Ucinet_Guide

UCINET5forWindowsSoftwareforSocialNetworkAnalysisUSER'SGUIDEBorgatti,EverettandFreeman1999Copyright©1999AnalyticTechnologies,Inc.Lastrevised:29April,2008UCINET5©1999AnalyticTechnologies.Allrightsreserved.AnalyticTechnologies11OhlinLaneHarvard,MA01451USAVoice:[phone]Fax:[phone]TableofContentsPreface0NotationalConventions0.1Acknowledgements0.2ProgrammingConsiderations0.3Content0.4MatrixOrientation0.5GettingStarted1Hardware1.1Installation1.2QuittingtheProgram1.3TechnicalSupport1.4CitingTheProgram1.5LicenseandLimitedWarranty1.6TheUCINETEnvironment2MenusandHelp2.1Forms2.2RunningAnAnalysis2.3TheLogFile2.4Datasets2.5ProgramOrganization2.6FileSubmenu2.7DataSubmenu2.8TransformSubmenu2.9ToolsSubmenu2.10NetworksSubmenu2.11OptionsSubmenu2.12ImportingData3RAWFiletype3.1ExcelFiletype3.2DLFiletype3.3FullMatrixFormat3.4RectangularMatrices3.5Labels3.6MultipleMatrices3.7ExternalDataFile3.8DiagonalAbsent3.9LowerhalfandUpperhalfMatrices3.10BlockmatrixFormat3.11LinkedListFormats3.12Nodelists3.12.1EdgelistFormats3.12.2EdgearrayFormat3.12.3UCINETSpreadsheetEditor3.15DataProcessing4SubgraphsandSubmatrices4.1MergingDatasets4.2PermutationsandSorts4.3TransposingandReshaping4.4Recodes4.5LinearTransformations4.6Symmetrizing4.7GeodesicDistancesandReachability4.8Aggregation4.9NormalizingandStandardizing4.10ModeChanges4.11Whereisitnow?5~0~Preface0.1NotationalConventionsUCINETismenu-drivenWindowsprogram.Thismeansyouchoosewhatyouwanttodobyselectingitemsfromamenu.Menusmaybenested,sothatchoosinganitemfromamenumaycallupasubmenuwithadditionalchoices,whichinturnmaycallupsubmenusoftheirown.Consequently,togettocertainchoices,youmayhavetoselectthroughanumberofmenusalongtheway.Torepresenttheoptionsyoumusttaketoagivenchoice,weuseanglebrackets.Forexample,torunthehierarchicalclusteringprocedure,youmustfirststartUCINET,thenclickonthetoptoolbarandpointtoTools,fromthedropdownmenuthatappearshighlightClusterandfromthesubmenuthatappearsclickonHierarchical.WewillrepresentthisseriesofchoicesasToolsClusterHierarchical0.2AcknowledgementsDozensofpeoplehavecontributedtoUCINET5forWindowsbymakingsuggestions,contributingtechnicalexpertise,findingbugs,andprovidingmoralandfinancialsupport.WeespeciallythankCharlesKadushin,DavidKrackhardt,RonRiceandLeeSailerfortheirearlyandcontinuedsupport.Wealsothankthosewhohavecontributedtechnicalexpertiseforspecificalgorithms,includingPipPattison(semigroups),KimRomney(correspondenceanalysis)andStanWasserman(p1).WearealsogratefultoBrianKneller(UniversityofGreenwich,London)fordoingmuchoftheinitialprogrammingontheWindowsinterface.Thenon-metricMDSprogramisadaptedfromUCINET3.0(MacEvoyandFreeman,1985),whichinturnwasdrawnfromtheUniversityofEdinburgh'sMINISSAprogram.ManyoftheproceduresmakeuseofroutinesfoundinNumericalRecipesinPascalbyPress,Flannery,TeukolskyandVetterling,andEISPACK(Smithet.al.1976,Springer-Verlag).0.3ProgrammingConsiderationsToparaphraseanoldsong(andreversethemeaning),UCINET5.0isbuiltforspeed,notforcomfort.OftentimesduringtheprogrammingofUCINET,wehadtochoosebetweenusingafastalgorithmthatusedalotofmemory(andthereforereducedthemaximumsizeofnetworkitcouldhandle),andaslowalgorithmthatsavedmemoryandcouldhandlemuchlargerdatasets.Inpreviousversionswetriedtostrikeabalancebetweenthetwo.Inthisversion,weusuallychosespeed.Onereasonforthisisthatitispreciselywhenworkingwithlargedatasetsthatspeedisessential:whatgoodisanalgorithmthatcanhandlethousandsofnodesifittakesdaystoexecute?Theotherreasonisthatadvancesinhardwareandoperatingsystemsoftwarecontinuallyextendtheamountofmemoryprogramscanaccess,soitseemsawasteofprogrammingtimetoworkoutwaystoeconomizeonmemory.Oneconsequenceofmenusystemsistheneedtoorganizeprogramcapabilitiesintocategoriesandsubcategoriesinawaythatislogicalandcomprehensible.Ofcourse,thishasprovedtobeimpossible.WithonlyalittleeffortonecandiscoverseveralcompetingschemesforclassifyingallthefunctionsthatUCINET5.0offers.Noneoftheschemesisperfect:eachdoesanelegantjobofclassifyingcertainitemsbutseemsforcedandarbitrarywithothers.Theschemewehavesettledonisnoexception.Thebasicideaisthatundernetworkweputtechniqueswhosereasonforbeingisfundamentallynetwork-theoretic:techniqueswhoseinterpretationisforcedwhenappliedtonon-networkdata.Anexampleofsuchatechniqueisacentralitymeasure.Incontrast,undertoolsweputtechniquesthatarefrequentlyusedbynetworkanalysts,butarealsocommonlyusedincontextshavingnothingtodowithnetworks.Multidimensionalscalingandclusteranalysisareexamplesofsuchprocedures.Inevitably,ofcourse,therearetechniquesthatareeitherdifficulttoclassifyorforsomereasonareconvenienttomisclassify.0.4ContentUCINET5.0isbasicallyare-engineeredversionofUCINETIV,andsousersfamiliarwithUCINETIVshouldeasilyadapttothenewenvironment.WehaveextendedUCINET'scapabilitiesandre-organisedtheroutinesintowhatwebelievetobemoresensiblecategories.Perhapsthemostfundamentaldesignconsiderationwehavefacedischoosingwhatcapabilitiestheprogramshouldhave.SinceFreeman'sfirstversionwasreleased,UCINEThasincorporatedadiversecollectionofnetworktechniques.Thetechniquesare