follows:

Every communityI have visited or studiedhas hada well-defined,elativelysmallgroup

of people who constitute the local powerstructure. They most often represent thelargest ocal industries, anks, awfirms,com-mercialhouses and newspapers.Major pro-jects must haveat least the informal anctionof the majority of these policy makers.Butthey are not a singlepyramid f power.Thereare revolvingcommitteeclusteringsof lead-ers who link with other committee cluster-ings on matters of major policy concern(Hunter, 1959:5, italics supplied).

In a national elite study, Guttsman foundthat a ruling class existed in Britain andthat movementsbetween elite groups tendedto increase (1963: Chapter 11). A similarfinding in the United States indicated toMills that he indeed was dealing with apower circle. But Guttsman noted that incontrast to the nineteenth century, the con-

Page 9: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 9/16

692 AMERICANSOCIOLOGICALEVIEW

temporary elite appears at first sight dis-

parate and lacking unity. Sereno explained

that it is precisely because the concept of

interpersonal nfluence is so intractable that

he believes there is no such thing as a ruling

class (1962:79).

CHARACTERISTICS OF SOCIAL CIRCLES

Whatever one's favorite method for locat-

ing elites and, hence, for studyingcommunity

or national power systems, the matter of

social circle connection must be dealt with

systematically. First, some working defini-

tion of the phenomenonmust be established.

A social circle has three defining charac-

teristics, two of which are positive and one,negative: (1) A circle may have a chain

or network of indirect interaction such that

most membersof a circle are linked to other

members, at least through a third party.

It is thus not a pure face-to-face group. (2)

The network exists because membersof the

circle share common interests-political or

cultural. (3) The circle is not formal-i.e.,

there are: (a) no clear leaders, although

there may be central figures; (b) no clearly

definedgoals for the circle, though it almost

always has some implicit functions; (c)

no definite rules which determine modes of

interaction, though there are often cus-

tomary relationships; and (d) no distinct

criteria of membership.

There are four kinds of common interests

that circle members may have, and thus

there are four kinds of circles: Cultural

circles draw members together on the basis

of valuational goals such as religion, psy-chotherapy (Kadushin, 1966) and other

philosophiesof life ; expressive goals such

as literature, art and recreation (Kazin,

1965; Wilson, 1964; Podhoretz, 1968); and

cognitive goals such as science and tech-

nology (Price, 1963; Mullins, 1966; Cole

and Cole, 1967). Utilitarian circles are char-

acterized by the need to trade goods and

services with other producers in external

economy industries (Vernon, 1960) such

as Wall Street, or Seventh Avenue or

Hollywood . Power and influence circles

are exemplified by our current discussion.

Finally, there are integrative circles which

are elaborationsof interactionresultingfrom

some common experience such as ethnic

membership, wartime experience, or mem-

bershipin an occupationalcommunity (Lip.

set,et al., 1954).10The structure of these circles tends to

differ. Cultural circles have a core of pro-

ducers of symbols and are surrounded by

a periphery of symbol consumers and valid'dators. Utilitarian circles are less concentric

in form and more overlapping in nature.

Power circles are more pyramidal, while

integrative circles are often the most loose

and egalitarian in structure. Their linkage

to formal organizationalstructures tends to

vary by type of circle. Cultural circles are

often completely unattached to formal or-

ganizations and hence have often relied

on a direct salon or a particular meeting

place as a unifying device. Utilitarian cir-cles form connections between formal or-

ganizations. Power circles do this too, but

also exist as units within some larger and

more amorphous political unit, and they

may or may not have a coherentor cohesive

ideology (Agger, 1964). For various reasons,

connections in utilitarian and power circles

are more covert and less legitimated. Inte-

grative circles frequently are hung upon or

grow into or develop from various forms ofvoluntary organizations. There are also

''supercircles which link the inner core of

two or more circles or different types. Fre-

quently, it is asserted that there are super

circles of a combinedutilitarian and power

axis.

An expositionof the theoryof social circles

is beyond the scope of this paper. But it

is important to observe that many of the

issues warmly debated in the field of power

and elite studies are systematically identi-

fied within this framework. The issue of

super circles and the relation between social

circles and formal organizations are ob-

vious problems. More generally, the kind

of structure which characterizes different

social circles is important to the power

field: Are circles closely knit? Are they

primarilyface-to-face units or are they com-

posed of a string of personswho know other

persons? Are there sociometric stars? Arethere inner and outer circles? What are

10 These types correspond to Parsons' latent pat-

tern maintenance, adaptation, goal achievement and

integrative functions (Parsons et al., 1953:179-

190). A full exposition of these types is beyond the

scope of this paper.

Page 10: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 10/16

POWER, INFLUENCE AND SOCIAL CIRCLES 693

recruitmentndexpulsionmechanisms? recirclesborn, and if so, how do they die?Do powerand influence ircleshave,as weclaim, structuresthat are differentfromothertypesof circles?Finally, it is evidentthat

the structure f circlesdepends n parton the natureof a particular ocialsystem,so there is no a priori reasonto assumethat power structures n all communitieswill have the same form (Agger, 1964;Clark, 1967a and forthcoming).

The Social Circle Method for StudyingPower and Influence. The majorissues inthe studyof powerand influencehave beenreconceptualizeds problems f socialcirclestructure, function, and development.If

systematicwaysof studying ocialcirclesofpowerfulor influentialpeoplecanbe found,then many of the problemsof studyingpowerand influencecan be located if notresolved. In general, the problem is to con-struct an open-ended sociometric, ratherthan a sociometric of a closed system.Further,the measurementmust allow forindirectas well as direct interaction.Thebasis for the interactionmust be specified.

Thoughsome of the samepersonsmay beinvolved,and hence the circlesmay to amajor or minor degree overlap, circlesformedon the basis of socio-political n-fluencewillnot be the sameas thoseformedon the basis of, say, literary influence.Finally,all circlesare in variouswayslinkedto formal structures,as suggestedabove.This meansthat in any institutionalarea,the location of informalcirclesmay serveto locate formalstructures, nd vice versa.Nevertheless,becausesocial circlesdo nothave a formal leadership,positional orformal eadershipvariablesare neveralto-gether satisfactorysubstitutes.

Any sociometricmust decide upon theparticular ink between personswhich isto be studied. Best friend, personyouwould most want to sit next to, and soon, havebeentraditionalinkages.Linkagesthathavebeenusedin paststudiesof power

includeparticipationn the same decisionand general nfluence.Obviously,discussionof politicaland economicssuesshouldalsobe includedas linkages.All of this pointsto using the concept nfluenceratherthanpower in studyinglinkages.Influence m-plies informality,he sort of thingwe wish

to measure with social circles. It also isdefined in terms of the probability thatsomeone will take into account the wishesof others. This is presumablywhat is meantby the expectation that true power is

exerted not by the person in the formalauthority position, but by some other per-sons or class of persons. As long as the

person in the formal position thinks thathe takes another into account, then thatother person must be said to have influence.This kind of influenceis not mere percep-tion as Rose (1967:Chapter 18) wouldhave it. This also suggests that some of thebest informants about influence must bethe very occupants of formal decision-

making positions (Katz and Lazarsfeld,1955). It follows that once a person holdinga formal power position is found, he mustbe asked about other persons who influence

him, much as has been done in studiesof individual decision-making (Kadushin,1968a). His perception about who is gener-ally influential is also important but is adifferent matter. On the other hand, a per-son may unwittingly take the desires of

others into account or may be placed in asituation in which the social frameworkserves to limit severely the course of his

action (or inaction). Here the direct useof informants may be of only limited value.It is not so much the individual influencerthat is so important as the entire socialcircle of influencerssurrounding he decision-

maker and of which he himself is a part.This again has a parallel in studies of indi-vidual decision-making.In many individualdepth decisions, such as going to a psychi-atrist, it is not the overt pressureof a singleinfluencer which makes a person act, butrather the pressure as expressed by thenorms of an entire social circle (Kadushin,1966, 1968b). The concept of social circlechanges the notion of influence from oneof perception to one of social structure.

