Saturday, June 13, 2015

Climate change and baseball

If I have one more person try to tell me that because global warming has paused over the last 17 years, human beings must not be contributing to climate change, I'm going to explode.  That argument is the product of searching for some evidence to validate a preconceived belief. 

In the study of baseball statistics, it's kind of an inside joke to mention sample size, because although it's important identifying where data stabilizes and becomes informative, it can be applied to literally every piece of information.  Any time something wonky happens, sample size is the easiest and often most accurate explanation, even if there's a far deeper cause for the deviation, often temporary, but occasionally permanent.  It's funny because in something that can be so mathematically complex, a child can watch Mike Trout go 0-4 with 4Ks and identify that's not indicative of future performance based on a wealth of past information, yet as the selection grows even slightly, many overreact.

With the evidence present for the rise in surface temperature since the beginning of human industrialization, it's ridiculous to select a 17-year sample of slowdown, and present it as negating the long term evidence of climate change.  There are probably a near infinite number of natural variables that could temporarily counteract the effects, and cause a hiatus.  I'm far too dumb to explain exactly why because I certainly don't understand the intricacies of thermodynamics.  There are plenty of theories out there, but I'm incapable of determining the accuracy of any of the work.  I'm at the mercy of the author. 

The best explanation I'm capable of giving is in terms of baseball statistics, of course.  You can't take surface temperature data from a 120-year sample, select the most recent 17 years, and reach conclusions from that tiny sample.  It's as unwise as taking ERA from a 120-inning sample, selecting the most recent 17 innings, and deciding that ERA is indicative of a true performance level, and of future performance.  There's just too much noise in that data.  If you strip it down further and analyze Velocity, K/BB, HR rate, pitch selection/sequencing, swinging strike rates, etc., you may begin to identify if the change in overall results (ERA) is just the result of natural variations in a small sample size, or if you're potentially seeing the beginnings of a change in skill level.  Ultimately, you're just making an educated guess until you have enough data.  It's the same when analyzing the data in climate change. 

If we're equating surface temperature to ERA, there are other components of a climate system that can provide a more complete climate picture, as pitchers have other indicators that provide a more complete picture of their true performance .  When studying the atmosphere, measurements at the top of the atmosphere show that Earth is receiving more energy than it is radiating back into space.  Scientists say the retained energy should be producing warming.  A baseball analogy for that would be, when a pitcher allows more fly balls, you'd expect to see a higher number of homeruns allowed.  Over 120 years we have seen that warming trend, and over 120 innings, you'd see an increase in homeruns allowed.  However, over the sample size of 17 years/innings, it's entirely possible natural fluctuations in a complex system could cause a stagnation that's misleading.  If you look at any pitcher in 17-inning samples, his HR/FB rate will fluctuate wildly without there being any real change in skill level.  It takes time for that data to stabilize.  Two other, often overlapping, components in the Earth's climate system is the hydrosphere and the cryosphere.  Despite the recent hiatus in surface temperature, sea level rise has not stopped in recent years, and Arctic sea ice decline has continued.  Again, this would suggest a continuation of climate change due to global warming, despite the pause in surface temperature.  In baseball terms, this is comparable to a rise in BB-rate combined with a decline in K-rate.  It's a clear indication of declining skill.  Over a small sample, naturally variations can lead to ERA being unaffected, masking the declining skill, but like before, the data will eventually stabilize and reveal a true performance level. 

In summary, you shouldn't just toss out 17 years of data, but you can analyze it, and combine it with past data, to make more reasonable assertions.  Just as you shouldn't unthinkingly panic if Mike Trout hits .200 over the next month, or if Clayton Kershaw has a 4.50 ERA in July, you shouldn't dismiss a clear trend just because there's a relatively brief hiatus in one aspect of the progression.  Try to understand natural variances in sample sizes.  Children do it watching Mike Trout.  Everyone is capable.

No comments: