FOR RELEASE OCTOBER 23, 2019
National Politics on
Twitter: Small Share of
U.S. Adults Produce
Majority of Tweets
Those who strongly disapprove of Trump produce majority of
tweets from U.S. adults that mention national politics
FOR MEDIA OR OTHER INQUIRIES:
Adam Hughes, Associate Director, Research
Nida Asheer, Communications Associate
202.419.4372
www.pewresearch.org
RECOMMENDED CITATION
Pew Research Center, October, 2019, “National Politics on Twitter:
Small Share of U.S. Adults Produce Majority of Tweets”
1
PEW RESEARCH CENTER
www.pewresearch.org
About Pew Research Center
Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes
and trends shaping the world. It does not take policy positions. The Center conducts public
opinion polling, demographic research, content analysis and other data-driven social science
research. It studies U.S. politics and policy; journalism and media; internet, science and
technology; religion and public life; Hispanic trends; global attitudes and trends; and U.S. social
and demographic trends. All of the Center’s reports are available at www.pewresearch.org. Pew
Research Center is a subsidiary of The Pew Charitable Trusts, its primary funder.
© Pew Research Center 2019
2
PEW RESEARCH CENTER
www.pewresearch.org
National Politics on Twitter: Small Share of
U.S. Adults Produce Majority of Tweets
Those who strongly disapprove of Trump produce majority of
tweets from U.S. adults that mention national politics
The social media platform Twitter plays a prominent role in how politicians, media outlets and
advocacy organizations promote their agendas and engage with political issues. Although these
entities represent a highly visible portion of the political Twitter ecosystem, less is known about
the political tweeting habits of the 22% of the American public that uses Twitter. A new Pew
Research Center analysis sheds light on this question by collecting and analyzing the tweets of a
random sample of U.S. adults with public Twitter accounts
1
over a period of one year surrounding
the 2018 midterm elections (from June 10, 2018, through June 9, 2019).
The findings of this analysis paint a nuanced picture of just how prevalent political speech is
among U.S. adults on Twitter. On one hand, 39% of users with public accounts tweeted at least
once about national politics – which includes
mentions of national politicians, institutions or
groups, as well as civic behaviors such as voting
– over the study period. On the other hand,
national politics is a relatively small element of
the total Twitter conversation among U.S.
adults. Content explicitly related to these issues
made up just 13% of all tweets analyzed over the
year studied.
And as our previous research found to be true
of Twitter content more broadly, this analysis
finds that a small share of users produce the
vast majority of content related to national
politics: 97% of tweets from U.S. adults that
mentioned national politics over the study
period came from just 10% of users.
1
Because tweets from users with private accounts are only visible to their followers, they are not included in this analysis. For more detail on
private accounts, see this post.
A small share of U.S. adults on Twitter
produce most public tweets about
national politics
Note: Tweets about national politics include those that reference
national politicians, political groups or institutions, or political
behaviors such as voting.
Source: Survey of 2,427 U.S. adult Twitter users with public
accounts conducted Nov. 21-Dec. 17, 2018. Tweets collected via
Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce
Majority of Tweets”
PEW RESEARCH CENTER
13%
Tweets about
national politics
97% of those tweets were
created by the most active
10% of users
All tweets from
U.S. adults
Tweets about other topics: 87%
3
PEW RESEARCH CENTER
www.pewresearch.org
Understanding the activity of U.S. Twitter users is challenging because the platform itself is global
in scope and is used not just by individuals but also institutions, corporations and a variety of
other entities. To address this challenge, the current project began with a large, nationally
representative survey of American adults, asking each whether they used Twitter and would be
willing to have researchers follow their activity on the platform.
A total of 2,427 respondents with public Twitter accounts participated in the survey; collectively,
they produced more than 1.1 million tweets over the course of the study period (June 10, 2018-
June 9, 2019). The Center used a guided machine learning model to classify each of these 1.1
million tweets as related to national politics or not (see sidebar and methodology for more details
on this process and the types of content that are included in our definition of political tweets).
The study design helps to illustrate which types of users are most drawn to tweet about national
politics in the first place. Most notably, those with intense views of President Donald Trump are
among the most prolific political tweeters.
Tweets from users who strongly disapprove of Trump are especially prominent: This group
generates 80% of all tweets from U.S. adults and 72% of tweets mentioning national politics.
Those who strongly approve of Trump (who produce 11% of all tweets from U.S. adults) create 25%
of tweets mentioning national politics. Taken together, strong disapprovers and strong approvers
of Trump generate 97% of all tweets mentioning national politics from U.S. adults on Twitter. In
interpreting these results, it is important to note that it was beyond the scope of this study to
classify the share of tweets that explicitly mentioned the president, or that expressed particular
views about his administration’s policies. And as noted above, the definition of political tweets
used in this analysis includes a wide range of phrases and hashtags not related to Trump.
