The famous mathematician Paul Erdös (perhaps best known to non-mathematicians for the eponymous Erdös number - see my post "Six degrees of Kevin Bacon") reportedly once said, "A mathematician is a device for turning coffee into theorems." I had to laugh at that, because I can totally relate. However, I can't imagine that there is any true, cause-and-effect relationship between coffee consumption and academic productivity. As any scientist (or mathematician) knows, causation and correlation are two completely different concepts. Confuse these two concepts at your own peril.
There is a famous study published in the New England Journal of Medicine that demonstrated a statistically significant correlation between a country's per capita chocolate consumption and the number of Nobel laureates (winners) in that country. The study's investigator, Franz Messerli, found a correlation coefficient of +0.791 (with positive 1 being perfect correlation, so +0.791 is actually quite good) between the number of Nobel laureates per 10 million inhabitants and the chocolate consumption in kilogram per year among 22 countries. He then suggested a potential reason, stating the potential (and still theoretical) benefits of dietary flavonoids found in chocolate on learning and memory.
Incidentally, another investigator found a similar correlation between chocolate consumption and Nobel laureates even after controlling for other factors, though he also found that per capita coffee consumption did not correlate with the number of Nobel laureates. It's an interesting idea, but one that demonstrates yet again that studies of this kind (analyzing large data sets) should only be used to generate new hypotheses, not prove them. I don't think anyone would use the results of this study to suggest that a government should encourage (or even mandate) its citizens to eat more chocolate!
I don't necessarily need data to convince me that Michael Jordan was the greatest basketball player that I've ever seen play in my lifetime. Jordan made everyone around him better. Actually, that's an understatement. He made everyone around him great! And it seemed like he could take over any game that he was playing in and exercise his will to win. Jordan's Chicago Bulls drafted power forward Stacey King out of the University of Oklahoma with the 6th overall pick in the 1989 NBA Draft. King averaged 8.9 points per game (PPG) in his first season with the Bulls and was named to the NBA All-Rookie Second Team that same year. However, it was his performance on March 28, 1990 that stands out for me. King and Jordan together combined for 70 points in the Bulls' win against the Cleveland Cavaliers. King was quoted in the L.A. Times and said, "I'll always remember this as the night that Michael Jordan and I combined to score 70 points."
That sounds great, but there's a catch. The stat sheet for the game on March 28, 1990 shows that Jordan scored 69 points, while King scored just 1 point. So, to be 100% honest, they both combined to score 70 points. But that doesn't come anywhere close to telling the real story. For the real story, you have to dive a little deeper into the data. It's almost always a good idea to have someone with content expertise review the data and the conclusions that are trying to be made.
When it comes to data analytics then, remember these important points. First, correlation is not causation. Second, analyzing large data sets should only be used to generate new hypotheses, not prove them. Third, don't jump to conclusions until someone with content expertise has reviewed the data and the conclusions that are trying to be made.
No comments:
Post a Comment