Our analysis of current methods of study-ing elites begins with the one closest to the

social circle method. In the reputationaltechnique, a preliminary list of the elite is

constructed with the aid of informants anddocuments. In practice, this first list oftenconsists of a positional elite. Under the

assumptionthat the best judges of influence

and power are the power elite themselves,

Page 11: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 11/16

694 AMERICANSOCIOLOGICALEVIEW

this list is then shown to those named onthe list and perhaps to other importantpersons. The judges may now rank the per-sons on the list (or vote for them) and/orthey may be asked to name other persons

who should qualify for such a list but werenot placed upon it. The other influentialswho are named (or who are named suffi-ciently often) are then interviewed andshown the list, and they too are asked tosuggest other influentials.The process stopswhen some criterionis reached,usually whena significantproportion of the same personsis repeatedlynamed.

The reputational method obviously meetsmany of the requirements or circlemeasure-

ment. Its most important virtue is that itcan be considered a snowball sample (Good-man, 1961). A snowball sample is a devicefor obtaining an open-ended sociometric.Starting with a given list, usually a sampleof some universe, each respondent is askedto name several others who are then inter-viewed, and so on. The problem with manyreputational studies is that they have notbeen formally conceptualized as snowball

samples and hence have not always main-tained adequate controls over the snowballprocess. The feeling that reputationalstudies err because the elite play favorites intheir nominations of other elite (Presthus,1964:110) or that they offer only percep-tions of power (Rose, 1967) is now seenas an advantage rather than as a disad-vantage. The error helps to trace thelinkages in the power circle. On the other

hand, some of the most serious criticism

of the reputationaltechniquestems from thelack of clarity with which the snowballmethod is used. In particular, the basis on

which the circle is extended is not alwaysclear, for the reputational technique at-tempts to discover four things simultane-

ously: the generalformal structureof power,the general informal structure, the percep-tion of formal and informal power, and theparticular formal and informal interaction

partners of members of the power circle.In principle, there is no reason why a singlestudy cannot accomplish all these aims,but if the goals have not been properlyconceptualized in advance, as they gener-ally have not, the result is likely to beconfused.

From a formal point of view, the majorcompetitor of the reputational technique,the decisional method, is simply anotherkind of snowball sample. The decisionmethod works not with some general or

hypothetical influence links, but with in-fluence upon specific historical issues aslinkages. The snowball starting points are

drawn in two steps: first, a list is madewith the aid of documents and informants

of important recent community decisions.

Then, with the aid of documents and/orinformants, a list is drawn of persons who

are thought to have participated in thedecision. Again, this is often a positionallist. These persons are then interviewed or

documentary evidence is consulted, in partto ascertain their exact role, if any, in thedecision or series of decisions. While reputa-

tional studies usually ask respondents torate the importance of other respondents,

most decisional studies do not do the

analogous thing-they do not ask respon-dents to rank the importance of the various

decisions (though see Agger, 1964). Butthey do ask respondents for a list of others

who participated in the decision in certainways. Should new names not previously onthe investigator's list be derived from theinterviewers, these new persons are also in-terviewed. Again, the process stops when

some criterion of repetitiveness is reached.

Once a decision study is seen as an open-ended sociometric a number of new prob-lems emerge. By failing to conceptualizedecision studies as snowball samples, in-

vestigators have tended not to conceptualizethe role of the decision-maker respondent.He may be asked to act as an informant and

to name people who generally participatedin a decision. But he may also serve as a

respondent and may report the persons who

influenced him, whom he influenced, and

with whom he interacted in particular ways

at particulartimes in the decision. Althoughthese distinctions may not matter much

in a very small power elite, more complex

situations will yield quite different lists ofpeople for different modes of interactionand influence. Decision methods thus tend

to give a false sense of precision. They too

are subject to sociometric error, and since

the dimension upon which the snowball is

expanded is usually not clearly noted, it is

Page 12: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 12/16

POWER,INFLUENCEAND SOCIALCIRCLES 695

difficultto ascertain the groundsupon whichpersons are drawn into the sample. Curi-ously, only to the extent that respondentsare asked to act exclusively as purely ob-jective informantsare the data both limited

and confused. In any case, the networkaspect of decision-making is usually un-available in decision studies. At best, thedata are examined only in terms of theoverlap between participantsin one decisionand participants in another. Finally, theproblem of sampling decisions has beenoften noted.

