Have you created a personal profile? Login or create a profile so that you can create alerts and save clips, playlists, and searches.
[MUSIC PLAYING][Communication Networks]
RONALD E. RICE: Hi.I'm Ronald E. Rice.I'm the Arthur N. Rupe chair in the Social Effectsof Mass Communication. [Professor Ronald E.Rice, Department Chair, University of California,Santa Barbara] I'm currently chairof the Department of Communicationand I'm also co-director of the Carsey-WolfCenter at the University of California Santa Barbara.I got interested in networks-- actually,when I was a little kid, my dad would take me downto the train station in Richmond, Virginia,and I'd always look at the tracksand how the tracks got bigger and smallerand they came together.
RONALD E. RICE [continued]: I was just always interested in that for some reason.And then I was very fortunate when I went to graduate school.Ed Rogers was there.He wrote a lot of work on communication networks.And they needed somebody to run the program called [INAUDIBLE],which was on a commercial CDC computer.I had been a systems analyst, and Iwas comfortable with computers and job control language,and so I took that job on, and I learned a lot about networks.
RONALD E. RICE [continued]: In fact, my dissertation involved network analysisof online communication data.So I have always enjoyed networksand I have pretty good training, although itcontinues to advance.It's becoming more and more sophisticated,and we'll talk a little bit about those issues.So in this tutorial, we're going to talka little bit about the concept of networks,what they consist of, the kind of data that'sassociated with them, the kind of measures that can come outof network analysis, and then some examples of some researchinvolving network analysis.
RONALD E. RICE [continued]: [Network analysis]Network analysis is the study of patterns of relationshipsbetween entities.That's a very simple definition.What do I mean by that?So patterns could be structures that endure over time.Or patterns could be how things cluster together,or particular directions of relationships.
RONALD E. RICE [continued]: Or patterns that you expect, that you predict.What are entities?Well, typically, most people are going to think about people.So what's a network?Well, it's a group of people who communicatewith each other a lot.But those entities can be anything.Those entities can be organizations.So for instance, organizations often have boards of directors.
RONALD E. RICE [continued]: Boards that report and also monitor the organizations.And there's a number of people on each of those boards,and sometimes people on one boardmaybe belong to another board.We call that an interlocking directorate.That's one form of a network relationshipbetween organizations.Usually people don't know about that.Other kinds of networks might be trade.
RONALD E. RICE [continued]: There are trade networks between companies, between nations.Transportation network.You can collect data on all the flights among all the airportsin the world, and you can see the patterns that develop,and they change over time and by season.Telecommunication networks.You can collect data on all the phone calls that are madeand see where the most calls come from and where they go.
RONALD E. RICE [continued]: You can look at relationships among families.So for instance, some of the early anthropologistswho studied networks, they considered the structureof kinship as a way of defining the culture of the tribe.So for instance, in some places when a boy reached puberty,they had to leave the village and go live with their uncle.
RONALD E. RICE [continued]: This was to avoid inbreeding in their tribes.In other places, inbreeding is actually considered valuable.It's a way of maintaining resources.So for instance, most of the European monarchswere all pretty much interrelated,because this is a way of keeping land in controlof a small number of families.Other kinds of networks could be relationships among words.
RONALD E. RICE [continued]: So that's called semantic network analysis.So words that co-occur together in textrepresent patterns of meaning that are common to that text.It might be different across other kinds of text.Links among websites.Those are also networks.And you can look at websites thatare clustered together that are morelikely to link to each other.
RONALD E. RICE [continued]: So the entities can be anything.Relationships can be friendship, advice, hate, violence,marriage.Any kind of thing that involves a relationship among entitiescan be measured, and therefore itcan be analyzed through network analysis.[Network data]What do we mean by network data?
RONALD E. RICE [continued]: So a network is a set of these entitieswithin some kind of boundary.So for instance, an alumni networkwould be all the people who graduatedfrom your program and their relationships among themselves.
RONALD E. RICE [continued]: Could be friends, or it could be all the peoplethat they co-authored with.So that's a co-citation or a co-authorship network.Citations.Any two journals that cite the same article represents a linkbetween those cited articles.And the field of bibliometrics looks at citation data.
