Twitter is a modern public square where many voices discuss, debate and share their views. Media personalities, politicians and the public turn to social networks for real-time information and reactions to the day’s events. But compared with the U.S. public overall, which voices are represented on Twitter?

To examine this question, Pew Research Center conducted a nationally representative survey of 2,791 U.S. adult Twitter users who were willing to share their Twitter handles.¹ The design of this survey provides a unique opportunity to measure the characteristics and attitudes of Twitter users in the United States and link those observations to actual Twitter behaviors, such as how often users tweet or how many accounts they follow.

The analysis indicates that the 22% of American adults who use Twitter are representative of the broader population in certain ways, but not others. Twitter users are younger, more likely to identify as Democrats, more highly educated and have higher incomes than U.S. adults overall. Twitter users also differ from the broader population on some key social issues. For instance, Twitter users are somewhat more likely to say that immigrants strengthen rather than weaken the country and to see evidence of racial and gender-based inequalities in society. But on other subjects, the views of Twitter users are not dramatically different from those expressed by all U.S. adults.

In addition to teasing out these differences between Twitter users and the population as a whole, this analysis also highlights the sizable diversity among Twitter users themselves. The median user tweets just twice each month, but a small cohort of extremely active Twitter users posts with much greater regularity. As a result, much of the content posted by Americans on Twitter reflects a small number of authors. The 10% of users who are most active in terms of tweeting are responsible for 80% of all tweets created by U.S. users.

Individuals who are among the top 10% most active tweeters also differ from those who tweet rarely in ways that go beyond the volume of content they produce. Compared with other U.S. adults on Twitter, they are much more likely to be women and more likely to say they regularly tweet about politics. That said, there are only modest differences in many attitudes between those who tweet frequently and those who do not.

Twitter users are younger, more educated and more likely to be Democrats than general public

U.S. adult Twitter users differ in significant ways from the overall U.S. adult population. Most notably, Twitter users are much younger than the average U.S adult and are also more likely than the general public to have a college degree. The median age of adult U.S. Twitter users is 40, while the median U.S. adult is 47 years old. Put differently, the U.S. adult population is nearly equally divided between those ages 18 to 49 and those ages 50 and older. But Twitter users are nearly three times as likely to be younger than 50 (73%) as to be 50 or older (27%).

Although less pronounced than these differences in age, Twitter users also tend to have higher levels of household income and educational attainment relative to the general adult population. Some 42% of adult Twitter users have at least a bachelor’s degree – 11 percentage points higher than the overall share of the public with this level of education (31%). Similarly, the number of adult Twitter users reporting a household income above $75,000 is 9 points greater than the same figure in the general population: 41% vs. 32%. But the gender and racial or ethnic makeup of Twitter users is largely similar to the adult population as a whole.

Twitter users more likely to be Democrats

Twitter users are more likely to identify with the Democratic Party compared with U.S. adults more generally: 36% do so, compared with 30% of U.S. adults, according to a national survey of all adults conducted in November 2018. Similarly, 26% of U.S. adults identify as Republican, versus 21% of adult Twitter users. Political independents make up a similar share of the general public (27%) and Twitter users (29%).

Of course, many political independents actually lean toward one of the two major parties. Of the Americans who lean toward either party, 52% of U.S. adults identify as Democrats or lean toward the Democratic Party, while 60% of U.S. adult Twitter users say the same. Similarly, 43% of U.S. adults identify as or lean Republican, compared with 35% of adult Twitter users.

These partisan differences between Twitter users and the general public persist when looking across certain age groups. Specifically, nearly two-thirds (63%) of Twitter users ages 18 to 49 identify as Democrats or lean toward the Democratic Party, compared with the 55% of 18- to 49-year-olds who identify the same way. Among older users, these differences are similar. Some 53% of Twitter users age 50 or older identify as Democrats or lean Democratic, a figure that is somewhat higher than the 47% of U.S. adults in this age group who identify with or lean toward the Democratic Party.