Even positional methods of studyingelites have used social circle methods. Inthe most ambitious application, closed sys-

tem sociometrics are used. The entire listof positional or decisional elite derived froma previous research step is shown to thoseon the list who are then asked to indicatewhom they know and in what capacities(Bonilla and Silva-Michelena,1968; Hunter,1959; Agger, 1964). This method indeedgives closure, but aside from its cumber-someness (examination of the list may useup a major part of the hour or so one can

get from most elites), this very closurecreates problems. One of the chief charac-teristics of a social circle-its informalboundaries-is lost. In practice, this meansthat once assembled, the initial list cannotbe self-correctingin the fashion of a snow-ball sample. There are many indirect waysof demonstrating common social circlemembership,however,and positional studieshave provedmost ingenious in trackingthesedown, though no one study has

used allmethods. Most of the indirect measurementsare based on some functional consequencesof circle membershipor upon some set ofcircumstances that might lead to a highprobability of membership in a commoncircle.

One may assume that social circle mem-bership is identical with membership in agiven subculture or social class. Studieswhich attempt to show that members of

a power elite come from an upper classare of this type. Merely demonstratingthiscommon origin is usually insufficient sincemost studies wish to talk about the inter-connection of the elite. Thus an attemptis made not only to show common valuesand origins but also participation in com-

mon organizations or, even better, common

activities or instances of having been in

the same place at the same time. Common

schools, clubs, resorts, coffee houses and

recreational or cultural interests, even ap-

pearances in the same journals, have allbeen used as indirect evidence of the con-

nection between elites. Some positional

studies use the organization as a unit of

analysis and produce organizational powersociometrics by showing that the sameperson is a member of two or more or-

ganizations. (This method was used byYoung and Larson, 1965:926-934, to

measure the structure of an entire small

community.)

In sum, once the social circle characterof elites and power elites has been recog-nized, and once it is seen that most studies

have, in fact, relied on this characteristic

to gather their data, it is necessary only to

formalize these techniques to gain superiordata. Sociometric techniques should be used

both to locate the sample and to analyze itsinterconnections. The more varied the

bases for the sociometric questions, the

richer the material and the more easily cansome of the issues outlined above be re-

solved. Items pertainingto general influence,discussion partners, and actual decisionsshould be included. The distinction betweenmembers of the same type of circles and

membersof other types should be maintainedso as to allow for the analysis of both func-tional circles and super circles. There re-

IThe methods of analyzing the sociometricsof

a snowball sample are somewhat different fromthose used with closed system samples.If most ofthe sampleis drawn as a snowball rather than asa positionalsample,then by definitionthe densityof interconnectionswill be higher, since to be amember of the sample in the first place one musthave been named by at least one person. The con-cept of a biased net is appropriate.For a reviewof appropriatemodels, see Abelson (1966), Cole-man (1964:Chapters14 and 17), and Rapoport(1963). It must be admitted that at present noneof these models is altogethersatisfactory for ourpurposes.The lack of techniques,muchless models,

for handling sociometricdata has probably beenresponsible or the failure formally to analyzenet-work data collected n such studies as Linz (1967)and Bell (1964), which have used snowball andcircle approaches.Agger (1964) representsa soph-isticated attempt to combine decisional,nomina-tion and informant data to arrive at a compara-tive study of power circles, but the snowballmethod s not formallyapplied.

Page 13: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 13/16

696 AMERICANSOCIOLOGICALEVIEW

mains, however, the questionof selectingthe right starting points for the snowball.If the elite is at all cohesive,then it pre-sumably makes no differencewhere onestarts, since all who count will eventually

be named.But this assumptions often ex-actly what one wishes to test. In such acase, a positional or decisionalelite as astarting point may be the only recourse.Unfortunately, here is no substitute forknowledge f the social system for selectingthe appropriate institutional commandposts or the importantdecisionsfor theinitial sample.This is the kind of informa-tion whichcan best be given by informants(Zelditch,1962) or which can be obtained

from documents,though as Bonilla andSilva-Michelena 1968) point out, a listof positions is, in fact, a reputationalstudy of positions.

