Why do some countries win most of the Olympic medals?

Medals won by top 15 winners at Beijing Olympics 2008 compared to prediction of model developed by Forrest et al

Most media attention so far has been on the liveability of the athletes' village, but inevitably it will turn to the Rio Games proper and, in particular, how one's country fares in the medal count.

After allowing for various drug-related disqualifications over the succeeding four years, the official medal table for the 2012 London Olympics now shows the usual suspects; USA on top, followed by China, Great Britain, and Russia. Australia ranked eighth and New Zealand fifteenth.

Why do some countries consistently win more medals than others? With the Games starting this weekend, it's timely to revisit a piece I wrote at the time of the London Games, How can a country win more Olympic medals? (1)

I noted then that those dissatisfied with their nation's absolute medal count could always get solace from one of the many places offering alternative medal tables. For example, The Guardian's Datablog shows New Zealand came fourth on the population-adjusted medals ladder for the London Olympics. That's much better than its official fifteenth position and, perhaps equally important from a Kiwi perspective, it's better than Australia did!

But whatever consolation population-adjusted rankings might offer, the number of citizens in a country is only indirectly related to winning Olympic medals. Sure, the third and first most populous nations on earth took out the first two places in the official count in London, but the second most populous country – India – only ranked fifty-fifth.

In a model they constructed to predict the share of medals won by countries at the Beijing Olympics, academics David Forrest, Ismael Sanz and J.D. Tena found that when GDP was included, population size had a limited direct role in explaining the relative performance of countries. What seems to matter far more are factors like national wealth and a country's preparedness to direct resources to fostering elite sport.

Their model was built on six variables (since they're closely related, they call them co-variates). For each country, they measured:

• Gross domestic product
• The number of medals won at the preceding Olympics
• Government spending on sport and recreation
• If the country is the Games host city or due to host the next Games
• If the country was a member of the former Soviet bloc or is a planned economy.

They calibrated the model using the medal results for 127 countries at each Olympic games from 1992 to 2004. They then forecast the total medals each would win at the Beijing Games and subsequently compared the prediction against the actual results post-Beijing.

It's evident from the exhibit that the forecasts understated China's, Australia's and the US's medal tally somewhat and overstated Japan's. However, the authors say the R-squared between the forecast and actual medals was nevertheless an impressively strong 0.970.