The following is cross-posted from the Pew Research Center. Photo credit: Flickr user Ken Lee.

Nearly two-thirds of U.S. adults who use Twitter get news on the platform, according to a recent survey. But we wanted a finer-grained understanding of how they use Twitter for news – not only whether they tweet about news and follow news organizations, but also what news topics they tweet about, and how many news media accounts they follow. We approached these questions through some exploratory research, leveraging one of social media’s greatest advantages for researchers: its openness.

In order to better understand how Americans are engaging with news on Twitter, we built a small but representative sample of 176 Twitter users from an earlier national survey of 3,212 Americans conducted by Pew Research Center in association with the John S. and James L. Knight Foundation. We then analyzed the Twitter activity of these users, with their explicit permission.

About half of the Twitter users in our sample ever tweeted about news, defined here (and in our recent social media work) as information about events and issues beyond just your friends and family. Users in the sample were more likely to send an original post than a retweet when tweeting in general, but when posting about news, the opposite was true. And while news media accounts made up a relatively small proportion of the accounts a Twitter user in our sample followed, tweets from this type of account made up a significant portion of a user’s feed. Our findings are summarized in the infographic below. 

It is important to put this exploratory research in context with past research on this topic. One way to study Twitter is to look at tweets, which allows us to say, for instance, that 86% of the Twitter conversation using the hashtag #Ferguson was, in fact, related to the events in Ferguson, Missouri, or to say that the Twitter conversation about Greek Prime Minister Alexis Tsipras was largely negative after he agreed to a new bailout deal for Greece in July. This type of analysis gives us information about the overall conversation on Twitter.

But we are also interested in the average behavior of Twitter users – not only those users who are the most active and retweeted, but also those who tweet less often. Indeed, in our sample of 176 representative American Twitter users, we found that sporadic posting was the norm: Most tweeted less than a few times a week.

This type of research allows us to also understand Twitter news behavior at the individual level – to be able to say what portion of individuals are sharing news on Twitter, rather than what share of tweets are about news.

A little more background on our research methods: We asked respondents in our national survey of 3,212 Americans, part of Pew Research Center’s American Trends Panel, if they would be willing to share their Twitter handles with us for analysis. Of the 486 self-reported Twitter users in the sample, 218, or 45%, provided their Twitter handles and permitted us to study their feeds. Of these, 19 handles were invalid and 23 were private, leaving us with 176 accounts for analysis. 

Researchers examined the tweets these 176 users sent over four randomly selected weeks – a total of 6,538 individual tweets. We automatically detected how many times each user tweeted; whether or not a tweet contained a URL and/or had a photo attached; and whether a tweet was a retweet, a reply, or an original post. Through hand coding, researchers analyzed various characteristics of the tweets to answer a few key questions: Was the tweet about current news? Did the Twitter user add his or her opinion to tweets about news? And finally, what were the topics of the tweets about news? Researchers also looked at the accounts users follow, hand coded them by type, and calculated the volume of tweets sent by the accounts users follow.

From a research perspective, our sample of Twitter users was both encouraging and disappointing. It is encouraging to us that half our respondents were willing to provide us with their Twitter handles, and encouraging to know that we had a representative sample at the user level. (We ran statistical comparisons between the group who gave their handles and all Twitter users to determine if they differed in any significant ways to rule out any bias in our sample. For the majority of qualities measured, no bias was found. The small differences we did find are detailed in the full methodology.

The disappointing aspect: Even a large survey yields a fairly small number of Twitter users. This small sample size means our estimates have relatively large margins of error (for example, for the sample of Twitter users the margin of sampling error is plus or minus 9.7 percentage points, and higher for those who tweeted about news). We also can’t take advantage of the main benefit of getting our sample of Twitter users through a survey: the ability to see how behavior on Twitter differs by demographic characteristics and survey responses. Researchers typically rely on a minimum of 100 cases in order to draw conclusions, so even a simple comparison between men and women is impossible – only 87 female respondents provided valid Twitter handles. Our small sample size of 176 Twitter accounts also prevents us from making comparisons by age, race or political ideology, as more than two groups yield even smaller subsample sizes.

Though our sample is small, we believe this analysis yields valuable insight into the news habits of American Twitter users and hope this exploratory research will provide some useful lessons to researchers navigating the complex world of social media research.

Note: The graphic was designed by Margaret Ng and Margaret Porteus. See our full methodology.