The advantage f a snowball tudyof theelite is that it is largely self-correcting.Nonetheless, differentelite structuresanddifferentsamplestrategiescan yield quitedifferent esults,and the procedure an be-comequitecomplex.Let us suppose hatwe

are planning a study of economic andgovernmentalelites in a moderate-sizedcountry.We constructan initial universelist composed f all firstandsecondofficersof all enterprisesabove a given size, alllegislators,and all civil servants above agiven level. We take a proportionalampleof these lists as our initial starting pointsand perhapsweight accordingto size offirm and importance f governmental osi-tion. 2Two dimensions overnpossibleout-comesof nominationsor the secondwaveof a snowball sample with this design:whether the nominations are containedwithin the initial universe ist or not, andwhetherthey tend to comefrom the sameinstitutionalarea or to crossover into theother one. Nominationsargelyconfined othe same nstitutionalectorsand the initiallist confirmhe existenceof two fairly tightcircles. Nominationscross-cutting he in-

stitutional ectorsshowa tight supercircle.The disproportionate vote received by

12 As with all sociometricand structuralstudies,there is the issue of generalcoverageversus struc-tural density.The resolutionof this problemmustalways be at the level of expediency, given theresources t hand.

certain ndividual an confirm r correct heweightingscheme.If the nominationsendto fall outsidethe initial universe ist, thenthere are basically three possibilities: theinitial startingpoints were simply wrong;

the initial ist wasincomplete; r a kitchencabinet phenomenons operating n whichpowerfulpersonsname their close advisorswho themselveshave no formalpowerposi-tion. Subsequentwaves can check some ofthese possibilities. f relativelyfew namesare added on subsequentwaves, then thenew universe ist formedby the initial listplus all new namesmentioned ay, by atleast two or threepersons, s likely to bethe right one. Constant expansionof the

lists shows a diffuselystructured lite. Thediscovery of new circles which nominateeach other on subsequentwaves but whichrarely refer to anyone on the initial listwould be of special nterestand, dependingon other information,would suggest thatthe true entersof powerare behind hescenes, that a self-delusionalgroup hasbeenfound,or thatsecond evels n a societyformtheirowncircles.Thesearenot all the

possibilitiesand only one basis for linkageis assumed,but some notion of the richnessof data obtained n this way hasbeen given.It is possible to generate empiricallygroundedhypothesesabout elite structurethat might otherwisehave never been evenconsidered,much ess verified.

Once the sample is obtained,then theelite's degree of poweror influenceshouldbe checked ccordingo reports f whattheysay they actuallydo. That is, what are theirinputs and outputs.Properlydeveloped,n.formation n these areas providethe ma-terials to fill in the reductionstatementsaboutpowerand influence.No doubt somepersonsobtainedin the snowballwill notbe producersof opinionswhich influenceothers,nor will they have made decisionsor produced ositionpaperswhichhave hadan impacton societalprocesses.These mat-ters can be solvedby consideringhe entire

classification f reduction tatementsaboutpowerwhich we earlierderived from theliterature. A full considerationof themethodologyor measuring owerafter theelite has been identified s impossiblehere.In any case, it will be apparentthat amultidimensionalndexof power, nfluence,

Page 14: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 14/16

POWER,INFLUENCEAND SOCIALCIRCLES 697

and opinion-makingability is likely toemerge rom such an effort.Someof thosedrawn ntothe elitesamplemayscoreso lowthat theywill be eliminated.Thusthe indexserves as an additionalsamplecontrol,as

well as a variableto explainand analyze.

