Day 17: Closing Ceremonies (Part 1)

Earlier in the week I posted this conversation about the medal count that I had with my German Friend. I would like, as part of my own closing ceremonies, to make a few remarks on the medal count.

The U.S. has won the total medal count with 37, followed by the Germans with 30 and the Canadians with 26. In terms of gold medals, the Canadians come out on top with 14, followed by the Germans with 10 and the U.S. with 9. Now, many favor the gold medal count as the preferred measure for Olympic success. And, I will not deny that this measure has merit. For instance, it acknowledges that silver and bronze medals are not the same as gold medals. It is not, however, the only one, and my reasons for problematizing it are not, strictly speaking, biased (at least in terms of patriotism). I would like to suggest a few other measures and offer reasons for their value over the "Gold only" measure.

1. Total Medals. We should recall that until very recently both summer and winter games were contested in the same calendar year. I would thus argue that for total medals we should combine this year's Winter Olympics with the 2008 Summer Olympics (we could, or course, wait until the 2012 London Summer Games). Doing that we come up with the following totals:

   U.S.A.: 45 / 53 / 49 // 147
   China: 56 / 23 / 32 // 111
   Russia: 26 / 26 / 3 // 87
   Germany: 26 / 23 / 22 // 71
   South Korea: 19 / 16 / 10 // 45
   Canada: 17 / 16 / 11 // 44

   We see that China leads overall in gold medals, but the U.S.A has a clear lead in overall medals. I would also add that no other nation is as balanced in terms of representation in the medals for both the Summer and Winter games. Russia and Germany are close, but nowhere near as dominant in both Olympics.


2. Total Points. Another measure, and one that gives full due to all medal winners, which the "Gold only" argument does not, is to award values to medals and combine those totals (i.e., 3 points for gold, two for silver, and one for bronze). For this year's Winter Olympics it works out this way:

   U.S.A.: 68 points
   Germany: 63 points
   Canada: 58 points

   
   For both the 2008 Summer and 2010 Winter Olympics we have the following totals

   U.S.A.: 290 points
   China: 246 points
   Russia: 165 points
   Germany: 146 points
   South Korea: 99 points
   Canada: 94 points

While, for personal reasons, I might prefer the total medal count translated into points, the larger point is that there are multiple ways of assessing the games. I prefer this last way of accounting because it counts silver and bronze medals without making them equal to gold medals, which addresses one of the reasons I feel the "Gold only" assessment method is preferred. It also highlights consistency and excellence across the entire Olympic spectrum. Clearly, one could use another measure when and where appropriate (e.g. if one wanted to make geographical claims). I make the following range of assessments available to suggest the need for alternatives and to contextualize these now completed 2010 Winter Olympic Games.

Day 8: A Stat(e) of Time

The hour is late. I have put this off until now. Time is of the essence. And upon time and the statistical analysis of time I will briefly turn.

Enough.

I am still on statistics in this post (as I was yesterday - UPDATE: John Shuster, skip of the US Curling Team, was benched).

I had a statistics professor (in psychology) who once used the Olympic sport of Luge to complicate the issue of statistics in research. In stats there is the concept of statistical significance. It is the way, for instance, that correlations are made. Is the number of times one thing correlates to another significant? Many things inevitably "correlate;" statistical significance is used to separate coincidences (things that happen by chance) from correlations (things are "actually" and consistently and predictably related).

Statistical significance, however, does not always manifest itself practically. In luge, for example, the differences in times between first and second and third and forth and fifth places rarely meet the conventions of statistical significance. Once you have to go out the thousandths of a second to determine the difference between first and second place, my professor would say, you are pushing statistical significance (of course there will be a difference in the times of each run, but are those differences significant?). That is, the differences between the times are not significant enough statistically to make any definite assertion about the relative excellence of individual lugers. To say it another way, a difference in race times that is one-one-thousandths of second doesn't "mean" anything statistically because it looks like simple chance. However, my professor continued half joking, it is the same damn German who wins every time. Statistically, the difference between first and second even third and forth and fifth shouldn't be significant - yet it invariably is, as we can point to dominant lugers who are always, on average, a few thousandths of a second faster than everyone else.

The point being, measurements are always choices made before the thing measured, and a good researcher, a good statistician, will be able to shift their measurement - to look around conventional statistical measures from time to time. I say this because too often, particularly in sports, stats are used as if they are argumentative trump cards. I do mean to say that stats are meaningless. Statistical significance is a positive boon in psychology, and it allows psychologists to accurately predict and describe (and thus positively impact) human behavior in many cases. Stats, however, are applied in order to certain achieve ends. Psychologists choose, based on convention, what level of significance is acceptable to reach a conclusion based on the results of a study. This does not make their results (or the choice) merely arbitrary, but it does mean that stats - far from settling arguments - are themselves arguments.

Day 7: On Streaks, Hot and Cold

In a brief aside in his The Invention Air, Steven Johnson argues (with respect to people who suddenly become exceptionally productive):

We are naturally inclined to see a hot hand here, some extra dose of inspiration that triggered the streak in the first place, but in fact the streak would just be an offshoot of that random distribution, no more magical than a repeated coin toss that every now and then turns up heads ten times in a row.

He builds on this with an example of statistical research in sports:

the fact that a player has just made a free throw makes him no more or less likely to sink the next one. (42)

I will grant Johnson his numbers. However, we could argue that made shots and the perception that made shots lead to more made shots must certainly affect other players (teammates and adversaries alike). I would like to know, for instance, that if the defense assumes a shooter is hot will they leave her teammates open for easy shots?

The "hot streak" is kin to the various superstitions one finds in sports, particularly baseball. I am sure that a wearing the same dirty socks for the entire playoffs itself does not affect the outcome (although I am sure that my standing with one foot in the kitchen and one foot in the living room won the Atlanta Brave the World Series in 1995 - and find a stat that disproves that, Mr. Johnson). However, I am also just as sure that washing those socks might vary well affect the performance of the superstitious player. That is, "real" effects can take various sometime unquantifiable forms (the placebo effect is invariably an effect).

This all (if you can believe it) leads me to John Shuster, the unfortunate skip of the American Curling Team who has now lost four straight matches on missed last shots. He is on one hell of an Olympic cold streak. And he is clearly shaken. Now, I don't know of any research on cold streaks (and I haven't looked that hard), but I assume, based on the Johnson's logic above, that missing a shot makes it no more likely one will miss the next one. But the cold streak strains the statistical argument against streaks even more. Whether it can be charted statistically, and what human behavior can be reliably charted statistically, things accumulate in the human mind. Statistics may have not have memories, but human beings certainly do. And those memories matter.

Shuster remembers each missed shot with each subsequent one, and those memories (emotional as they are) form a part of his decision-making process. He becomes like drivers who die in head on collisions after trying to steer back onto the road from the shoulder. This syncs up with arguments about high performing athletes who are said to be high performing precisely because they can forget previous mistakes (forgetting - repressing - isn't always a bad thing).

And we remember what other people have done. As I argue above, the perception of a hot streak could change defensive approaches. Anyone who watched downhill skier Lindsey Vonn win the gold could safely assume she benefited twice from the crash of the skier who followed her. The next competitor after the crash, who was the last and greatest threat to Vonn's gold medal, clearly played the mountain safer after a serious of crashes before her run.

The statistical argument downplays human memory and human frailty. However we describe the pattern after the fact, we must admit it only becomes a pattern after the fact. In the moment it is humans acting with bodies, memories, and the emotions they spawn. It isn't magic; it's human.