New Media and Politics; There's a Hashtag for That

I am going to talk about opinion leadership via Twitter. And we need to start with the numbers.

The Numbers

This is the growth curve for Twitter messages. In 2008 when political commentators were in awe about the spread of the Obama candidacy via social media Twitter was running 1 million messages a day. Only a year later Twitter messages numbered 50 million a day. In April 2010 it was 55 million messages a day, and the growth continues -- we have surely passed 60 million by now. That is impressive growth.

I do not believe we know how many of those messages are political because 'political' is a very slippery idea and Twitter messages are often not easy to decipher. However, the Pew Foundation researchers claim that 6% of the messages are political, and I will not quibble with their number. So 6% of 60 million messages is 3.6 million.

That is a lot of politics tweeting. But there is one additional bit of information one needs about these messages. I have collected several million of them, and this number comes from that collection. People who send messages on Twitter have followers -- followers are people who get every message the leader writes. People who use Twitter for politics have, on average, more than 1,000 followers. Multiply that by 3.6 million and you get a circulation of messages that numbers 3.6 billion. That is the reach of political messages -- 3.6 billion every day. That is, of course, 10 times the population of the United States, and begins to reach Chinese population proportions. And that is every day.

This is a big deal!

Those of us who might study Twitter and politics are not data poor.

Of course, there is the argument that these messages are somehow trivial. It would take a lot of ingenuity to make a political message out of what I had for breakfast, which seems to be the stereotypical view of Twitter messaging. Ingenuity like this tweet: "Obama ordered mayo on his corned beef on rye. Dude's not Kosher, he's Muslim, right?" Appropriate to the controversy of the day.

I can assure you that political messages on Twitter have a very different character than the personal or entertainment oriented messages that are the standard stereotypes, but that was another paper. [Boynton, March 2010]

The Two Step Flow of Communication

First in the 1940s Lazarsfeld and colleagues and then Katz and Lazarsfeld in the 1950s found that mass media to audience was not an adequate mapping of the flow of communication. To do an adequate mapping of the flow you needed to recognize the important role played by intermediaries -- opinion leaders. They demonstrated quite conclusively that a two step was needed to grasp how communication between mass media and people happened.

As political scientists became addicted to the atomism of sample surveys the two step flow became almost impossible to investigate. Sprague and Huckfeldt and then Huckfeldt and other colleagues took on the task of tracing the flow of communication with snowball sampling. That is the best we have had until now.

Twitter is the 'perfect' platform for the two step flow of communication. And because it is almost wholly public it is therefore the perfect platform for investigating the two step flow of communication in great detail.

One, you know who is following whom. That is public knowledge, and can be accessed without much difficulty.

Two, you also have a reasonably good idea about the media that the opinion leaders are calling to the attention of their followers. With considerable frequency Twitter messages about politics include a url that identifies the document to which they are referring. So, we have information on mass media, opinion leaders, and followers publicly accessible.

And we can track the messaging over time to notice how the messages and documents referred to change.

And there are lots of twists and turns making it multi-step or an elaborate network. But two step is good enough to call our attention to the reality beyond the overly simple media to audience that is the standard practice of interpretation by political scientists and news media personnel.

Opinion leaders:

1. Filter: call attention to some events and not others
2. Frame: call attention to the event in a way to lead one to think about it one way and not another

The tweets of the day are:

"There are three times as many Americans who oppose the war in Afghanistan as believe Obama is a muslim. Why isn't that important mass media?"

"Building a mosque at ground zero"

That instead of building a 12 story office building housing a community center that includes a [list] in a neighborhood of bars, off track betting, hamburger joints, a strip club, and street sales that is two blocks from ground zero and you cannot see one from the other.

It is true that 20% of Americans believe Obama is a muslim.

Given a generous interpretation of 'at' "at ground zero" is correct as well.

Both filtering and framing are inevitable; they are the human condition. It is who is doing it and how they are doing it that is important.

Filtering

In this presentation I want to focus on filtering rather than on framing. And I want to do this by comparing The Washington Post version of breaking news with the attention these events/matters received in Twitter messages.

Mass media

The Washington Post is, for the present purpose, the mass media. It aspires to an audience of the political leadership of the nation. It wants to be their first read in the morning. When it was only delivered on paper what you saw in the morning was what you got. But when it went internet, as well, it has both a standard news cycle edition delivered at 3:45 a.m. CDT and breaking news or news alerts that it delivers off cycle. Breaking news is news that they expect is so important to their audience that it needs to delivered post haste as well as in the standard news cycle edition. 'Breaking news' can be taken to be their version of what is important at the moment.

I have collected breaking news alerts for the past year. The mean per month was 36. The distribution is pictured in the figure.

Early there are 20 or fewer and then it climbed to a mean of 45 a month from November 2009 on. That was more searches through the Twitter-verse than I could handle. So I will examine Twitter messages for only some.

If we treat The Washington Post breaking news stories as important stories according to the mass media then the differential interest in those stories is the filtering we can see in action among persons using Twitter.

Streams -- there's a hashtag for that

It is possible to write 140 characters and send them off to an unknown fate -- not knowing how the message might be found or who might read it. But humans like communication, and that is interaction. So Twitter users developed conventions that facilitate interaction. One of the more important of these conventions is the hashtag. A hastag is a combination of characters preceeded by a #. The characters might be a word or words mashed together, but they also could be an abbreviation that would be easily recognized.