CONCLUSIONS

Thoughpoweris a hallowedconceptofpolitical and sociological theory, it hasprovedmostdifficult o define n wayssuffi-ciently preciseso that it can be measured.We have shown that the difficulty s notcausedby the ineffectivenessf the scholarswho have attemptedthese definitions,but

ratherby the natureof poweras a disposi-tion concept.As such, it can only be speci-fiedthrougha largeseriesof reduction en-tences.There can be no singledefinitionofpower.Direct measurements f power turnout, therefore, o be a somewhatmpracticaldeviceforlocatingpowerelites.In fact, thecurrentmethodsfor studyingelites makeslittle use of classic notions of power. Inpart,this stemsfromthe difficulty f apply-ing the conceptof powerand in part fromtheimprecisewayin whichelites havethem-selvesbeendefined.Atpresent, hetechniquefor the studyof social power s to locateanelite and then to observe ts use of power,or to locate a decisionand to ascertainwhodeterminedits outcome. Although theseare not unreasonableprocedures,currentmethodshave suffered roman inadequateconceptualizationf one of the central the-oreticalissues about powerelites, namely,

the extent to whichthey areinterconnected.We have shownthat the conceptof socialcirclenot only servesto locatemost of theimportantssues in the study of elites butalsoserves o unifyand to formalizemethodsfor ocating hem.

REFERENCES

Abelson,RobertP.1966 Mathermaticalmodels in social psychol-

ogy. Pp. 1-25 in Leonard Berkowitz(ed.), Advances in ExperimentalSocialPsychology,Volume 3. New York: Aca-demicPress.

Agger,Robert,et al.1964 The Rulers and the Ruled. New York:

Wiley.Aron,Raymond.

1966 Social class, political class, ruling class.

Pp. 201-210 in ReinhardBendixand S. M.Lipset (eds.), Class, Status and Power(2nd ed.). New York: Free Press. Firstpublished n EuropeanJournal of Sociol-ogy 1 (1960):260-281.

Bachrach,Peter and MortonS. Baratz.1962 Two faces of power. AmericanPolitical

Science Review 56 (December)947-952.Bell,Wendell, t al.

1964 Jamaican Leaders: Political Attitudes ina New Nation. Berkeley: University ofCaliforniaPress.

1965 Public Leadership.San Francisco: Chand-ler.

Bonilla, Frank and J. A. Silva-Michelena.1967 Politics of Change in Venezuela. Volume

I: Strategy of Research on Social Policy.Cambridge: Massachusetts Institute ofTechnology.

Bottomore,T. B.1964 Elites and Society. New York: BasicBooks.

Cartwright,Dorwin.1965 Influence, eadership, control. Pp. 1-47

in James G. March (ed.), Handbook ofOrganizations.Chicago: Rand McNally.

Clark,TerryN.1967a Power and community structure: who

governs,where, and when? The Sociologi-cal Quarterly8 (Summer):291-316.

1967b The concept of power: some overem-phasizedand underrecognized imensions.

Southwestern Social Science Quarterly(December)271-286.1968 The concept of power. Pp. 45-87 in

Clark (ed.), Community Structure andDecisionMaking. San Francisco:Chandler.

Forthcoming. Community structure, decision-making, budget expendituresand urbanrenewalin 51 Americancommunities, nCharlesM. Bonjean, Clark, and RobertL. Lineberry (eds.), CommunityPolitics:New York: FreePress.

Clark,TerryN., et al.Forthcoming. Discipline, method, community

structure and decision-making. AmericanSociologist.

Cole,S. andJ. R. Cole.1967 Scientificoutput and recognition:a study

in the rewardsystemin science. AmericanSociological Review 32 (June) :377-390.

Coleman,JamesS.1964 Introduction to MathematicalSociology.

New York: FreePress.Dahl,RobertA.

1958 A critique of the ruling elite model.AmericanPolitical ScienceReview 52:463-

469.1961 Who Governs?New Haven: Yale.Danzger,Herbert.

1964 Community power structure: problemsand continuities. American SociologicalReview 29 (October):707-717.

Domhoff, G. William.1967 Who Rules America? Englewood Cliffs:

Prentice-Hall.

Page 15: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 15/16

698 AMERICANSOCIOLOGICALEVIEW

Edinger,LewisJ.1967 PoliticalLeadershipn IndustrialSocieties.

New York:Wiley.Etzioni,Amitai.

1961 A ComparativeAnalysis of ComplexOr-

ganizations.New York: Free Press.Freeman,L., et al.

1963 Locating leaders in local communities:a comparison of some alternative ap-proaches. AmericanSociologicalReview28 (October):791-798.

Friedrich,CarlJ.1963 Man and his Government. New York:

McGraw-Hill.Goodman,LeoA.