In terms of political ideology, Twitter users are less likely than U.S. adults more broadly to characterize their views as very conservative. On an 11-point scale ranging from 0 (“very conservative”) to 10 (“very liberal”), 14% of Twitter users place themselves between 0 and 2, compared with 25% of the general public. At the same time, similar shares of Twitter users and U.S. adults identify as very liberal. And although Twitter users are somewhat more likely to report having voted in the 2018 midterm elections, these differences are relatively modest: 60% of Twitter users reported that they definitely voted in 2018, compared with 55% of all U.S. adults.

Twitter users have somewhat different attitudes than the general population

Twitter users as a group express distinct opinions relative to the public as a whole on some political values, particularly when it comes to views having to do with race, immigration and gender. A larger share of Twitter users – who as noted above are more likely to identify as Democrats relative to the population as a whole – say that blacks are treated less fairly than whites (64% of Twitter users vs. 54% of Americans). They are also more likely than the U.S. general public to say that immigrants strengthen the U.S. (66% vs. 57%) and that barriers exist in society that make it harder for women to get ahead (62% vs. 56%).

In other ways, the views of Twitter users differ only slightly from those of all U.S. adults. Twitter users are somewhat more likely to say that people take offensive content they see online too seriously (59% say this, compared with 54% of U.S. adults), and are somewhat less likely to report being “very attached” to their local community (12% vs. 17%).²

Those most active on Twitter differ from the rest of adult U.S. users

In addition to these differences between Twitter users and the rest of the population, there are also significant differences between the most active Twitter users (as measured by the quantity of tweets they post) and those who post less frequently.

By definition, the most active tweeters produce a large amount of content relative to the rest of the Twitter population. But the scope of these differences is profound. The median Twitter user posts just two times a month, but the most prolific 10% of Twitter users in terms of tweet volume produce a median of 138 tweets monthly. In fact, this analysis estimates that the top 10% of tweeters are responsible for 80% of the tweets created by all U.S. adults on Twitter.

The behaviors of these highly active tweeters also differ from the rest of the Twitter population in ways that go beyond tweet volume. The median user in the top 10% by tweet volume creates 138 tweets per month, “favorites” 70 posts per month, follows 456 accounts, and has 387 followers.³ By comparison, the median user in the bottom 90% of tweeters creates just two tweets per month, “favorites” one post per month, follows 74 accounts, and has 19 followers. And when asked to report how often they use the platform, fully 81% of these highly active tweeters say they do so every day; 47% of other Twitter users visit the platform with this regularity.

Members of the top 10% of tweeters also have distinct attitudes, behaviors and personal characteristics compared with those who use the platform less often. These prolific tweeters are more likely to be women: 65% are, compared with 48% of the bottom 90% of tweeters. And these most active tweeters are much more likely than others to say they post about political issues. Fully 69% of the top 10% most prolific tweeters say they have tweeted about politics, compared with 39% of Twitter users generally. And 42% say they have tweeted about politics in the last 30 days, compared with just 13% of other users.

The Twitter platform provides multiple ways to post and share content, but the top 10% of tweeters are more likely to report using automated methods that allow others to post tweets on their behalf: 25% of highly prolific tweeters have done so, compared with only 15% of Twitter users in the bottom 90%.

Despite the differences between highly active tweeters and those who are less active, other instances show these active users differ only modestly – or not at all – from the rest of the Twitter population. Although prolific tweeters report tweeting about politics with great regularity, their overall partisanship is not out of sync with other Twitter users. Overall, 61% identify as Democrats or lean toward the Democratic Party, compared to 60% among other users.

And there are only modest differences between the top 10% of tweeters and the bottom 90% in other views. Identical shares of both groups (64%) say blacks are treated less fairly than whites. But the top 10% of tweeters are somewhat more likely than the bottom 90% to say that immigrant newcomers to this country strengthen American society (70% vs. 65%), or to say there are still significant obstacles in society that make it harder for women than men to get ahead (69% vs. 62%).

This report is made possible by The Pew Charitable Trusts. It is a collaborative effort based on the input and analysis of the following individuals.

