Skip to main content
Search form
  • 00:01

    [MUSIC PLAYING][Studying Public Opinion of Immigration on TwitterUsing Network Analysis]

  • 00:10

    JIM WITTE: My name is Jim Witte and Iam a professor of sociology and directorof the Institute for Immigration Researchand the Center for Social ScienceResearch at George Mason University in Fairfax,Virginia.[Jim Witte, Professor of Sociology,George Mason University] What I'd like to talkabout today is some work we've beendoing on the use of social media as data for social scienceresearch.

  • 00:31

    JIM WITTE [continued]: I think pretty much everyone is awarethat the internet has changed justabout every way we live, everything we do.And just like the Industrial Revolutionin the 19th century transformed Western society,the internet has done the same and, in part, global society.It's affected everything.And that extends to politics and government.

  • 00:55

    JIM WITTE [continued]: If we think about the weapon of choice or tool of choice,whatever you want to think about it, in terms of social media,it's Twitter.And I think one of the things we have to realizeis the internet's been around since the late '60s,but it was really in the early 2000s

  • 01:16

    JIM WITTE [continued]: when the internet became social, whatpeople refer to as Web 2.0.And earlier parts of the internetwere about connecting devices to one another.Web 2.0 connects people to one another.It's inherently social.[What are some ways social media has been used as a politicaltool?]

  • 01:37

    JIM WITTE [continued]: Yes.From Barack Obama's first use of Twitter as a political toolin the 2007 election right on through todaywhen we have President Trump's morning tweets,Twitter really has played a key role in American politics.And as we see in the recent election,the social media is not always for good.

  • 01:60

    JIM WITTE [continued]: It can also be used for a variety of reasons,as we've seen the allegations of Russian meddlingin the 2016 election.And so clearly, the way people use social media in politicsis important for social science.It helps us understand the events that are going on.But in another way, and this is what first attracted meto the analysis of social media, there's

  • 02:22

    JIM WITTE [continued]: important methodological advantages for those of uswho use it in our research.Most social behavior is ephemeral.It disappears.And if we're trying to study it, wehave to figure out ways to capture it.And even if it's captured through notes or observations,people can differ in their interpretationand there's no way to really get back to the original behavior.

  • 02:44

    JIM WITTE [continued]: What we do online, for better or worse,leaves electronic footprints.And this is where we see we can use those footprints in tryingto understand social behavior.And so the case study I'd like to talk to youabout today is some of our use of tweets about immigrationand how we can use those to understand the way

  • 03:05

    JIM WITTE [continued]: the political discourse about this important topicis happening in America today.[How would you explain network analysis?]Before we start talking about our case study,I'd like to mention a few things about a network perspective,which is slightly different than the way people havetraditionally looked at social behavior.

  • 03:28

    JIM WITTE [continued]: Traditionally in the social sciences,we focused on individuals or organizationsand the characteristics of those individuals and organizationsand how they interact with one another.In a network perspective, what we focus onis the relationships between people, the tiesand links between people.

  • 03:48

    JIM WITTE [continued]: And one way to think about the differenceis to understand that these ties are notnecessarily hierarchical.As we see in this slide here, the network perspectiveis very different.On one hand, we have a traditional organizationalchart, very hierarchical and with well-specified linksbetween one another.

  • 04:10

    JIM WITTE [continued]: In a network graph, on the other side of the slide,we see the connections are quite varied.They're not hierarchical.You have multiple connections with different units,different people in this case.And it shows the flexibility of human life today.When we think about networks, there are two main components.We talk about the vertices and the edges.

  • 04:32

    JIM WITTE [continued]: The vertices are the nodes.They could be the agents, the entities, individuals, groups.And the edges are the ties between them.These are the things that link individuals.And these are the relationships we want to look at.We can think about these ties as being of different types.We won't go into a lot of detail,but they can be directed, someone follows me.

  • 04:56

    JIM WITTE [continued]: And so that's a unidirectional tie. [unidirectional] Or theycan be bidirectional, where we follow one another.[bidirectional]Or we may not know the direction at all.They can just be linked to one another.And so those imply different types of relationships.And then we can also think about thembeing weighted, the weight--and you can think about the thicknessof the line increasing.

  • 05:18

    JIM WITTE [continued]: And that would emphasize that the relationshipbetween these two vertices, these two edges,is more intense.[How is Twitter an example of a complex social network?]Now, we want to think a little bitabout how we would then apply this networkperspective to Twitter.