1961 Snowball sampling. Annals of Mathe-matical Statistics32:148-170.

Guttsman,W.L.1963 The British Political Elite. New York:

BasicBooks.Hempel,CarlG.1952 Fundamentalsof concept formation in

empiricalscience. P. 24 in InternationalEncyclopediaof Unified Science (VolumeII, No. 7). Chicago:Universityof ChicagoPress.

Hunter,Floyd.1959 Top Leadership,U.S.A. ChapelHill: Uni-

versity of North CarolinaPress.Kadushin,Charles.

1966 The friends and supportersof psycho-therapy: on social circles in urban life.

AmericanSociologicalReview 31 (Decem-ber):786-802.

1968a Reasonanalysis. Pp. 338-343 in the In-ternationalEncyclopediaof the SocialSci-ences (2nd ed.). New York: The Mac-millanCompanyand The FreePress.

1968bWhy People Go to Psychiatrists. NewYork: Atherton.

Katz,Elihuand PaulLazarsfeld.1955 PersonalInfluence.New York: Free Press.

Kazin,Alfred.1962 StartingOutin the Thirties.Boston:Little,

Brown.Keller,Suzanne.

1963 Beyond the Ruling Class: StrategicElitesin Modern Society. New York: RandomHouse.

Kornhauser,William.1966 'Power elite' or 'veto groups. 'Pp. 210-

218 in ReinhardBendix and SeymourM.Lipset (eds.), Class,StatusandPower (2nded.). New York: Free Press.

Lasswell,Harold.1950 World Politics and Personal Insecurity.

Glencoe:Free Press (Originallypublished,1934).

1965 Introduction: the study of politicalelites. P. 4 in HaroldLasswelland DanielLerner (eds.), WorldRevolutionaryElites:Studies in Coercive Ideological Move-ments. Cambridge:Massachusetts nstituteof TechnologyPress.

Lasswell,Harold D. and AbrahamKaplan.1963 Power and Society.New Haven: Yale Uni-

versityPress.

Linz,Juan.1967 Local elites and social change in rural

Andalusia.Unpublishedreport (July).Mills,C.Wright.

1956 The Power Elite. New York: OxfordUni-versityPress.

Mullins,N. C.1966 Social Networks among Biological Scien-

tists. Harvard University Ph.D. disserta-tion.

Lipset,SeymourM., et al.1954 Union Democracy.New York: FreePress.

Parsons,Talcott.1960 The distributionof power in American

society. Pp. 199-225 in Parsons (ed.),Structureand Process n Modem Societies.New York: Free Press. [First published n1957,WorldPolitics 9 (October).]

1966 On the concept of political power. Pp.

240-265 in ReinhardBendixand SeymourM. Lipset (eds.), Class, Status and Power(2nd ed.). New York: Free Press.[First published in 1963: Proceedingsof the American Philosophical Society197 (June).]

Parsons,Talcott,et al.1953 WorkingPapersin the Theory of Action.

New York:FreePress.1954 Toward a General Theory of Action.

Cambridge:HarvardUnivesrityPress.Podhoretz,Norman.

1968 Making It. New York: Random House.

Presthus,Robert.1964 Men at the Top: A Study in Community

Power. New York: Oxford UniversityPress.

Price,Derek.1963 Little Science, Big Science. New York:

ColumbiaUniversityPress.

Rapoport,Anatol.1963 Mathematicalmodels of social interac-

tion. Pp. 495-545 in R. Duncan Luce,et al. (eds.), Handbook of MathematicalPsychology.New York: Wiley.

Riker,William.

1964 Someambiguitiesn the notion of power.American Political Science Review 58(June):341-349.

Rose,ArnoldM.1967 The Power Structure.New York: Oxford

UniversityPress.Rosenau,JamesN.

1963 National Leadershipand Foreign Policy:A Case Study in the Mobilizationof Pub-lic Support.Princeton:PrincetonUniver-sity Press.

Rustow,DankwartA.1966 The study of elites: who'swho, when and

how. WorldPolitics18 (August) 690-717.Sereno,Renzo.

1962 The Rulers.New York: Praeger.Spinrad,William.

1965 Powerin local communities.SocialProb-lems 12 (Winter):335-356.Also, pp. 218-231 in ReinhardBendix and S. M. Lipset(eds.), Class,Statusand Power (2nd ed.).New York:FreePress.

Page 16: Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

8/12/2019 Power, Influence and Social Circles a New Methodology for Studying Opinion Makers

http://slidepdf.com/reader/full/power-influence-and-social-circles-a-new-methodology-for-studying-opinion 16/16

NATIONALPOLITICALDEVELOPMENT 699

Vernon,Raymond.1960 Metropolis, 1985. Cambridge: Harvard

UniversityPress.Walton,John.

1966a Discipline,methodandcommunitypower:a note on the sociology of knowledge.American Sociological Review 31 (Octo-ber) :684-689.

1966b Substance nd artifact: the currentstatusof research on community power struc-ture. American Journal of Sociology 71(January) 430-438.

Wilson,Robert(ed.).1964 The Arts and Society. Englewood Clifs:

Prentice-Hall.

Young, Ruth C. and Olaf F. Larson.1965 A new approach to community struc-

ture. American Sociological Review 30

(December)926-934.Zelditch,Morris,Jr.

1962 Some methodologicalproblems of fieldstudies. American Journal of Sociology67 (March):566-576.

MULTIVARIATE ANALYSIS OF NATIONALPOLITICAL DEVELOPMENT *

MARVINE. OLSEN

Indiana University

The purposeof this study is to exploreempiricalrelationshipsbetweenpoliticaldevelopmentand socioeconomicmodernization,with emphasis on the multidimensionalnature of bothprocesses.The process of national political development s conceptualizedas consisting ofseveral nterrelated nd separatelymeasurable imensions.Data from115 independentnationssupportthe argumentconcerning he importanceof overall socioconomicmodernization, ndparticularly ransportationacilities, for political development.Each dimension of nationalpoliticaldevelopment s stronglypredicted by variouscombinationsof socioeconomicvari-ables. The findingsmay be explainedby use of Karl Deutsch's dea of social mobilization-the involvementof individuals n non-traditionalpatternsof collectiveactivity. It is sug-gested that industrializations necessary o nationalpoliticaldevelopmentbecause t createsthe economic oundationnecessaryor both socialmobilizationof the populationand ensuingpoliticaldevelopment.

THE past years have witnessedrapidlygrowing nterestamong ociologists ndpoliticalscientists n the links between

the processesof politicaland socioeconomicdevelopment.'The basicproposition nder-lying this work is that all dimensionsof

modernization end to be interrelated, o

that no one type of development-especiallypolitical-can proceed ar withoutconcomi-tant growthin many other areas (Caplowand Finsterbusch,n.d.) More specifically,the intriguing hesis has been proposedbySeymourMartinLipset (1960:27-63) thatpoliticaldemocracy s highly dependenton

socioeconomic expansion in such realms asindustrialization, urbanization, and educa-tion.

The existing empirical studies of socio-economic and political development are con-

siderably less conclusive than are Lipset'stheoretical arguments,however.The purpose

of this research is to explore empirical rela-tionships between these two forms of soci-etal modernization more extensively thanhas been done in previous work, with par-ticular emphasis on the multidimensionalnature of both processes.

The theoretical reasoning underlying Lip-set's thesis has been widely discussed, andneed not be repeated here, except for onepoint of clarification. Whereas Lipset dealt

explicitly with the growth of politicaldemocracy, our concern here is with thebroaderprocess of national political develop-ment, of which democratization is onlyone possible aspect. Following Huntington(1965:386-430), we definepolitical develop-ment as the institutionalization of politicalorganizationsand procedures, with level of

* Revised version of a paper read at the 1967annual meetingsof the AmericanSociologicalAs-sociation. This researchwas partially conducted nconjunctionwith the 1966 SummerSeminarof theInstitute for ComparativeSociology, held at In-dianaUniversity.

1For a representative amplingof this literature,see Almond and Coleman (1960), Apter (1965),Braibantiand Spengler(1961), Cutright (1963and1967), Deutsch (1961), Lerner (1958), von derMehden (1964), Rostow (1960), and Shannon(1958 and 1959).