RONALD E. RICE [continued]: Other kinds of data could be-- I could have,for instance, all people in an organization,if it was small enough.I could list them all on a surveyand I can ask each person to indicatehow much they work with each of these other people.So that would be what we call a full network roster studycapturing the entire network.
RONALD E. RICE [continued]: But we could also just ask individual people,list the three people you talk with most frequentlyand tell me something about them.That would be what's called an ego network.We have data on the links from that particular ego.But depending on how we analyze that other data,we don't really know how all the egos or individualsare related.They could be talking to three people whoare outside the organization.
RONALD E. RICE [continued]: Or they, of course, could be talking to a lot more peoplethan just those three, and you're notgoing to capture those links.Depending on how you collect the data and what the data is,then you either have individuals with ego level data,the number of links, the characteristicsof those other people.Or you have a full network matrix.
RONALD E. RICE [continued]: And the simplest matrix would be all the people and allthe people they talk to.And what the data are is how each reports howthey talk to the other person.From those two kinds of data-- ego level data and networklevel data, and there are other kinds--you can then compute different kinds of measures.
RONALD E. RICE [continued]: So for instance, at the ego level if you ask simplyname all the people that you talk with about this topic,some people might name one, some people might name six or seven.So one measure might just be links.So we might expect that the people who name more linksmight be more informed about a topicor they might have higher status.
RONALD E. RICE [continued]: We can also look at the types of content they talk about.You could say, tell me the six peopleyou communicate most with and what you talk about with them.And you might find that some people talk aboutwork related things, some people talk about social things,some people talk about family.We can measure different characteristics thenof the individual.
RONALD E. RICE [continued]: At the network level, if we have the whole roster,we can actually then measure characteristicsof the whole network.So for instance, what's the density of the network?Density is a particular measure, and itmeasures the percentage of all possible links thatactually occur.So we might look at a group of people worktogether when they're just new and a year later.
RONALD E. RICE [continued]: You might expect that the density of the networkmight actually increase over time.That as more people were talking to more people,and that could be good.But you also might design it so that people actuallytalk to fewer people.You might want density to decrease.Early on, people don't know their roles, who their supposedto relate to, who they ask.Later on, they do.
RONALD E. RICE [continued]: They don't need to talk to as many people.You actually want it to be less dense.Another measure is reciprocity.I ask all the people how much theytalk with each other person.Well, person A and person B might disagree.Person A might say, oh, I talk to person B three times a week.Person B says, oh, I talk to person A seven times a week.So those numbers differ, and we can actuallymeasure the extent to which they differ.
RONALD E. RICE [continued]: And to some extent, that can be measured by reciprocity.You might expect a high reciprocalityin a more collegial, collaborative environment,and less reciprocality in a more authority or hierarchy,or maybe just a pure trade relationship.There are other kinds of measuresthat can be computed, as well.
RONALD E. RICE [continued]: And then, within that full network datayou can actually measure network level measuresfor each individual.So for instance, in this network of people in an organization,who's the most central person?And that can be defined in a number of ways.A simple measure of centrality mightbe who's the person that the most other people talk to?
RONALD E. RICE [continued]: Or who's the person that most other people wouldhave to go through if they had to reach somebody else?That's a different kind of centrality.Centrality tends to be related to a lot of factors--knowledge, innovation, power-- for instance,longevity in the organization-- and various other factors.
RONALD E. RICE [continued]: So you collect the network data of the whole network,you compute the individual level measureof how they fit in the network, and thenyou can use that variable along with other variablesabout the individual to predict outcomes like promotion,success, satisfaction, contribution, turnover,whatever other organizational factors.Or the extent to which each individual personis influenced by other people.
RONALD E. RICE [continued]: And therefore, why some people adopt innovation or engagein a certain kind of behavior and other people don't.[Importance of networks]Networks are ways in which resourcesare allocated and obtained.And it doesn't have to be manipulative,and doesn't have to be instrumental.So for instance, friendship is a resource.
RONALD E. RICE [continued]: The kinds of people you know affect what kinds of friendshipyou have.And the kind of friendships you have affect the networkstructure that you're in.But networks are also ways of allocating resources.In many countries, the networks--are you linked to the ruling party?Are you linked to the ruling family?