The prevalence of tweets mentioning national politics from U.S. adults who strongly disapprove of
Trump is tied to two factors. First, this group makes up a larger share of Twitter users than of the
general public. The share of U.S. adults on Twitter who strongly disapprove of Trump (55%) is 7
percentage points higher than the share of the general public that holds this view (48%). Second,
this group is more likely to tweet about national politics than others on Twitter. Strong
disapprovers of Trump make up 55% of all U.S. adults on Twitter, but their frequency of tweeting
about national politics results in them generating 72% of all the tweets about this topic from U.S.
adults.
4
PEW RESEARCH CENTER
www.pewresearch.org
By contrast, U.S. adults who strongly approve of Trump are underrepresented on Twitter relative
to their presence in the population as a whole: 15% of Twitter users strongly approve of Trump,
compared with 29% of the general public. However, those who strongly approve of Trump are
similar to those who strongly disapprove of him in that they also are more likely than other groups
to tweet about national politics. Twitter users who strongly approve of Trump generate 25% of the
political tweets that come from U.S. adults – larger than the share of all Twitter users they
represent (15%).
15%
14
12
55
Strongly disapprove
Not so strongly
disapprove
Not so strongly
approve
Strongly approve
11%
5
3
80
25%
2
1
72
The share of all tweets
produced by those who
_____ of President Trump
The share of all tweets mentioning
national politics produced by those
who _____ of President Trump
The share of users who
_____ of President Trump
5
PEW RESEARCH CENTER
www.pewresearch.org
Those who tweet about
politics differ from
those who avoid the
topic or tweet
infrequently
Pairing a survey with an
examination of the Twitter
behaviors of U.S. adults also
allows for a comparison of the
attitudes and characteristics of
political tweeters with those of
other Twitter users. Political
tweeters – defined as those
who tweeted at least five times
in total, and at least twice
about national politics, over
the year of the study period –
are almost twice as likely as
other Twitter users to say the
people they follow on Twitter have political beliefs similar to their own (38% of political tweeters
say this, compared with around one-in-five among nonpolitical tweeters). By contrast, nonpolitical
tweeters are more likely to say they aren’t sure about the political beliefs of the people they follow
or that they follow people with a mix of political views.
And greater engagement with national politics on Twitter is tied to political behaviors that extend
beyond the social media platform. Those who tweet about politics are more likely than other users
to report engaging in a variety of political activities, such as attending rallies and contacting
elected officials.
Prior research by the Center has found that politically engaged Republicans and Democrats in the
general public are more likely than less-engaged partisans to hold a range of negative attitudes
about the opposing party. And this new analysis indicates that partisans who tweet about national
political issues tend to have colder views of the opposing political party relative to other users who
are less politically active on the site. This difference is especially pronounced among Republicans:
51% of Republicans and those who lean toward the Republican Party who tweet about national
politics give a very cold rating to Democrats, compared with about a third of Republicans on
Twitter who don’t tweet about politics.
U.S. adults who tweet about national politics more
likely to say that most of the people they follow share
their political views
% of U.S. adults with public Twitter accounts who say most of the people they
follow on Twitter have ___ political beliefs
Note: Tweets about national politics include those that reference national politicians,
political groups or institutions, or political behaviors such as voting. Political tweeters posted
at least two such tweets (and at least five tweets total) during the study period.
Source: Survey of 2,427 U.S. adult Twitter users with public accounts conducted Nov. 21-
Dec. 17, 2018. Tweets collected via Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce Majority of Tweets”
PEW RESEARCH CENTER
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
Total
Similar Different A mix of beliefs Not sure
25%
38
17
20
4%
3
4
4
37%
33
40
39
34%
25
39
37
6
PEW RESEARCH CENTER
www.pewresearch.org
Measuring whether or not tweets talk about ‘national politics’
A central challenge of identifying and analyzing political content (on Twitter or elsewhere) is that many
types of speech or online activity might be deemed “political” in some form or fashion. The general concept
of politics might encompass numerous actors at the federal, state or local levels, as well as issue-specific
concepts ranging from abortion to immigration to taxes. In an effort to create a broadly reasonable and
functional definition of what tweets are considered “political” for the purposes of this analysis, we chose to
focus on tweets that
specifically mention or engage with national politics
.