  • 05:39

    JIM WITTE [continued]: And Twitter is often referred to as a conversational microblog.It's something that's evolved.When I first got on Twitter, it waspeople just taking funny photos of their food and thingslike that.It's now become a real source of informationfor people and a way that people get messages outto different groups of followers.The vertices are the Twitter users,

  • 06:01

    JIM WITTE [continued]: that you have a Twitter handle.So for example, my Twitter handle is @JamesCWhitte.We'll talk about two other Twitter handlesthat have been in the news quite a bit latelyand that's @realDonaldTrump and @POTUS.Now, if we think about the edges, the edges thatare linking these different Twitter handles,

  • 06:23

    JIM WITTE [continued]: those are the tweets.And originally, the tweets were limited to 140 characters.In 2017, they were increased to 280 characters, where you canput more information in there.And what's very important is the content of the tweetis not limited to words.It can be a link.It can be a URL.

  • 06:43

    JIM WITTE [continued]: So really, you can put a whole movie inside a tweet.And that's one of the reasons it'sbecome so powerful for political messaging.So we have the @ users.These are the handles.These are the vertices.The tweets are the edges.And within a edge, you can refer to another user.

  • 07:05

    JIM WITTE [continued]: So this is a way to direct your attentionor direct the attention of the Twitter userstoward another person that you can mention themor you can retweet something that theyhave tweeted on their own.So it's one way to organize the conversation on Twitter.Because if you can think about, it's

  • 07:25

    JIM WITTE [continued]: fully, to begin with, completely unstructured.There are literally millions of Twitter usersblasting out these short tweets.But you can use the mentions by putting a handlein there or a retweet someone elseto tie the conversation together.And then there's another tool that's used,

  • 07:47

    JIM WITTE [continued]: which is a hashtag.So #BlackLivesMatter, for example,is very influential hashtag.And this is also a way to structure the conversation,that people can search for a particular hashtagand then hear the different voices that are speakingon that particular topic.[How have you used network analysis in your own research

  • 08:07

    JIM WITTE [continued]: on Twitter and immigration?]And now we're going to look at a few specific examplesof our work on immigration.And there, what we see is typically the waypeople analyze tweets is one of three types.You can look at a network, a collection of tweets,

  • 08:28

    JIM WITTE [continued]: based on a particular hashtag and the users whohave employed that hashtag to call attention to the messagethat they are trying to get across.Then you can also get a network based on a list of users.So you can put in a collection of Twitterhandles-- so for example, we've identified

  • 08:49

    JIM WITTE [continued]: using network statistics about 40 people who are particularlyinfluential on Twitter when it comesto discussions of immigration.And so there, what we do is we put in those 40 handlesand see how they're interacting with oneanother at any particular point in time.And then the other thing you can do, which is a little moretraditional social science, but you can then

  • 09:12

    JIM WITTE [continued]: take the content of the tweets and doan analysis of the language different usersor different clusters of users employto talk about a topic such as immigration.So now let's go on to the first example.And we started to use the collecting data on immigration

  • 09:36

    JIM WITTE [continued]: on Twitter in about 2013.And we had set up a tool, somethingcalled NodeXL, which is a way that you can automaticallyscrape the tweets and then visualize them, put theminto a graph and do network statistics to try to find outwho's connected with whom and who has the mostinfluence within a network based on the type of people

  • 09:57

    JIM WITTE [continued]: who follow them.[How did you collect the Twitter data for your analysis?]In about March of 2013, we set up our tool, NodeXL, as a wayto collect tweets automatically.And it would collect them and organize them through a script

  • 10:17

    JIM WITTE [continued]: that we had constructed to portray, visualize,the conversations around the word "immigration."And I had it set up so about every four hours,I would get a text that had picture of the graph thatwould come to my phone.And so this is two weeks before--this first slide shows what the network looks

  • 10:39

    JIM WITTE [continued]: like in the beginning of April.And so April 15th, 2013 was the day of the Boston Marathonbombing.And I got my message to my phone and the networklooked completely different.This is the afternoon of the Boston Marathon bombing.And the network structure-- it's as if you had a pool there

  • 11:00

    JIM WITTE [continued]: and someone threw a big stone in it.And the clusters scattered.On the one side of the graph, you see those little dots.Those are individual people tweeting.The size of the icon there in the graphreflects the number of followers you have.And so as you can see, that little grid there,

  • 11:22

    JIM WITTE [continued]: there are lots of individuals with not many followers whoare tweeting using the word "immigration."They're probably people who neverwould tweet about immigration.But then suddenly in the context of a bombingwhere people didn't know what-- at this point, noon that day,no one really knew what had happened.People were automatically linking it to immigration.