RONALD E. RICE [continued]: Does your kinship structure have accessto certain kind of expertise?Are you in a community that has a lot of resourcesthat you can get to because you know people?So one of the important concepts about networksis you are who you are and you cando what you can do because of the networkthat you're part of.
RONALD E. RICE [continued]: What you know about, who you can contact, what kind of resourcesyou can get, what kind of resourcesyou can give, how easy it is for you to havethose interrelationships.How diverse your network is.Do you have different kinds of friends?Do you have different kinds of resources?So the network that you're in reallyaffects, and even creates, your life.
RONALD E. RICE [continued]: It can be changed, and it often is.And it's constantly changing.And new media expand our networksand can make them more diverse, but it can alsobe used to let us focus more in and have just morereinforcement with the people and the ideasthat we're already familiar with.Another important aspect of networksare the conceptual issue of how we actually act in the worldand how much choice and agency we have.
RONALD E. RICE [continued]: So we call that the agency structure tension.And there are different sociological and psychologicaltheories.Some argue that structure is more important.That we are heavily constrained by structureand we don't have nearly as much choice as we think we have.We Tend to do the things we do because we're limited in termsof mobility or social exposure or ideasthat are available to us.
RONALD E. RICE [continued]: Others would argue for much greater agency.That is, we are individuals who can shape our structureand shape our networks.And different people do that to different extents.So network analysis is very interestingbecause it brings together both structure and agency.For instance, in studies I've doneI've measured actual organizational structure.That is, who you report to.
RONALD E. RICE [continued]: We would call that the position that they're in.Also, physical structure, how far awayyou are from everybody.There's a lot of research on physical proximityand how that affects who you communicate withand how much you even like people.And that would be influenced by the actual distanceand number of turns you have to make,because the research is clear that distance is not linear.
RONALD E. RICE [continued]: You could also have communication proximity,the extent to which you talk to other people about work relatedmatters, or about a new technology, or whatever.And then use of other media, which would overcomea lot of these constraints.And so you can see different kinds of structuresand see the effects that they haveseparate from the other ones.And you can see whether networks change over time.
RONALD E. RICE [continued]: So some research shows that in organizations or other settingswhere people adopt one of these new media thatovercome time and space constraints,that their network changes.Sometimes power relation change, because now Ihave access to information.Before, you didn't let me have it.Or now I can provide some more support or get more support,and therefore, actually I'm healthieror my community is stronger.
RONALD E. RICE [continued]: So that structure-agency debate is inherent in network studies.So one example of this kind of studyis represented in this figure.It looks like a bunch of amoebas,but actually, it's based upon a small organizationthat I was very fortunate to be able to study.And the dots are the people who workedat this small organization.
RONALD E. RICE [continued]: And the location of the dots and the circles around the dotsare at time one.And this was before the organization implemented email.And this actually uses a variety techniques.It uses what's called multidimensional scaling aswell as hierarchical clustering.We have the matrix of who talks to whom about work relatedmatters.
RONALD E. RICE [continued]: From that, we can create a correlation matrixor some other matrix, feed it into the multidimensionalscaling program, and that aligns those dots, those people,in relation to how they interact with each otherand all other people.So the people who communicate more with each otherare closer, and the people who communicate less with othersare farther away.
RONALD E. RICE [continued]: We can also use hierarchical clustering,which talks about the extent to which those people are actuallyclose to each other.And what you can see there, it saysthere's three groups that are clustered together,and then there's a couple people who are in between.They don't really strongly belong to one group or another.So that's the picture of what the network lookedlike at time one before the email system was implemented.
RONALD E. RICE [continued]: Now the dots are a year later.The black dots are those who adopted and were usingthe email system, and the clear dotsare the people who were still not using the email system.And what you can see is-- without lookingat any statistics-- that people whoare closer together and in one or two of those groupswere more likely to adopt the email system,because they communicated with each other more early on,and therefore this communication system was more useful to themand they were more likely to all adopt it.
RONALD E. RICE [continued]: In another group where there are fewer peopleand they didn't quite communicateas much with each other, but they didn't communicate muchwith anybody else, they didn't adopt, because they didn'tfeel there was a need to adopt.So here's an example of collecting data, analyzingthrough network analysis, using it to predict an outcome,and then also using that to help reinforce a theory about whypeople adopt a communication medium.