For the purposes of this analysis, a tweet is considered to discuss national politics if it mentions or
expresses support or opposition toward national politicians or elected officials, political parties, ideological
groups or political institutions, or specific political behaviors like voting. This definition of political content
excludes
tweets that are solely about state or local politics. It also does not include general discussions of
issues (even if those issues may be of national interest) unless they also include a mention of a national
politician, party, institution or organized political behavior.
Researchers identified tweets about national politics using a two-step process. First, they used a custom
text classification model to classify political tweets based on the language used in the tweet. As an
example, tweets encouraging Americans to vote in a national election or using common political hashtags
would be classified as political during this step. Second, researchers searched for mentions of the Twitter
usernames of members of Congress and other notable national politicians. All tweets mentioning these
usernames were classified as political. As a result, tweets that engage with national politicians on Twitter
are classified as political even if the content of the tweet is otherwise not obviously political. For instance, a
brief reply to a member of Congress saying “happy birthday” would be classified as being related to
national politics.
It is important to note that this method may fail to correctly classify tweets that use irony, sarcasm, memes
or images to convey feelings or ideas about national politics. In other words, this analysis captures only the
specific subset of the larger political conversation on Twitter that involves overt or explicit discussion of, or
engagement with, U.S. national politics.
Despite the bounded nature of this definition, the estimates produced by this model largely align with the
self-reported tweeting behavior of participants in the survey. For those who say they never tweet about
politics or have, but not recently, the model estimates that just 2% of their tweets mention national politics.
But among those who say they have tweeted about politics within the previous 30 days, an estimated 26%
of their tweets mentioned national politics.
7
PEW RESEARCH CENTER
www.pewresearch.org
Older Americans produce sizable share of tweets by U.S. adults
mentioning national politics
Reflecting the modest share of all Twitter users they represent, older adults on Twitter don’t
produce a great deal of the total content. But this group is much more inclined to talk about
national politics in their tweets than younger users. Those ages 65 and older produce just 10% of
all tweets from U.S. adults, but these older Twitter users contribute 33% of tweets related to
national politics. More broadly, those 50 and older produce 29% of all tweets but contribute 73%
of tweets mentioning national politics. By contrast, 18- to 29-year-olds produce 20% of all public
tweets from U.S. adults, but just 4% of political tweets.
In addition to these age-related differences, Republicans who describe themselves as conservative
and Democrats who indicate that they are liberal are more likely to tweet about national politics
than others in their parties. Researchers used an 11-point scale ranging from 0 (“very
conservative”) to 10 (“very liberal”) to measure how individuals self-identified ideologically.
Among Republicans and Republican leaners who identify as conservative (0-4) on the scale, 25%
of tweets mention national politics. But just 9% of tweets from Republicans at the midpoint or
liberal end of the scale (5-10) do the same.
65+
50-64
30-49
Ages 18-29
20%
52
18
10
The share of all tweets
produced by those who are ...
4%
23
40
33
The share of all tweets
mentioning national politics
produced by those who are ...
The share of users
who are ...
30%
43
18
8
8
PEW RESEARCH CENTER
www.pewresearch.org
Among Democrats and Democratic leaners, there is a similar pattern: 17% of tweets from
Democrats who identify as liberal (6-10) on the scale mention national politics, compared with 5%
among those who do not identify in this way (0-5).
The different types of Twitter users examined in this report
To compare the attitudes and behaviors of different types of Twitter users, researchers classified
respondents to the survey into one of three categories based on how often they tweet in general, as well as
whether they tweet about national politics.
Political tweeters
are defined as those who tweeted at least five times over the study period, with a
minimum of two of those tweets discussing national politics. About three-in-ten (31%) U.S. adults are in this
group. Political tweeters are particularly likely to be Democrats: 67% of political tweeters identify as
Democrats or lean toward the Democratic Party; 28% identify as Republicans or lean toward the GOP. The
model used in this report estimates that 13% of all tweets produced by this group are political. Among
those who say in the accompanying survey that they have tweeted about politics in the last 30 days, 78%
meet our definition of a political tweeter based on their Twitter behavior. This group also contains some of
the most active tweeters in the sample in terms of overall tweet volume.
Nonpolitical tweeters
are defined as those who tweeted at least five times over the study period, but had
no more than one tweet in that time that mentioned national politics. This group makes up 26% of U.S.
adults on Twitter. Overall, 63% of nonpolitical tweeters identify as Democrats or lean toward the
Democratic Party; 33% identify as Republicans or lean toward the GOP. Less than 1% of all tweets from this
group are political in nature.
Infrequent tweeters
are defined as those who tweeted fewer than five times in total over the year and
includes those who did not tweet at all. Infrequent tweeters make up 43% of all U.S. adults on Twitter.
