Discussion Board

FiveThirtyEight

For those interested in statistical analyses and simulation, or just sports or politics, I highly recommend checking out the website www.fivethirtyeight.com. Run by Nate Silver, the site has brought simulation and probabilistic-based analysis into the public domain. Directly inspired by Silver, for instance, is the New York Times run The Upshot blog, among other sites.


Especially regarding political and sports simulations, FiveThirtyEight is superb. Silver first gained media attention by developing PECOTA, a model that forecasts the success of MLB players. He first came to my attention in 2008, when he successfully called the state election outcomes in 49 of 50 states. This was widely covered in the media and netted him a spot at one of Time Magazine’s 100 most influential people.


Since then, Silver has stayed with his convictions despite an increasing spotlight. In 2012, Silver was roundly criticized by the right for giving Obama what they felt was an unfair chance of victory. One critic even started a competitive website called Unskewed Polls, which claimed that if polls were not skewed, they would show Romney as the favorite. In 2016 he again came under fire, this time from the left. FiveThirtyEight gave Donald Trump a 35 percent chance at winning the 2016 election. The New York Times, The Huffington Post, and other sites had Trump’s chances at 10 percent or less. Some openly accused him of keeping the race artificially tight in order to drive more web traffic to the site. Of course, in both 2012 and 2016, he had the last laugh. He was actually the first to say that his predictions were CLOSER in 2016 than in 2012 in terms of vote totals, but in 2012 he called the correct winner so no one noticed the disparity.


The strength of FiveThirtyEight is that everything is probabilistic. Instead of just calling a race, such as the 2018 midterms, everything is given a percent chance. This contribution to society alone is huge. The average person needs to be taught that things like polls involve uncertainty, and it is irresponsible to make a prediction without an accompanying statement of confidence. Probabilities aren’t as exciting, but much more accurate than a simple yes/no.


My one complaint would be that in recent years there are more and more fluff posts on the site. Worse yet, some articles rely on the exact time of bad empirics Silver originally made his name discounting. On the whole, though, FiveThirtyEight has done a great deal on popularizing simulation and statistical methods, and I highly recommend everyone take a look.

E-mail me when people leave their comments –

You need to be a member of UNH Economics Collective to add comments!

Join UNH Economics Collective