Vote

TweetCast your vote

Can Tweetcast predict who you will vote for?
Give it a try...



Clinton or Trump? TweetCast predicts which candidate you are most likely to vote for in the 2016 U.S. presidential election based on an analysis of your Twitter activity.

Our technology examines the words, hashtags, websites, and usernames in a user's tweets to predict future voting behavior. To help understand how these features are useful in distinguishing between supporters of the two candidates, we have provided a list of the most discriminating terms used by Twitter users supporting Clinton and those supporting Trump:

Trump • Pence
Clinton • Kaine
  • Terms

    • trump
    • lying
    • left
    • corrupt
    • liberal
    • islam
    • race
    • money
    • country
    • illegal
    • hillary
    • bernie
    • republicans
    • humanity
    • low
    • single
    • yall
    • rights
    • nra
    • progressive

  • Hashtags

    • #makeamericagreatagain
    • #trump2016
    • #trumppence16
    • #neverhillary
    • #boycottt
    • #crookedhillary
    • #greta
    • #trumptrain
    • #soundcloud
    • #kyrieirving
    • #imwithher
    • #demsinphilly
    • #basketofdeplorables
    • #nobillnobreak
    • #easychoice
    • #bernie2016
    • #nevertrump
    • #famousmelaniatrumpquotes
    • #michelle
    • #uniteblue

  • websites

    • hannity.com
    • dailymail.com
    • breitbart.com
    • reddit.com
    • bbc.com
    • townhall.com
    • foxnews.com
    • rt.org
    • allenbwest.com
    • glennbeck.com
    • huffingtonpost.com
    • deletealltweets.com
    • cnn.com
    • dailykos.com
    • apple.news
    • hillaryclinton.com
    • dailynewsbin.com
    • instagram.com
    • buzzfeed.com
    • moveon.org

  • users

    • @realdonaldtrump
    • @breitbartnews
    • @msnbc
    • @oreillyfactor
    • @anncoulter
    • @real_peerreview
    • @seanhannity
    • @drudge_reportvote
    • @tedcruz
    • @megynkelly
    • @hillaryclinton
    • @berniesanders
    • @huffpostpol
    • @moveon
    • @change
    • @untappd
    • @billmaher
    • @lawrence
    • @buzzfeed
    • @colbertlateshow
What if Twitter Voted? : State-by-State Predictions

We adapted Tweetcast to poll individual states. Our method samples users from each state using geographic information available on Twitter. Each users' tweets are assessed by a collection of Tweetcast models each created at a different time in the election cycle. Results obtained in this manner are displayed as Clinton's margin, in bold, next to a benchmark from FiveThirtyEight.com.


Tweetcast Prediction (Bolded)
FiveThirtyEight Polls-Only Average
  • Virginia

    • 6.11
    • 8.30
  • New Mexico

    • 7.65
    • 11.00
  • Nevada

    • 13.10
    • 4.00
  • Florida

    • -0.71
    • 3.40
  • North Carolina

    • 5.45
    • 2.50
  • Ohio

    • 2.30
    • 1.60
  • Georgia

    • -4.06
    • -3.10
  • Arizona

    • 8.73
    • -0.30
  • Rhode Island

    • 10.97
    • 19.80
  • New Hampshire

    • 16.61
    • 7.60
  • Iowa

    • 11.53
    • 0.90
  • New York

    • 30.80
    • 20.70
  • California

    • 13.36
    • 24.00
  • Vermont

    • 43.50
    • 27.20
  • Illinois

    • 11.75
    • 16.50
  • Massachusetts

    • 16.70
    • 23.30
  • Maryland

    • 1.39
    • 28.70
  • Washington

    • 28.29
    • 14.50
  • New Jersey

    • 6.65
    • 13.20
  • Oregon

    • 29.87
    • 13.60
  • Maine

    • 23.36
    • 9.70
  • Delaware

    • 3.04
    • 16.50
  • Colorado

    • 20.36
    • 6.40
  • Wisconsin

    • 19.18
    • 7.60
  • Pennsylvania

    • 13.34
    • 6.90
  • Michigan

    • 3.60
    • 8.30
  • Connecticut

    • 6.91
    • 16.40
  • Minnesota

    • 21.75
    • 7.60
  • Missouri

    • 13.19
    • -5.40
  • South Carolina

    • -0.54
    • -6.50
  • Texas

    • -6.77
    • -7.00
  • Arkansas

    • 0.08
    • -15.70
  • Louisiana

    • -8.87
    • -12.60
  • Indiana

    • 7.40
    • -7.20
  • Kentucky

    • 8.13
    • -13.30
  • Nebraska

    • 9.81
    • -16.10
  • Tennessee

    • 9.35
    • -12.30
  • Kansas

    • 8.67
    • -9.00
  • Mississippi

    • -15.27
    • -11.80
  • Alabama

    • -7.92
    • -20.30
  • West Virginia

    • 13.48
    • -21.30
  • Oklahoma

    • 3.29
    • -19.00

TweetCast is one demonstration of our broader effort to develop technology that can predict future activity and preferences based on social media analysis. It was created for the 2012 election by Shawn O'Banion and Larry Birnbaum at Northwestern University with help from the John S. and James L. Knight Foundation, and Google, and updated for 2016 by Jason Cohn with support from the John S. and James L. Knight Foundation and the National Science Foundation. Special thanks to our undergrad students Shawn Cairo, Sam Cohen, and Athif Wulandana for their work on the initial prototype of this version in our Spring 2016 joint projects class in technology and journalism.

Please send general comments or feedback to infolab@northwestern.edu.