Overall, 54% of infrequent tweeters identify as Democrats or lean toward the Democratic Party; 42%
identify as Republicans or lean toward the GOP. This group accounts for fewer than 1% of all tweets
produced by U.S. adults; 6% of their tweets are classified as political.
Political tweeters report visiting the site with great regularity: 73% of this group says they use the site at
least once a day. The other two groups contain a smaller share of regular visitors to the site: 48% of
nonpolitical tweeters and 33% of infrequent tweeters say that they use the site on a daily basis.
9
PEW RESEARCH CENTER
www.pewresearch.org
Political tweeters are
more likely to follow
other people who
share their political
views
Most U.S. adults on Twitter
report that they follow people
with a mix of political views
(37%) or that they don’t know
the political views of the
people they follow (34%). A
quarter (25%) say most of the
people they follow on Twitter
hold similar political views to
their own, while just 4% say
that most of the people they
follow have political views that
are different from theirs.
The above relationship – in
which a majority of users
follow either a mix of people
with different beliefs or say
they aren’t sure of the political
views of the people they follow
– holds for the three types of
Twitter users identified in this
analysis (political tweeters,
nonpolitical tweeters and
infrequent tweeters). Still, the
share of Twitter users who say they follow people with political views similar to their own is much
higher among those who tweet about national politics (38%) than it is among nonpolitical tweeters
(17%) and those who tweet infrequently (20%). In part, this dynamic is driven by the fact that
nonpolitical tweeters (39%) and those who tweet infrequently (37%) are more likely than political
tweeters (25%) to say they don’t know the political views of the people they follow.
U.S. adults who tweet about national politics are more
likely than other users to say they follow people with
similar political beliefs to their own
% of U.S. adults with public Twitter accounts who say most of the people they
follow on Twitter have ___ political beliefs
Note: Tweets about national politics include those that reference national politicians,
political groups or institutions, or political behaviors such as voting. Political tweeters posted
at least two such tweets (and at least five tweets total) during the study period.
Source: Survey of 2,427 U.S. adult Twitter users with public accounts conducted Nov. 21-
Dec. 17, 2018. Tweets collected via Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce Majority of Tweets”
PEW RESEARCH CENTER
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG DEM/LEAN DEM
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG REP/LEAN REP
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
Dem/Lean Dem
Rep/Lean Rep
Total
Similar Different A mix of beliefs Not sure
25%
15
31
38
17
20
25
10
13
45
21
27
4%
6
2
3
4
4
8
6
6
2
2
2
37%
48
32
33
40
39
48
50
48
30
35
33
34%
29
34
25
39
37
19
32
33
24
42
38
10
PEW RESEARCH CENTER
www.pewresearch.org
Democrats and Democratic leaners on Twitter are about twice as likely as Republican and
Republican-leaning users to say that most of the people they follow hold similar political views to
theirs (31% vs. 15%). A previous Pew Research Center report found that Democrats make up a
larger share of all Twitter users than of the general public overall.
Both Republicans and Democrats who tweet about national politics are more likely to say they
follow people with similar political views than those who don’t tweet about politics. However,
reflecting differences between partisans among all Twitter users generally, a larger share of
Democrats (45%) than Republicans (25%) who tweet about national politics say they mostly follow
people with political views that are similar to their own.
Political tweeters have
‘colder’ views of
members of the other
party
Partisans who tweet about
national politics are more
likely than those who do not to
give members of the other
party a “very cold” rating (a
rating of less than 25 degrees
on a 0-100 “feeling
thermometer” scale).
This phenomenon is
particularly prominent among
Republicans: 51% of
Republicans and Republican
leaners who tweet about
national politics give a very
cold rating to Democrats –
significantly higher than the
shares of Republicans who
don’t tweet about politics.
Political tweeters are colder toward the other party
than those who do not tweet about national politics
% of U.S. adults with public Twitter accounts who rate the other party as …
Note: Tweets about national politics include those that reference national politicians,
political groups or institutions, or political behaviors such as voting. Political tweeters posted
at least two such tweets (and at least five tweets total) during the study period.
Source: Survey of 2,427 U.S. adult Twitter users with public accounts conducted Nov. 21-
Dec. 17, 2018. Tweets collected via Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce Majority of Tweets”
PEW RESEARCH CENTER
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG DEM/LEAN DEM RATING REPUBLICANS
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG REP/LEAN REP RATING DEMOCRATS
Dem/Lean Dem
rating Republicans
Rep/Lean Rep
rating Democrats
38%
49
51
34
34
54
47
46
21%
22
20
22
20
19
23
24
29%
19
23
34
30
16
20
21
6%
4
4
6
8
2
4
5
3%
4
1
4
4
6
4
3
Very cold Cold Neutral Warm
Very
warm
11
PEW RESEARCH CENTER
www.pewresearch.org
Among Democrats and Democratic leaners who tweet about national politics, 54% give a very cold
rating to Republicans. Somewhat smaller shares of Democrats who don’t tweet about national
politics give an equally cold rating for members of the other party.
12
PEW RESEARCH CENTER
www.pewresearch.org
Greater levels of political activity and engagement among those who tweet
about national politics
Those who tweet about
national politics are more
likely than those who do not to
report engaging in a range of
behaviors related to political
activity and engagement.
For example, 35% of national
political tweeters say they have
contacted an elected official
within the last year, compared
with 14% of nonpolitical
tweeters and 21% of
infrequent tweeters. Political
tweeters are also more likely
than the other two groups to
say they have attended a
political rally or protest or
contributed money to a
candidate running for office in
the last year.
Within both political parties,
national political tweeters are
more likely to report engaging
in political activity than
nonpolitical tweeters.
Larger shares of Democratic
political tweeters than
Republican political tweeters
say they participated in each of the three types of political behavior measured in the survey. This
pattern is consistent with surveys of the general public conducted during the 2018 midterm
election campaign, which found higher levels of political engagement among Democrats than
Republicans.
Among U.S. adults on Twitter, tweeting about national
politics is tied to higher levels of political activity
% of U.S. adults with public Twitter accounts who have __ in the last year
Note: Tweets about national politics include those that reference national politicians,
political groups or institutions, or political behaviors such as voting. Political tweeters posted
at least two such tweets (and at least five tweets total) during the study period.
Source: Survey of 2,427 U.S. adult Twitter users with public accounts conducted Nov. 21-
Dec. 17, 2018. Tweets collected via Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce Majority of Tweets”
PEW RESEARCH CENTER
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG DEM/LEAN DEM
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG REP/LEAN REP
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
Dem/Lean Dem
Rep/Lean Rep
Total 12%
6
16
18
6
11
8
3
6
22
8
16
23%
18
27
35
14
21
28
11
16
39
15
26
13%
8
17
18
8
12
12
4
8
22
11
16
Attended
political rally
or event
Contacted
elected official
Contributed
money to
campaign
13
PEW RESEARCH CENTER
www.pewresearch.org
National political tweeters also are somewhat more likely than other Twitter users to say they
regularly follow the news. About two-thirds of national political tweeters (65%) say they follow the
news closely whether or not something important is happening, slightly larger than the shares of
infrequent tweeters (59%) and nonpolitical tweeters (53%) who say the same.
14
PEW RESEARCH CENTER
www.pewresearch.org
U.S. adults on Twitter skeptical
about the accuracy of news from
social media sites
Most Twitter users (57%) say they expect the
news they see on social media sites to be largely
inaccurate; 41% expect the news they see on
social media to be largely accurate.
However, political tweeters are more evenly
divided: 48% expect the news they see on social
media to be largely accurate, while a
comparable share (51%) expects it to be largely
inaccurate. Other Twitter users with public
accounts are more skeptical: 64% of
nonpolitical tweeters and 58% of infrequent
tweeters say they approach news stories from
social media sites with the expectation that they
will be largely inaccurate.
Differences on this question are driven largely
by views among Democrats.
Democrats and Democratic leaners who tweet
about politics are more likely than Democrats
who don’t tweet about politics and Democrats
who are infrequent tweeters to say they expect
news stories from social media sites to be
largely accurate (54% compared with 37% and
46%, respectively).
However, there is no comparable relationship
among Republicans and Republican leaners.
About two-thirds of Republicans across all
three groups of Twitter users say they expect news stories from social media sites to be largely
inaccurate.
Democrats who tweet about national
politics less skeptical of social media
news than other users
% of U.S. adults with public Twitter accounts who say
they expect news stories from social media sites to be …
Note: Tweets about national politics include those that reference
national politicians, political groups or institutions, or political
behaviors such as voting. Political tweeters posted at least two such
tweets (and at least five tweets total) during the study period.
Source: Survey of 2,427 U.S. adult Twitter users with public
accounts conducted Nov. 21-Dec. 17, 2018. Tweets collected via
Twitter API, June 10, 2018-June 9, 2019.
“National Politics on Twitter: Small Share of U.S. Adults Produce
Majority of Tweets”
PEW RESEARCH CENTER
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG DEM/LEAN DEM
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
AMONG REP/LEAN REP
Infrequent tweeters
Nonpolitical tweeters
Political tweeters
Dem/Lean Dem
Rep/Lean Rep
Largely
Inaccurate
Largely
accurate
57%
67
52
51
64
58
67
69
67
45
62
50
41
31
47
48
35
39
32
29
31
54
37
46
Total
15
PEW RESEARCH CENTER
www.pewresearch.org
Among the general public, Republicans are far more likely than Democrats to express distrust in
news organizations and rate the impact of the national news media on the country negatively.
