Huffpost Media

Featuring fresh takes and real-time analysis from HuffPost's signature lineup of contributors

Maria Ehsan Headshot

Political Junkies' Newest Fix: Election Modeling

Posted: Updated:
Flickr: Randy Stewart
Flickr: Randy Stewart

One of the most followed topics among election watchers is election modeling via statistics. Believe it or not, this esoteric topic, which usually isn't the most exciting or interesting, Moneyball notwithstanding, fires up as one of the hottest topics in blogs and articles before an election.

The numerous blogs that review elections start to increase postings on poll statistics this time of the year and really get followers engaged. Even if you're a political news follower who is not knowledgeable about statistics, start reading these blogs and once you follow them you'll be hooked. The lure of a model that can provide probabilities of various election outcomes is downright exciting to wonks.

Let's start with the A #1 most popular election modeling blog.

With legions of addicts, began in 2008 when baseball statistician Nate Silver decided to apply his expertise to analyzing presidential election polls. A few years ago FiveThirtyEight got so popular that it was purchased by the New York Times and is now a regular part of the newspaper's election coverage.

The blog looks at polls and their historical accuracy and analyzes their patterns and error rates over time to predict who's going to win the presidential election as well as congressional and gubernatorial seats.

FiveThirtyEight doesn't tell you by how much someone is going to win. Instead, what it focuses on is whether or not that candidate will win. For example, rather than "Obama is predicted to win by 75 electoral seats," you may read that the president has a 75 percent chance of winning the 270 votes needed. Or that a Senate candidate has a 60 percent chance of winning. The blog takes a specific outcome of 270 electoral votes, or 50 percent +1 in other elections, and assigns a probability to it, rather than predict a specific outcome based on the polls.

Another popular blog with similar analysis is the Monkey Cage. It's done by a group of political science professors from D.C., mostly men, who work at GWU, Georgetown University, and others, and share research, including election models. Another popular one is Princeton Election Consortium.

'Tis the Season

Among others, RealClearPolitics and Public Policy Polling conduct a lot of good polls and analysis, and they produce volumes of data. It's polling season these days, providing fodder for a lot of statistical analysis. When polling becomes popular, these blogs extrapolate the polls into further statistical analysis. And they rate polls to determine how accurate they are.


If you like following the numbers at all you can get sucked into this. Every day you'll be checking the sites, seeing how candidates are doing, trying to find out what the latest polls show, and wondering what Nate Silver will deduce from all the data.

But it isn't just the candidates and their policy positions that matter; it's how well they run their campaigns. This is reflected in polling and these sites' analysis, which graphs newsworthy events and developments over time, going into depth to really show what makes candidates win.

However, these sites will not treat all polls equally. They'll note that certain polls consistently lean in a certain direction or are consistently inaccurate or accurate.


Tracking the stats can turn into a game for readers. Each blog that covers this is like a team. As in sports, you could follow each team's success (accuracy) and analysis. And at the end of the election you could determine who was the best "team" based on how well they predicted the outcome. The reigning king so far is Nate, though I'm sure he'll welcome some competition.

As for accuracy, correctly predicted the 2008 presidential outcome in 49 states. He got Indiana wrong -- where yours truly spent the last two weeks in a rural county helping push Barack Obama over the finish line.

If only they had polled me.