Primary researchers

Stefan Wojcik, Computational Social Scientist Adam Hughes, Computational Social Scientist         

Research team

Emma Remy, Data Science Assistant Lee Rainie, Director, Internet and Technology Research Aaron Smith, Associate Director, Research Brian Broderick, Data Science Engineer Patrick van Kessel, Senior Data Scientist Andrew Mercer, Senior Research Methodologist Onyi Lam, Computational Social Scientist Nick Hatley, Research Analyst

Editorial and graphic design

Travis Mitchell, Copy Editor Alissa Scheller, Information Graphics Designer

Communications and web publishing

Rachel Weisel, Communications Manager Shawnee Cohn, Communications Manager Hannah Klein, Communications Associate Calvin Jordan, Communications Associate Travis Mitchell, Digital Producer

In addition, the report benefited from feedback provided by the following Pew Research Center researchers: Claudia Deane, Courtney Kennedy, Carroll Doherty, Jocelyn Kiley, Alec Tyson, Brad Jones, Besheer Mohammed, Andrew Perrin, Skye Toor, Dennis Quinn and Peter Bell.

The analysis of Twitter users in this report is based on a nationally representative survey conducted from Nov. 21 to Dec. 17, 2018, among a sample of 2,791 U.S. adults ages 18 years and older who have a Twitter account and agreed to allow researchers to access that account. The margin of error for the full sample is plus or minus 3.0 percentage points.

The survey was conducted by Ipsos in English using KnowledgePanel, its nationally representative online research panel. KnowledgePanel members are recruited through probability sampling methods and include those with internet access and those who did not have internet access at the time of their recruitment (KnowledgePanel provides internet access for those who do not have it, and if needed, a device to access the internet when they join the panel). A combination of random-digit dialing (RDD) and address-based sampling (ABS) methodologies have been used to recruit panel members (in 2009 KnowledgePanel switched its sampling methodology for recruiting members from RDD to ABS). KnowledgePanel continually recruits new panel members throughout the year to offset panel attrition.

All active members of the Ipsos panel with an active Twitter account were eligible for inclusion in this study. In all, 4,829 panelists responded to the survey. Of that group, 3,649 (76%) confirmed that they used Twitter. Among confirmed Twitter users, 3,293 (90%) agreed to provide their Twitter handle and complete the survey. Next, researchers reviewed each account and removed any that were nonexistent or belonged to institutions, products or international entities. This report is based on the remaining 2,791 respondents who both completed the survey and provided a valid handle (76% of confirmed Twitter users).

The final sample of 2,791 adults was weighted using an iterative technique that matches gender, age, race, Hispanic origin, education, region, party identification, volunteerism, voter registration, and metropolitan area to the parameters of the American Trends Panel November 2018 survey of Twitter users. This weight is multiplied by an initial sampling, or base weight, that corrects for differences in the probability of selection of various segments of Ipsos’s sample as well as by a panel weight that adjusts for any biases due to nonresponse and noncoverage at the panel recruitment stage using all of the attributes described above.

Sampling errors and statistical tests of significance take into account the effect of weighting at each of these stages. In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of opinion polls.

The American Trends Panel survey methodology

The American Trends Panel (ATP), created by Pew Research Center, is a nationally representative panel of randomly selected U.S. adults. Panelists participate via self-administered web surveys. Panelists who do not have internet access at home are provided with a tablet and wireless internet connection. The panel is managed by Ipsos.

The ATP was created in 2014, with the first cohort of panelists invited to join the panel at the end of a large national landline and cellphone random-digit-dial survey that was conducted in both English and Spanish. Two additional recruitments were conducted using the same method in 2015 and 2017, respectively. Across these three surveys, a total of 19,718 adults were invited to join the ATP, of whom 9,942 agreed to participate.

In August 2018, the ATP switched from telephone to address-based recruitment. Invitations were sent to a random, address-based sample (ABS) of households selected from the U.S. Postal Service’s Delivery Sequence File. In each household, the adult with the next birthday was asked to go online to complete a survey, at the end of which they were invited to join the panel. For a random half sample of invitations, households without internet access were instructed to return a postcard. These households were contacted by telephone and sent a tablet if they agreed to participate. A total of 9,396 were invited to join the panel, and 8,778 agreed to join the panel and completed an initial profile survey.

Of the 18,720 individuals who have ever joined the ATP, 13,569 remain active panelists and continue to receive survey invitations.