The most prolific political tweeters have behaviors and attitudes distinct
from other U.S. adults on Twitter
As a group, political tweeters differ in numerous ways from those who tweet about nonpolitical
topics and from those who rarely tweet at all. It’s also the case that there are meaningful
differences within the group of users who tweet about national politics. The most prolific political
tweeters (defined as those who tweeted at least 10 times and mentioned national politics in at least
25% of those tweets) stand out even from other political tweeters (whose political tweets make up
a smaller share of their total output). Prolific political tweeters make up just 6% of all Twitter users
but generate 20% of all tweets and 73% of tweets mentioning national politics.
Compared with other political tweeters, even larger shares of these highly prolific users say they
largely follow others on Twitter who share their political views (51%) and that they follow the news
closely whether or not something important is happening (92%). And this group is highly active in
more traditional forms of political activity: 57% have contacted an elected official in the last year,
while 38% have made a political donation and 34% have attended a rally.
16
PEW RESEARCH CENTER
www.pewresearch.org
Acknowledgments
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
Adam Hughes, Associate Director, Research
Brad Jones, Research Associate
Alec Tyson, Senior Researcher
Emma Remy, Data Science Assistant
Aaron Smith, Director, Data Labs
Research team
Patrick van Kessel, Senior Data Scientist
Regina Widjaya, Computational Social Scientist
Stefan Wojcik, Former Computational Social Scientist
Editorial and graphic design
David Kent, Copy Editor
Alissa Scheller, Information Graphics Designer
Communications and web publishing
Nida Asheer, Communications Associate
Calvin Jordan, Communications Associate
Rachel Weisel, Senior Communications Manager
Travis Mitchell, Digital Producer
In addition, the report benefited from feedback provided by the following Pew Research Center
researchers: Claudia Deane, Jocelyn Kiley, Carroll Doherty, Bruce Drake, Elisa Shearer, and Galen
Stocking.
17
PEW RESEARCH CENTER
www.pewresearch.org
Methodology
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 follow and report on that
account. This study examines only the subset of respondents (N=2,427) who had public-facing
accounts; private accounts are excluded. The margin of error for the full sample is plus or minus
3.3 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 username and
complete the survey. Next, researchers reviewed each account
and removed any that were nonexistent or belonged to
institutions, products or international entities. Among the
remaining 2,791 respondents who both completed the survey
and provided a valid username (76% of confirmed Twitter users),
87% had public accounts. Of that set, 73% tweeted at least once
during the time period of interest, a total of 1,760 individuals.
The complete 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
Twitter survey sample
characteristics
Number of
respondents
Invited to
participate
4,829
Confirmed
Twitter users
3,649
Provided
handle
3,293
Provided valid
handle
2,791
Public account
2,427
Tweeted at
least once
1,760
Source: Survey of U.S. adult Twitter users
conducted Nov. 21-Dec. 17, 2018. Tweets
collected via Twitter API, June 10, 2018-
June 9, 2019.
PEW RESEARCH CENTER
18
PEW RESEARCH CENTER
www.pewresearch.org
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.
Determining whether tweets mention national politics
Researchers started by choosing to collect tweets for one year surrounding the 2018 U.S. midterm
elections (Nov. 6, 2018) and the dates of survey administration (Nov. 21-Dec. 17, 2018). All tweets
posted by respondents with public accounts between 12 a.m. on June 10, 2018, and 11:59 p.m. on
June 9, 2019, were included in the analysis, for a total of 1,147,969 individual tweets.
Next, researchers developed a definition of the concept of interest: national politics. Through an
iterative process, researchers arrived at the following classification instructions: Political tweets
are “defined as tweets that mention, or express support or opposition toward: national politicians,
political parties, ideological groups, or political institutions, as well as mentioning formal political
behavior.”
Researchers used a custom support vector machine text classification model – similar to the one
used to classify news articles as mentioning different topics in this report – to label tweets as
political or not. The model began with a set of keywords, hashtags and word stems related to
national politics. That list included: realdonaldtrump, barackobama, mcconnell, obama, trump,
president, republican, gop, democrat, congress, senat, politi, government, campaign, elect, vote,
voting, bluewave, bluetsunami, #maga, #resist, resistance, conservative, liberal, conservative, and
bipartisan. Researchers used regular expressions (a method of text searching) to avoid keyword
false positives (such as catching words like “#magazine” when searching the term “#maga”).
