Can Wikipedia Predict the Elections?

You can almost picture it: The voter picks up the voter guide — or maybe even an absentee ballot — sits down at his or her computer and gets ready to decide the crucial Senate race. First stop: Wikipedia.

Yes, the community-edited online encyclopedia is hardly the most thorough (or fair) source of information on political candidates, but it’s not a bad start. (Hey, we’ve done it, and we’re professionals.)

So what does it tell us about the crucial Senate fights? A look at traffic statistics on Wikipedia over the last 30 days (using this website, which seems authoritative enough) seems to give a little more hope to Democrats than you might expect.

Now, keep a few things in mind. Nate Silver, the nerd king of big data prediction modeling, gives the Republican Party a 74.4% chance of taking the Senate. Lesser-known third-party candidates typically don’t have Wikipedia pages, while incumbents often have had them for longer than previously unknown challengers.

And the types of voters who look up candidates on Wikipedia may not be representative of voters overall.

All that said, here’s which candidates in key Senate races were looked up most on Wikipedia over the last month.

Arkansas

Tom Cotton (R) 14,899

Mark Pryor (D) 12,327

New Hampshire

Scott Brown (R) 34,563

Jeanne Shaheen (D) 18,400

Iowa

Joni Ernst (R) 39,895

Bruce Braley (D) 15,186

Colorado

Cory Gardner (R) 20,932

Mark Udall (D) 26,274

North Carolina

Thom Tillis (R) 20,176

Kay Hagan (D) 31,058

Alaska

Dan Sullivan (R) 7,446

Mark Begich (D) 12,084

Louisiana

Bill Cassidy (R) 9,942

Mary Landrieu (D) 23,952

Kansas

Pat Roberts (R) 14,960

Greg Orman (I) 33,965

Georgia

David Perdue (R) 17,147

Michelle Nunn (D) 30,896

Kentucky

Mitch McConnell (R) 40,072

Alison Lundergan Grimes (D) 41,546

So there you have it. Overall, the searches show Democrats being searched more in six of 10 key races, Republicans in three and an independent in one.

Science, people. Place your bets now.