  • 11:45

    JIM WITTE [continued]: And so it's almost like taking the pulse of the countryor at least that segment of the country that's on Twitterand seeing how they try to make sense of something thatwas pretty senseless.And what we see in this slide hereis again, that 10 days after the Boston Marathon bombing,things returned to normal at least in Twitter conversations

  • 12:07

    JIM WITTE [continued]: about immigration, that there aren't as many people whoare suddenly inserting themselves in the conversationbecause they're linking immigrationto the Boston Marathon bombing.But instead, we have our main clustersof discussants of the topic representing really--one way you can look at this graph,

  • 12:28

    JIM WITTE [continued]: and we'll see it even more clearly in a minute,is this is a visualization of the echo chamber.These are the people who are talking to one another,but primarily to one another and not to other groups.There's another cluster-- there'sanother cluster that are having these internal conversations.Now here in this graph, what we've doneis over a longer period of time, we have pulled out--

  • 12:52

    JIM WITTE [continued]: there's some ways that we can filter out the noise,that we say OK, we're only going to look at those Twitterusers using #immigration who havea large number of followers or tweet frequentlyover the topic.And there you can see main clusters that come out.You see the progressive media.

  • 13:12

    JIM WITTE [continued]: You see a mainstream conversation.You see a reform conversation.You see institutional conservatives.And you can see the Tea Party.I think the Tea Party is a very interesting cluster therebecause there are very few-- within the Tea Partywhen they're talking about immigration,there are lots of small voices, but not

  • 13:33

    JIM WITTE [continued]: many people with a large number of followers.And one of the things--we use this at my Institute for a couple purposes.We're actually very interested in the conversation itself.But another thing we try to do is think abouthow can we get our message--our Institute focuses on the positive economic contributionsof immigrants because we think that's one

  • 13:56

    JIM WITTE [continued]: of the important aspects of immigration today in Americathat's often overlooked.And so we've been thinking about,well, how do we get these real facts, as opposedto fake news--how do we get real facts out into the Twitter feedsof people who might be in a institutional conservative

  • 14:19

    JIM WITTE [continued]: cluster or in the Tea Party cluster?And so what we do is we look at the content,the way people in these different clusterstalk about immigration.And then we think about, well, how can weuse some of that language to, in a sense,infiltrate the echo chamber to push the facts to people

  • 14:41

    JIM WITTE [continued]: who otherwise might not be open to hearing them?What we've done, as I mentioned this earlier,is we collected over the course of the fall of 2017.What we did is on a regular basis,we looked at our collections.And now we have, I think, 13 or 14 million tweets

  • 15:02

    JIM WITTE [continued]: on immigration.And we went into that collection during the fall of 2017and we said, OK, based on network analytics, whoare the people who have the most influence?Who are the people who have large number of followersand have positions in networks where the things they tweet

  • 15:24

    JIM WITTE [continued]: are most likely to get into the hands of a large numberof people and in particular to a large number of Twitterusers who themselves are influential?And so we came up with this collection of 39 people.There are two key players in there.One of them is @realDonaldTrump and the other

  • 15:45

    JIM WITTE [continued]: is @POTUS, which is his official account.So I don't know if many people know this,but the president tweets from two different accountsand the @realDonaldTrump is one that hehas had since the campaign and even before.And the @POTUS account was createdwhen he became president.And that's going to be archived through the National Archives.

  • 16:09

    JIM WITTE [continued]: There's a White House project that collects--they started with Barack Obama and now willcollect all future presidents' tweets and social mediapresence.But what we see here is on the one side of the graph,we see the real Donald Trump and both accountshave millions and millions--I think at this point, the real Donaldhas about 44 million followers and @POTUS 20-some odd million

  • 16:31

    JIM WITTE [continued]: followers.But what we see here is the cluster of influential peoplearound @realDonaldTrump tends to be peoplefrom the conservative media.And around the other side, we getsome of the more traditional mediawho are more closely aligned with the @POTUS account.And so you have these two sources of information

  • 16:56

    JIM WITTE [continued]: and different important people on Twitter around themwho then are interacting with them.The reason they're clustered together,again, is because they tend to mention or retweetthe information found in either @realDonaldTrumpor the @POTUS accounts.One interesting thing about the Twitter handle @realDonaldTrump

  • 17:19

    JIM WITTE [continued]: is that in this account, the president does notfollow very many people.He doesn't follow many people in either of his two accounts.But here, the 40 or so followers he has, most of themare either members of his family,a few conservative media folks, and oddly enough,a golf course designer, Gary Player.

