Though political theorists have emphasized the importance of political discussion in non-political spaces, past study of online political discussion has focused on primarily political websites. Using a random sample from Blogger.com, we find that 25% of all political posts are from blogs that post about politics less than 20% of the time, because the vast majority of blogs post about politics some of the time but infrequently. Far from being taboo topics in those nonpolitical blogs, political posts got slightly more comments than non-political posts in those same blogs, and the comments overwhelmingly engage the political topics of the post, mostly agreeing but frequently disagreeing as well. We argue that non-political spaces devoted primarily to personal diaries, hobbies, and other topics represent a substantial place of online political discussion and should be a site for further study.
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
The Prevalence of Political Discourse in Non-Political Blogs
1. e Prevalence of Political Discourse
in Non-Political Blogs
Sean A. Munson, Paul Resnick
School of Information, University of Michigan
2. Concerns about homophily
in online political discourse
• Political bloggers predominantly link
to like-minded bloggers (Adamic &
Glance 2005)
• High levels of agreement in comment
threads on many blogs, including
political blogs (Gilbert et al 2009)
• Challenge aversion prevalent in
preferences for political news
Adamic & Glance 2005 aggregators (Munson & Resnick 2010)
7. 3-4562'67)$
People who come together for other reasons
• may be more politically diverse
• listen and frame arguments to protect their relationships
87"19$:&;-9&/$
… but they may also avoid political discussion or disagreement,
also to protect relationships.
<3$=-'-9."$ >16&9?@$
8. signs that politics is not taboo
in non-political spaces online.
• Discussion of political candidates by Twitter users
(Tumasjan et al 2010; Diakopoulos & Shamma 2010; Conover et al 2011)
• 8% of US adults posted political content to a social
network site during 2010 midterm elections, 11%
discovered for whom their friends voted (Smith/Pew 2011)
• People say they encounter cross-cutting political
discussion online, but in non-political spaces (Wojcieszak &
Mutz 2009)
9. … but some cautions
• Facebook users underestimate their friends’ political
diversity – is this a sign they don’t talk about it?
(Goel, Mason, Watts 2010)
• Posting political, partisan messages has been cited as
one of the top reasons for unfollowing or unfriending
people on Facebook or Twitter.
(Sibuna and Walczak 2010; Kwak, Chun, Moon 2011)
10. questions for our study
• How prevalent are political blog posts on non-
political blogs?
• What is the distribution of political blog posts across
di erent categories of blogs?
• When readers of non-political blogs encounter
political posts, do they treat them as taboo, or do they
engage with the political content of the post?
12. Poll Blogger.com’s recently updated list. 23,904 blogs
6-20 January 2008
Kept only blogs that:
• had at least 5 posts,
• were written in English, and
• had existed since at least 31 August 2007.
13. Poll Blogger.com’s recently updated list. 23,904 blogs
Research team inspects & eliminates 8,861 blogs
spam blogs and blogs only partially in
English.
stick gures from XKCD
14. Poll Blogger.com’s recently updated list. 23,904 blogs
Research team inspects & eliminates 8,861 blogs
spam blogs and blogs only partially in
English.
Code posts as political / nonpolitical, 8,765 blogs
classify blogs by genre. Eliminate 2.3M posts
additional spam & partially English
blogs.
15. classifying posts as political or not
Broad de nition of political: any mention of public policy,
campaigns, and elected or appointed o cials, and did not
restrict this de nition to US politics.
16. classifying posts as political or not
Broad de nition of political: any mention of public policy,
campaigns, and elected or appointed o cials, and did not
restrict this de nition to US politics.
Research team categorized 6,691 posts as political
or not. (oversampled political posts; κ = 0.969)
17. classifying posts as political or not
Broad de nition of political: any mention of public policy,
campaigns, and elected or appointed o cials, and did not
restrict this de nition to US politics.
Research team categorized 6,691 posts as political
or not. (oversampled political posts; κ = 0.969)
Used to train multinomial naïve Bayes classi er, classify
remaining posts. (κ = 0.902)
217,727 political posts and 2,136,551 non-political posts.
18. estimating prevalence
Simple tally is problematic:
• overestimate percent political on blogs with few political
posts (more opportunities for false positives)
• Underestimate political posts on blogs with many political
posts (more opportunities for false negatives)
19. estimating prevalence
Simple tally is problematic:
• overestimate percent political on blogs with few political
posts (more opportunities for false positives)
• Underestimate political posts on blogs with many political
posts (more opportunities for false negatives)
Consider a blog that is always political:
Political posts"
False negatives (12.6%)"
Consider a blog that is never political:
Non-political posts"
False positives (0.5%)"
20. estimating prevalence
Simple tally is problematic:
• overestimate percent political on blogs with few political
posts (more opportunities for false positives)
• Underestimate political posts on blogs with many political
posts (more opportunities for false negatives)
When reporting about blogs or bins, we handle this with
revised estimates (p*):
prevalence ! (1! specificity)
p* =
sensitivity ! (1! specificity) (Zhou et al 2002)
21. categorizing blogs
Classify blogs into seven categories: diary,
hobby & fan, professional & sales, politics,
religion, civic & issue, health &wellness,
and ethnic / cultural.
At least 5 categorizations per blog,
+ determine nal category with get-
get-another-label another-label (Sheng, Provost, Ipeirotis 2008)
Check against ratings from research team
(overall κ = 0.72)
22. coding comments
244 threads classi ed by research team.
Inter-rater reliability on 56 comments on 42 blog posts.
23. coding comments
244 threads classi ed by research team.
Inter-rater reliability on 56 comments on 42 blog posts.
24. coding comments
244 threads classi ed by research team.
Inter-rater reliability on 56 comments on 42 blog posts.
45. this is a start. (and a challenge.)
ere’s a lot of political talk
happening on non-political
blogs.
46. this is a start. (and a challenge.)
ere’s a lot of political talk
happening on non-political
blogs.
Not taboo: readers engage with
this political content in replies.
47. this is a start. (and a challenge.)
ere’s a lot of political talk
happening on non-political
blogs.
Not taboo: readers engage with
this political content in replies.
?
What is the actual discourse
quality? Civility? Arguments
expressed?
How does this stack up against
other spaces?
48. e Prevalence of Political Discourse
in Non-Political Blogs
Sean A. Munson @smunson samunson@umich.edu
Paul Resnick @presnick presnick@umich.edu
balance.projects.si.umich.edu
Funded by the National Science Foundation under award IIS-0916099
and a Yahoo! Key Technical Challenge Grant.
anks to Liz Aderhold, Erica Willar, and Emily Rosengren.