...Nate Silver or my own lying skin?
After a slightly warmer than usual May, the last seven weeks have been unusually cool here in the Twin Cities. On the 17th this month we set a new record for the lowest high temp for that date. The old one was 66° F and we only made it to 65. The average high for that day is 84° F. The cool summer so far prompted John at Power Line to write a post titled "A Year Without A Summer?" on Friday. Now I don't know about that, and John actually says the same thing, but he's right that if things don't change, 2009 will go down as a record cool summer here.
John's post prompted Nate Silver at FiveThirtyEight to write this:
I guess this guy Nate is supposed to be a statistics expert because he writes, "I'm tired of people who don't understand statistics." Well, I'm no expert, but I think one of the things that Nate might want to do is brush up on some basics, like using matching data sources to do comparisons. After doing a little investigation into what Nate was telling me is my faulty perception, it is now clear that Nate's post is a classic case of statistics dumbassery.
His mistake is a textbook example of fail. To get his average high temperature he uses data from weather.com. I spot checked those numbers today and they appear to be right on with the averages from the National Weather Service's Twin Cities station in Chanhassen, Mn. However, for his actual high temperature data he uses a reporting station at the University of Minnesota, smack in the middle of the urban heat island effect. Doh!
Actually comparing apples to apples* as far as the high temp for the day, we get(June 21-July 17) 9 days above average, 17 days below average, and two days right at average. In fact, if you go back to the start of meteorological summer, June 1, the trend is even more pronounced. June saw 18 days below average, 11 above average, and 1 day at average. July so far(through yesterday) has given us 14 days below average, 3 days above average, and 1 day at average. Here are the numbers if anyone wants to proof them:
The bottom line is that since June 1, we have had 32 days with below average high temps, 14 days above average, and 2 days right at average highs. So I guess I actually can trust my own skin. What a relief. One caveat here is that the NWS site says that the numbers for June and July may still be adjusted, but that's where they are as of today. I'm not going to take Nate up on his silly challenge, but he just might want to take another look at his assumptions.
*NWS Chanhassen records.