  • 17:39

    JIM WITTE [continued]: Now, if we go to the other profile,the @POTUS profile, what we see hereis he also does not follow very many people,but there, it's a couple of family members and primarilygovernment agencies.So this is where he gets a lot of the information-- whenhe wants to put something about policies and programs

  • 18:02

    JIM WITTE [continued]: out into the Twitter sphere, this is what he's doing,is he's following these agencies.And of course, there's a lot of discussionabout how often the tweets that are on either of these accountsactually come from the president himselfor from other people who are tweeting on his behalf.There is one thing we know, though,

  • 18:22

    JIM WITTE [continued]: is that a large number of the tweets that come outof the @POTUS account are actuallyretweets of @realDonaldTrump.And so there's, again, a way that he uses very strategicallythese two Twitter accounts to reach some different audiencesby putting things from @realDonaldTrumpinto the @POTUS feed.

  • 18:45

    JIM WITTE [continued]: To sum up, I think the results presented in this case studycan give you a clearer idea of how social media, in this caseTwitter, can be used to study social behavioror political behavior.And in this case, we looked at Twitter as a wayto understand the conversations about immigration.

  • 19:07

    JIM WITTE [continued]: We began to look at the Boston Marathonas a case study of the permanence of conversationson Twitter, that the bombing literallyblew apart the existing networks on Twittertalking about immigration.But then they managed to re-establish themselvesvery quickly.

  • 19:27

    JIM WITTE [continued]: Interestingly, about a month after the marathon--after we had seen the networks come back together, therewas a terrorist act in London involving two soldiers whowere attacked and a very similar phenomena happened.The network conversation about immigrationfell apart and then in a matter of 10 weeks, 2 days--

  • 19:48

    JIM WITTE [continued]: I mean 10 days, 2 weeks, it pulled itself back together.So you see the enduring characterof these conversations.We also saw examples of how you can look at--using hashtags, like #immigration,to see how people--how they cluster, form echo chambers,around a particular topic and that they only

  • 20:10

    JIM WITTE [continued]: hear one another, like-minded people ratherthan being open to a wide variety of viewpointson a topic such as immigration.We saw also that there are people of particular influencein the conversation because of wherethey are in the network and the number of followers we haveand we saw that those could be examined and also

  • 20:31

    JIM WITTE [continued]: perhaps used strategically.And we're doing it, but by far, we're not the only ones.This is a common technique now not only in political circles,but in a lot of marketing and advertising,using Twitter to try to figure out who are the key playersthat you can then somehow get them to speakabout a product of yours.

  • 20:53

    JIM WITTE [continued]: And then we looked at, again, the content--within different echo chambers, different clusters,what sort of language do they use?And then finally, we examined a coupleof accounts, those of the president, @realDonaldTrumpand @POTUS.And there, what we see is the way one user can

  • 21:14

    JIM WITTE [continued]: use two accounts for very strategic purposesand really to send a message to perhaps different audiences,but then also source information from a different set of people.And that's one of the other thingsthat's often discussed with--for a Twitter user, you want as many followers as you can get,

  • 21:36

    JIM WITTE [continued]: but you also want to think about limiting the number of peopleyou follow.The president does it to an extreme.People often talk about a 2 to 1 ratio,that you follow about half as many peopleas are following you.And occasionally, you can go back and you can--it sounds heartless-- but prune the people you're following.

  • 21:59

    JIM WITTE [continued]: Because what you want to think about is the people you followare the ones who provide you with information.And again, one of the things we see from a network perspectiveis we want to reduce the noise-to-signaland we also want to think about reducing redundant information.

  • 22:20

    JIM WITTE [continued]: So if you have a group of people you'refollowing who tend to retweet the same people, each of themretweet the same people, then your Twitter feed,which is really your attention span,is being limited and cluttered with the same messageover and over again.So it's a way to diversify.So I hope this example encourages

  • 22:43

    JIM WITTE [continued]: you to think about ways that you canuse social media such as Twitter to study social behavior.And I think it has some real advantages substantivelybecause of the importance of social media in our societytoday, but then also methodologically because you'reable to capture and hold onto--what used to be vanishing data is now permanent.

  • 23:06

    JIM WITTE [continued]: [MUSIC PLAYING]


Jim Witte, PhD, Professor of Sociology and Director of the Institute for Immigration Research and the Center for Social Science Research at George Mason University in Fairfax, discusses his network analysis research on public opinion on Twitter of immigration, including some of the ways social media has been used a political tool, network analysis of social media, Twitter as an example of a complex social network, and collecting Twitter data for analysis.

Looks like you do not have access to this content.

Studying Public Opinion of Immigration on Twitter Using Network Analysis

Jim Witte, PhD, Professor of Sociology and Director of the Institute for Immigration Research and the Center for Social Science Research at George Mason University in Fairfax, discusses his network analysis research on public opinion on Twitter of immigration, including some of the ways social media has been used a political tool, network analysis of social media, Twitter as an example of a complex social network, and collecting Twitter data for analysis.

Copy and paste the following HTML into your website