Researchers also developed a list of keywords that were likely to occur in false positives (tweets
that might appear to be about national politics, but that did not actually mention the topic). These
terms included words like celeb, entertainment, food, gossip, leisure, movie and music, among
others.
When training the classification model, researchers removed all tweets that contained both
matches to the list of political keywords and matches to the list of false positive keywords.
Researchers applied sampling weights to correct for changes in the sample’s representativeness
19
PEW RESEARCH CENTER
www.pewresearch.org
that occurred as a result of this decision. Tweets that included words from the false positive list
were included in the sample when making predictions and when evaluating the accuracy of the
model.
Researchers estimated the model for an initial set of likely positive cases. Next, they applied a rule
to override the model’s predictions based on the Twitter usernames (official, personal or
campaign) of members of Congress, automated account usernames (which individuals can use to
tag members of Congress), as well as the usernames belonging to President Donald Trump and
former President Barack Obama. Any tweet that mentioned the username of a member of
Congress, Trump or Obama met the definition of the concept described above and was
automatically classified as a positive case.
Finally, four researchers coded the same set of 1,000 tweets such that each tweet was coded twice
by different individuals. Researchers calculated the Cohen’s Kappa for agreement between the
human coders. The Cohen’s Kappa for inter-rater reliability ranged from 0.82 to 0.88 across the
pairwise comparisons. Researchers also calculated inter-rater reliability with respect to the model
predictions. The Cohen’s Kappa for these pairwise comparisons ranged from 0.80 to 0.83.
Next, researchers combined the human codes in order to generate a list of 1,000 tweets that had
been classified by two separate human coders (though not always the same two). In comparisons
with the model’s predictions, the first set of human codes had a Kappa of 0.81 and the second had
a Kappa of 0.82.
The model avoided false negatives and consistently identified true positives. Using the first set of
human codes, the model’s precision was 0.89 and recall was 0.81. Using the second set of codes,
precision was 0.87 and recall was 0.84.
Twitter data collection and prolific tweeters
The Twitter API limits how many tweets researchers can collect from individual accounts. Across
the sample of users included in this study, the research team could not collect a complete timeline
of tweets created between June 10, 2018, and June 9, 2019, for 2% of the sample (a total of 44
accounts). Even when excluding these accounts from the analysis, the results remain consistent.
20
PEW RESEARCH CENTER
www.pewresearch.org
Appendix: Demographic profiles of Twitter user types
Demographic composition of national political
tweeters and other users
% of U.S. adults with public Twitter accounts in each group who are …
Note: Tweets about national politics include those that reference national politicians,
political groups or institutions, or political behaviors such as voting. Political tweeters posted
at least two such tweets (and at least five tweets total) during the study period.
“National Politics on Twitter: Small Share of U.S. Adults Produce Majority of Tweets”
Source: Survey of U.S. adult Twitter users with public accounts conducted Nov. 21-Dec. 17,
2018. Tweets collected via Twitter API, June 10, 2018-June 9, 2019.
PEW RESEARCH CENTER
Dem/Lean Dem
Rep/Lean Rep
65+
50-64
30-49
Ages 18-29
HS or less
Some college
College grad +
Women
Men 50%
50
44
37
20
24
46
19
10
42
54
54%
46
36
34
30
37
42
16
5
33
63
52%
48
46
29
25
32
40
19
8
28
67
Infrequent
tweeters
Nonpolitical
tweeters
Political
tweeters
21
PEW RESEARCH CENTER
www.pewresearch.org
Topline questionnaire
PEW RESEARCH CENTER
SURVEY OF TWITTER USERS
TOPLINE FOR USERS WITH PUBLIC ACCOUNTS
NOVEMBER 21 – DECEMBER 17, 2018
TOTAL N=2,427
ASK ALL:
TWITTER_USE About how often do you use or visit Twitter?
Nov 21-
Dec 17
2018
35
Several times a day
14
Once a day
21
A few times a week
6
Once a week
13
A few times a month
11
Once a month or less
*
Refused
ASK ALL:
POL1DT Do you approve or disapprove of the way Donald Trump is handling his job as
President?