The first stream I captured was #hc09. It was initiated by Organizing for America, which is a project of the Democratic National Committee to stay in touch with Obama supporters. Supporters were asked to send a tweet to their senator supporting the passage of health care reform. It started in July of 2009 and continued through the passage of the legislation in April 2010. During that period there were 121,009 Twitter messages containing #hcr09. The expectation was -- if everyone used #hc09 it would be easily found and easily recognized by others. But there is no monopoly control on a hashtag. So very early there were messages that were opposed to health care reform that used #hc09 as well as the messages supporting reform. This tweet addressed to Senator Vitter

To @DavidVitter: NO! NO! NO! http://bit.ly/IlDEn #hc09 #LA #70119

illustrates how the hashtag was 'captured' by an opponent, and that the person using the hashtag believed it would be recognized by whoever read the tweet for Vitter. Redbeanqueen, the name taken by the writer, only wrote No and the hashtag. That was enough.

Not all streams are constituted by a hashtag. When they are not it may become necessary to search for more than one word or phrase. The stream about Obama's statement about the "mosque at ground zero" is an instance. A search for Obama and mosque found 47,890 Twitter messages. The search for Obama and ground zero found 27,316 messages. But there were 22,260 messages that contained Obama and mosque and ground zero. By doing the double search 5,000 messages were found to add to those found with only Obama and mosque.

The important point here is that messages are connected. People are reading them and passing them along to their followers as well as responding to them with their own reactions. As this proceeds through time it constitutes a stream of messages that is the pattern of interaction in the Twitter-verse.

Between the end of July 2009 and present I have captured 192 streams of messages that can be used to assess the similarities and differences when comparing The Washington Post's breaking news and the news interesting persons outside the beltway when they are communicating about government and politics.

The number of messages per stream is such a skewed distribution that neither mean nor a figure/picture of the distribution is helpful. So, I have divided the streams into quintiles by numbers of messages.

The first quintile is streams that range from 12 to 777 messages. The only reason for noticing a stream that has as few as 12 messages is that it was a Washington Post breaking news story. It is only in the fourth quintile that the messages reach tens of thousands, and only in the fifth quintile that they reach hundreds of thousands of messages. Pride of place goes to #teaparty that has been in 1 million messages between December 9, 2009 and August 21, 2010. A million messages is a lot of communicating. It is now 7,000+ a day. Since April there have been 650 thousand messages with a reach of 1.4 billion. Both pro and con are engaged in communicating with each other about #teaparty at a high rate.

Filtering the news

If all of the Post's breaking news stories were passed along with an equal number of Twitter messages then we could conclude there was no filtering. The Post produced the list of important stories, and Twitter messages followed.

If there are substantial differences in the Twitter messages that follow from the Post identification of a bit of news as important then we can conclude that there is filtering going on from outside the beltway.

The null hypothesis is no difference. If it is clearly wrong then we have a good basis for asserting that opinion leaders outside the beltway are busy filtering -- calling attention to some and not to others.

Most of the 192 streams were connected with Post breaking news stories, but 58 were not. So we can look at a 2 by 5 table to make the assessment.

Clearly there was filtering going on. The row of streams derived from Post breaking news stories fall into all 5 quintiles rather than all being equal. For the total set of 192 the breaking news related streams are over represented in the quintiles with fewer twitter messages and have, relatively, the fewest among streams with many messages.

What does the Washington Post think is important for its audience that people outside the beltway do not find worth producing twitter messages about?

Afghanistan: there are 7 breaking news stories that fall into the first quintile for Twitter messaging. These are stories that received fewer than 800 tweets.

Local political changes: in Washington D.C. who the actors are is very important. But most people in the country were not interested in who won the Democratic primary for governor in Alabama, for example. Five breaking news stories are about local political change and fall into the lowest quintile.

Executive changes: those actors are equally important inside the beltway. But a new head of minerals management in the US bureaucracy only produced 45 tweets. Five breaking news stories are about executive change and fall into the lowest quintile.

The rest are scattered.

What were both pushing? Here are the top five streams that were also breaking news stories of the Post.

There is quite a lot more that could be done with this, but I need to address how this is important

Conclusion

For many decades the mass media have been the monopoly suppliers of the public domain -- the 'space' in which our life as a nation is discussed. If you wanted to 'speak' in this domain you had to do it through the mass media, and not many people were invited.

What has been the advantage of the mass media? Reach. They did not have to be right next to you for the communication to happen. They could reach millions.

If I return to the numbers with which I began we have the beginning of understanding the reach of Twitter messages. In four months #teaparty was in 650 thousand messages with a reach of 1.4 billion. If there are 3.6 million political messages a day then the reach is 3.6 billion.

We can be confident that there is great redundancy in those numbers -- though no one has done the work to determine how much.

Even with great redundancy the reach of messaging with Twitter begins to equal the reach of the mass media. And as Twitter use continues to increase so does the reach.

Even at this moment the mass media no longer have a monopoly on the public domain. That just has not dawned on us, yet.

© G. R. Boynton, 2010

G. R. Boynton (2010) Politics Moves to Twitter: How Big is Big and Other Such Distributions, paper presented at the Midwest Political Science annual meeting.