The questions from the ATP that were used in this report were asked on four different waves of the panel, which are fielded roughly once a month. All panel waves were fielded in 2018. Estimates for each question are calculated using the respondents and weights that correspond to the wave in which it was asked. Some items, such as demographics, were measured at recruitment and updated periodically, in which case they do not belong to any individual wave. For these kinds of questions, estimates are based on the respondents and weights from Wave 39.

The field dates and sample sizes for each ATP wave are presented in the table below. Cumulative response rates account for nonresponse to the recruitment surveys and panel attrition in addition to nonresponse to the individual panel waves.

Weighting

The ATP data were weighted in a multistep process that begins with a base weight incorporating the respondents’ original survey selection probability and the fact that in 2014 and 2017 some respondents were subsampled for invitation to the panel. The next step in the weighting uses an iterative technique that aligns the sample to population benchmarks on the dimensions listed in the accompanying tables.

Sampling errors and statistical significance tests take into account the effect of weighting. Interviews are conducted in both English and Spanish, but the American Trends Panel’s Hispanic sample is predominantly U.S. born and English speaking.

In addition to sampling error, one should bear in mind that question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of opinion polls.

Defining the ‘engaged’ or ‘heavy’ Twitter user

Examining a small group of high-volume tweeters – the technique used in the body of this report – does not capture the broader set of activities that may define highly engaged users – such as following, liking and being followed by others. To better understand engaged users in general, and to ensure that the picture drawn by the behavior of heavy tweeters was broadly resonant with a more nuanced definition of engagement, researchers used a clustering model based on an array of behaviors to classify all respondents’ accounts into two groups: light users (who tweet less, follow less, like less and have fewer followers) and heavy users (who are more engaged on all these dimensions).

To create these two groups, researchers used a model that incorporates information about how often people tweet, how many accounts they follow, how many accounts follow them, how many tweets they “like” and how long they’ve been active on the platform. Researchers were not able to measure how frequently users logged into Twitter and passively viewed content, however. The Twitter API only provides access to posts, favorites, which accounts users follow, which accounts follow users, retweeting and quote-tweeting.

The research team used a technique called K-means clustering to identify highly active users using this broader set of measures. The method takes a set of variables and a predetermined set of clusters and classifies users into groups based on whether a given user’s behavior is most similar to the mean of one cluster or another.

The cluster model used as inputs:

• Number of posts per day
• Number of favorites per day
• Number of followers
• Number of accounts following
• Account age

Researchers took the log of each of the above variables and fit a K-means model to classify users into groups. Using this technique, they classified 60% of Twitter users as heavy users and 40% as light users. The heavy user cluster showed a higher correlation across all the input variables relative to the light users cluster. Researchers tested alternative clustering models and increased the cluster size to three, but this created an additional very small group of users without distinctive behavioral patterns.

Unsurprisingly, heavy users are more likely to report using Twitter frequently in daily life than those labeled as light users, but the top 10% of users by tweet volume still eclipse them by a substantial margin, with 61% of heavy users saying they use the site daily, compared with 33% of light users. And 81% of the top 10% of tweeters by volume say they use the platform daily.

Heavy users are more likely to use automated tools to post compared to light tweeters: 18% of heavy users say they have given permission for automated tools to post on their behalf, compared with 13% for light users. Among the top 10% of tweeters, 25% say they have given permission to post on their behalf.

Heavy users spend more of their time tweeting about politics. The study found that 22% report discussing politics on Twitter within the last 30 days, compared with just 6% of light users. For comparison, 42% of those in the top 10% of tweeters say they have tweeted about politics in the last 30 days. Indeed, 74% of those in the light user group say they never tweet about politics, compared with 46% for heavy users. Just 28% of users in the top 10% of tweeters say they never tweet about politics.

Heavy users are more likely to be Democrats and to identify as liberal than light users. 64% of heavy users identify as Democrats or lean toward the party, compared with 55% of light users. By comparison, 61% of the top 10% of tweeters by volume identify as Democrats or lean toward the Democratic Party.

Similar to top 10% tweeters, heavy users are also more likely to be women (52% compared with 46% for light users). The study also finds that 65% of users in the top 10% by tweet volume are women.