ASK IF ANSWERED POL1DT (POL1DT=1,2):
POL1DTSTR Do you [IF POL1DT=1: approve; IF POL1DT=2: disapprove] of the way Donald
Trump is handling his job as President…
Nov 21-
Dec 17
2018
30
NET Approve
15
Very strongly
14
Not so strongly
*
Refused
68
NET Disapprove
55
Very strongly
12
Not so strongly
*
Refused
3
Refused
ASK ALL:
NEWSIMPT Which of the following best describes you? I follow the news closely…
Nov 21-
Dec 17
2018
40
Only when something important is happening
59
Most of the time, whether or not something important is happening
1
Refused
22
PEW RESEARCH CENTER
www.pewresearch.org
ASK ALL:
SNSSKEP Which of the following best describes how you approach news stories from social
media sites, even if neither is exactly right? I expect the news I see on social
media will…
[RANDOMIZE RESPONSE LIST; CAPTURE RANDOMIZATION]
Nov 21-
Dec 17
2018
41
Largely be accurate
57
Largely be inaccurate
2
Refused
ASK ALL:
QBELIEF3 Now thinking about your friends or people you follow on Twitter:
Do most of the people you follow on Twitter have…
[RANDOMIZE ORDER OF RESPONSE OPTIONS 1 AND 2; CAPTURE
RANDOMIZATION]
Nov 21-
Dec 17
2018
25
Similar POLITICAL beliefs to you
4
Different POLITICAL beliefs from you
37
A mix of POLITICAL beliefs
34
I’m not sure about their POLITICAL beliefs
1
Refused
ASK ALL:
THERMO We'd like to get your feelings toward a number of groups in the U.S. on a
“feeling thermometer.” A rating of zero degrees means you feel as cold and
negative as possible. A rating of 100 degrees means you feel as warm and
positive as possible. You would rate the group at 50 degrees if you don’t feel
particularly positive or negative toward the group.
Please enter the “degree” or number between 0 and 100 that reflects your
feelings in the box.
[Enter the number in the box between 0 and 100 that reflects your feelings]
[RANDOMIZE ORDER OF RESPONSE OPTIONS, KEEPING A AND B TOGETHER
AND F, G, H TOGETHER; CAPTURE RANDOMIZATION]
23
PEW RESEARCH CENTER
www.pewresearch.org
Rating of
0 to 24
Rating of
25 to 49
Rating of
50
Rating of
51 to 75
Rating of
76 to 100
No
Answer
Mean
rating
a. Republicans
Nov 21-Dec 17,
2018
33
16
22
13
13
2
40
b. Democrats
Nov 21-Dec 17,
2018
19
11
24
19
25
2
53
ASK ALL:
PARTY In politics today, do you consider yourself a:
Nov 21-
Dec 17
2018
20
Republican
36
Democrat
29
Independent
12
Something else
2
No Answer
ASK IF PARTY=3,4 OR DID NOT RESPOND TO PARTY [N=863]:
PARTYLN As of today do you lean more to…
Nov 21-
Dec 17
2018
35
The Republican Party
55
The Democratic Party
10
No Answer
ASK ALL:
CIVIC_ENG_ACTYR Here’s a list of activities some people do and others do not. Please
indicate if you have done each of the following activities in the past year.
[RANDOMIZE ITEMS AND RECORD ORDER]
Yes, in the
past year
No, not in
the past
year
No
Answer
a. Attended a political rally, protest or campaign event
Nov 21-Dec 17, 2018
12
87
1
24
PEW RESEARCH CENTER
www.pewresearch.org
b. Contacted any elected official
Nov 21-Dec 17, 2018
23
76
1
c. Contributed money to a candidate running for public
office or to a group working to elect a candidate
Nov 21-Dec 17, 2018
13
86
1
PROGRAMMING NOTE: RANDOMLY ASSIGN HALF OF RESPONDENTS ‘VERY LIBERAL’ AT THE
TOP AND ‘VERY CONSERVATIVE’ AT THE BOTTOM, OTHER HALF SEES THE REVERSE, ALWAYS
KEEPING 10 AT THE TOP AND 0 AT THE BOTTOM EACH TIME.
ASK ALL:
IDEOSELF Please click where you would place YOURSELF on the scale below. [RANDOM HALF
VERY LIBERAL AT THE TOP AND VERY CONSERVATIVE AT THE BOTTOM, OTHER
HALF SEES THE REVERSE, ALWAYS KEEPING 10 AT THE TOP AND 0 AT THE
BOTTOM EACH TIME. KEEP THE SCALE THE SAME FOR IDEODEM, IDEOREP, AND
IDEOSELF.]
Nov 21-
Dec 17
2018
7
10 – Very liberal
5
9
8
8
13
7
10
6
26
5
9
4
7
3
5
2
3
1
5
0 – Very conservative
2
No Answer
ASK ALL:
POLTWEET Have you ever tweeted or retweeted about politics?
Nov 21-
Dec 17
2018
58
No
26
Yes, but not in the last 30 days
15
Yes, in the last 30 days
1
No Answer