RONALD E. RICE [continued]: So there are other interesting aspects about network data,and I just want to emphasize it's not-- I mean,the data is important and there area lot of issues to understand and to analyze.But the data represent what you wantto find out about the relationshipsamong these entities, whether people or organizationsor words.So when I talk about kinds of data,it really means ways of finding outabout this social environment.
RONALD E. RICE [continued]: So certainly, in the era of online communicationand now what's called big data, thereare ways of collecting huge amounts of network data.All the relationships among all the peoplethat you link on your Facebook page,or all the people who sent mails to each other.For instance, in the Enron scandal,as part of the court proceedings they actuallygot the computer record of all the email messages thatwere sent among everybody, and that data setis publicly available.
RONALD E. RICE [continued]: And you can actually use that to seewho actually knew about what kindsof things because who they shared information with.You didn't have to ask anybody.That's actually a record of the real behavior.And in my dissertation in 1982, I actually looked at two yearsworth of communication among about 870 people communicatedthrough the nationwide EIES, E-I-E-S, network.
RONALD E. RICE [continued]: And I did a network analysis over time of that.So I could look at the changes in structure over timedepending upon which group they werein and whether the groups changed or not.So big data-- that is, collecting online data--is a big development in survey and other kinds research,but particularly network data.We can collect data we never could have collected beforeand we can analyze huge amounts and lookat very, very subtle patterns that we couldn't see before.
RONALD E. RICE [continued]: [Organizational communication]One of the areas where networks are particularlyrelevant-- network data, network analysis--is in organization and communication.Because after all, organization and communicationis a patterned over time set of interactionsamong members of a system.
RONALD E. RICE [continued]: And you can have small local networks,the people you work with.You could have unit based relations.That is, what are the formal reporting relationshipsacross units?You can have upward communicationwith supervisors and co-workers and more senior executives.You can have relationships across people,across organizations.
RONALD E. RICE [continued]: So in these studies of organizations,network analysis is actually a very, very powerful tooland appears more and more in research.And there's a lot of consulting companies that look at this.That's one factor maybe affectingwhether the communication is effective and satisfyingto people.Maybe the people who should know aren't finding out.
RONALD E. RICE [continued]: And you can look at that, because some people areleft out of the network.And then you need to find out well, why is that?Did they not know they should be part of this network?Do they have certain behavioral traits?Do people not trust them?It's also the case that the flow of informationaffects certain outcomes in organizations.If everybody talks to the same people all the time,then you can't innovate.
RONALD E. RICE [continued]: So Ron Burton and others have shownthat if you are a kind of person whois on a path where you're exposedto information from different people,but those people don't actually communicate among themselves,you are more likely to come up with new ideas.And you are more likely to actually be promotedand proceed through the organization faster,because you have unique informationthat other people don't have.
RONALD E. RICE [continued]: And that can either be manipulated and usedin negative ways, or it can be used to bring peopletogether and foster new ideas.So a lot of people looking at networks in organizationsfor all sorts of reasons.And a lot of the really interesting network datacome from organizations, as well.[Key points-- Networks and network analysis these arerelationships among anything(not just people).Measures at individual and group levels.Importance of networks.Conceptual issues of what networks represent.]So some of the topics that we discussed in this tutorialis a definition of networks and network analysis.
RONALD E. RICE [continued]: The kinds of data that can be collected.The kinds of entities that can be considered.And it's very important to keep in mind it's not just people.It could be relationships among anything.The kinds of measures that you canderive from that at both the individual level and the grouplevel, but also at other levels--at dyads, triads, larger systems.
RONALD E. RICE [continued]: The importance of networks in predicting and explainingbehavior, particularly in organizations.And also some conceptual issues about what networks reallyrepresent and how they operate in our livesand how we operate them.[MUSIC PLAYING]
View Segments Segment :
Professor Ronald Rice outlines various types of networks, both social and virtual, and describes how to analyze data from them.
Professor Ronald Rice outlines various types of networks, both social and virtual, and describes how to analyze data from them.