Who Will Win the Most Medals in the 2012 Summer Olympics?

(Photo: momopeche)

Dan Johnson, an economist at Colorado College, has been predicting Olympic medal counts for years with a model that uses metrics like population count, income per capita, and home-country advantage. In the past six Olympics, his model has a correlation of 93 percent between predictions and actual medal counts, and 85 percent for gold medals.

For the Games in London this summer, Johnson predicts that the U.S. Will be the top medal winner, followed by China, Russia, then Britain — the same order they finished in the 2008 Beijing Olympics. Of course, no one wants to hear that their country won’t be at the top of the podium. But Johnson explains in a press release why such a projection might be helpful, and also why you should cheer for the underdog:

We all have expectations about how our own nation, or other nations, will perform so we attempt to quantify those  expectations, so that each nation can celebrate victory if they exceed the model’s predictions. For a small nation, winning three medals is an amazing accomplishment. For the U.S. or Germany or Russia, it’s appropriate to expect a lot more. “How much more?” That’s where the model comes in.

The Olympics are a celebration of the exceptional, and the fact that an economic model can predict medal counts so accurately simply points to the fact that there are underlying patterns that favor certain nations over others. I watch for excellence, wherever it occurs, and I cheer most loudly where it is unpredicted.


Mike B

Someone needs to decide on how the medals are weighted so one can properly score the Olympics. I think the 5-3-1 system works best for individual events with team events getting a bit more weight like 10-5-3. Either way once we have a scoring system then we can work through the UN to actually give the winner something until the next Olympic games like a Security Council veto (or veto override) or exclusive rights to arctic or antarctic mineral rights.

Due to the intense competition to host Olympic Games, hopefully the IOC will get around to expanding their franchise again to include both Spring and Fall games. There are a whole slew of games that are Olympic eligible, but simply not included in the event. Adding two more game slots would help give them exposure. For example, Olympic Tug of War.

Ben

That sounds like a terrible idea.

Andreas Moser

It won't be Iran: http://andreasmoser.wordpress.com/2011/03/01/iran-olympics-2012/

Stefan K.

Mr. Johnsons predictions might not be as good as they look on first seight:

http://leastthing.blogspot.com/2012/03/accuracy-and-skill-in-predicting.html

Ryan

Something is not really right here,

The statement that says "correlation of 93%" is meaningless, and shows a real lack of understanding statistics- is it a r = .93 or is it an r^2=.93 ? and this is red flag, perhaps showing that the author does not understand that statistics he is trying to use.

Correlation is a measure of sameness in relativity, so its hardly surprising the observed medal counts highly correlate with predicted medal counts from a model that uses the observed medal counts as the dependent variables...Why would you correlate model predicted & observed values? that shows a lack of the understanding of independence that these methods require.

Mango Punch

More likely is that Ryan doesn't understand statistics and didn't notice that the article is in layman's terms and based on a peer reviewed paper.

ryan

Clearly, Mango doesn't know statistics because I am correct on what I said.

How does aiming at paper at layman excuse methodological & statistical errors? It makes it worse.

Jennifer

Thanks for sharing, very interesting indeed. I will make a point of checking up on these predictions after the London Olympics.

Just a quick general thought: If you guys offered "flattr", that would be a nice way to contribute to the good work all of you do. I thoroughly enjoy reading your blog and would like the option to show some appreciation.

MW

How well do these models compare with, for example, an Olympic sports reporter's predictions? Without knowing how hard it is to predict the results, we can't really tell whether 93% is good or not. For example if I predicted sun rise times for the next week and had correlation coefficient of 0.93, I'd be doing very poorly.

Allan

Jamaica will win most track and field medals. Ok not quite but we will win most per capita of population for sure

wasserball

Spell check got it wrong. Will should not have been capitalized after S.

For the Games in London this summer, Johnson predicts that the U.S. Will be the top medal winner, followed by China, Russia, then Britain

Axel

i say mexico would

HawkGirl18

Team USA!!!! Woot woot!!!

kevin

i say china

David Hoover

One issue that is somewhat overlooked is where the athletes train, rather than who the represent. Several of the Caribbean athletes who excelled this year trained at American schools. OTOH, this could also reflect the income of the representative nation -- only those individuals wealthy enough from a country could train outside their own country, rather than locally.