- 98 -fits in a given industry of firms . The extent to which each sector of the American economy is problematicfor obtaining profits in two-digit <strong>and</strong> four-digit manufacturing industries is estimated from researchlinking industry profits with the form of the pattern of relations defining the industry as a position inthe network of dollar flow transactions given in the 1967 Input-Output Study for the United States . A twostage process is described for sampling firms representative of large corporations involved in Americanmanufacturing . Measures of alternative strategies for interlocking across sectors are described . Twoclasses of hypotheses are derived : (1) Firms in an industry should interlock with firms in some othersector in proportion to the extent to which the sector constrains the industry's profits . (2) Controllingfor production <strong>and</strong> market differences, the ability of firms in an industry to obtain unusually high profitsreflects their success in creating interlocks with those sectors most problematic for their industry'sprofits .Abstracts form SOCIOLOGICAL METHODS AND RESEARCHNetwork Analysis, edited by Ronald S . Burt .7 (No . 2) (November, 1978) . Special Issue on AppliedBurt, Ronald S . (Sociology, U . California-Berkeley) "Applied Network Analysis : An Overview .""Manuscripts have been solicited for this collection so as to represent six modes of network analysiscurrently popular in the social sciences . The collection will hopefully expose the range of practicalissues relevant to applied network analysis . The six modes of network analysis are distinguished by twoparameters ; (1) the level of aggregation of actors - individuals versus subgroups within a system versuswhole systems, <strong>and</strong> (2) the approach taken to linkages between actors - the relational approach versus thepositional approach . The relational approach is typified by traditional sociometry in its focus on therelations between actors . Subgroups within this approach are 'cliques .' Actors are aggregated into aclique to the extent that the actors are connected to one another by intense relations . The positionalapproach focuses on the pattern of relations in which an actor is involved . All an actor's relations to<strong>and</strong> from other actors in a system define the actor's position in the system . In terms of academictradition, the positional approach is associated with social stratification ; positions operationalizingstatuses <strong>and</strong> role-sets . Subgroups within this approach are 'statuses' or 'jointly occupied positions .'Actors are combined into a single position to the extent that they have identical patterns of relationswith other actors in a system . Such actors are 'structurally equivalent' to one another ." (excerpted fromarticle ; for full text, see <strong>Connections</strong> 1 (3) : 9-12) .McCaZZister, Lynne <strong>and</strong> Claude S . Fischer (Sociology, U . California-Berkeley) "A Procedure for SurveyingPersonal <strong>Networks</strong>"The application of network analysis to certain issues in sociology requires measurement of individuals'personal networks . These issues generally involve the impact of structural locations on persons' sociallives . One such case is the Northern California Community Study of the personal consequences of residentialenvironments . This article describes <strong>and</strong> illustrates the methodology we have developed for studyingpersonal networks by mass survey . It reviews the conceptual problems in network definition <strong>and</strong> measurement,assesses earlier efforts, presents our technique, <strong>and</strong> illustrates its applications .Lin, Nan (Sociology, SUNY-Albany), Paul W. Dayton <strong>and</strong> Peter Greenwald (New York State Department of Health)"Analyzing the Instrumental Use of Relations in the Context of <strong>Social</strong> Structure"Employing a variation of the Small World technique for tracing social relations in the context of alarger social structure, the instrumental uses of social relations are examined in terms of the prestige<strong>and</strong> types of relations characteristic of participants in the search process . The results show thatsuccessful chains tend to involve participants of higher occupational prestige as the chains progressbefore "dipping" down toward the target prestige level at the last link . Also, the successful chainstend to utilize weak <strong>and</strong> infrequent social relations rather than strong <strong>and</strong> frequent social relations .Alba, Richard D . (Sociology, Cornell U .) <strong>and</strong> Gwen Moore (Sociology, SUNY-Brockport) "Elite <strong>Social</strong> Circles"We describe a method for locating the denser, or more cohesive, parts of networks . The method startsfrom the identification of cliques, or maximal complete subgraphs . Since there are numerous such subgraphsin most networks, they are then aggregated when they overlap sufficiently . The resulting aggregatedsubgraphs are frequently large, not necessarily disjoint, <strong>and</strong> comparatively denser regions of thefull network ; frequently, they have the characteristics of social circles . Applied to an interactionnetwork containing nearly 900 individuals from the American Leadership Study, we identify a number ofthese denser regions, most of which are small cliques, based on the shared interests <strong>and</strong> institutionallocations of their members . One large central circle also results, <strong>and</strong> we discuss its interpretation interms of integration of the overall network .
- 99 -Burt, Ronald S . (Sociology, U . of California-Berkeley) "Cohesion versus Structural Equivalence as a Basisfor Network Subgroups ."Two basic approaches to network analysis are compared in terms of the network subgroups each produces .The relational approach, developing from traditional sociometry, focuses on relations between actors(individuals, groups, or corporations) <strong>and</strong> aggregates actors connected by cohesive bonds into "cliques ."The positional approach focuses on the pattern of relations in which an actor is involved <strong>and</strong> aggregatesactors with similar patterns, i .e ., "structurally equivalent" actors, into jointly occupied positions .There are several questions that can be posed for a specific project that might lead an individual toanalyze subgroups in terms of cohesion versus structural equivalence . Here, considering a series of suchquestions, I conclude that subgroups based on structural equivalence are to be preferred to those based oncohesion . Cliques can be analyzed as a special type of jointly occupied network position . Illustrationis provided by data on the elite experts in methodological <strong>and</strong> mathematical sociology circa 1975 .Breiger, Ronald L . (Sociology, Harvard U .) <strong>and</strong> Philippa E . Pattison(U . Melbourne) "The Joint Role Structureof Two Communities' Elites"Blockmodel analysis offers a perspective for developing operational theories of role interlock acrossmultiple networks . We identify precisely those features of role interlock that are shared by the elitesof two small cities . This joint role structure is then interpreted with the aid of an algebraic modelthat we formulate on the basis of Granovetter's (1973) "strength of weak ties" argument . Our discussionillustrates the operationalization of substantive <strong>and</strong> theoretical concepts in the form of idealized rolestructures, <strong>and</strong> their application via blockmodel analysis to observed network data .Holl<strong>and</strong>, Paul W. (Educational Testing Service) <strong>and</strong> Samuel Leinhardt (SUPA, Carnegie-Mellon) "An OmnibusTest for <strong>Social</strong> Structure Using Triads ."A general or omnibus test of structure in social network data is proposed . The test exploits all ofthe information contained in the triad census . Analogous to the classical F-test for contrasts amongmeans, the proposed test involves finding a weighting vector which maximizes a test statistic, T 2 (max),in the context of an empirical data matrix <strong>and</strong> then determining whether this quantity is statisticallysignificant by reference to a table of the chi-square distribution . An insignificant value of T 2 (max)implies that the structure of the network data matrix is r<strong>and</strong>om, <strong>and</strong>, therefore, that the search forrecognizable or substantively meaningful pattern in the data may be subject to artifactual discoveries .Empirical results are presented which indicate that, of the networks commonly studied by social researchers,some have r<strong>and</strong>om structure, others have nonr<strong>and</strong>om structure <strong>and</strong> exhibit strong indications of transitivity<strong>and</strong> still others, with strong indications of nonr<strong>and</strong>om structure, do not exhibit strong indications oftransitivity .