Indeed, it's been pretty cool in Minneapolis for the past couple of days; the temperature hasn't hit 70 since midday Thursday. But has it been an unusually cool summer? No, not really. Since summer began on June 21st, high temperatures there have been above average 15 times and below average 13 times. The average high temperature there since summer began this year has been 82.4 degrees. The average historic high temperature over the same period is ... 82.4 degrees. It's been a completely typical summer in Minneapolis, although with one rather hot period in late June and one rather cool one now. (Note: actual high temperatures can be found here and historical averages can be found here.)
The terms of his bet:
You are eligible for this challenge if:
1. You live in the United States and provide me with your home address and telephone number (I will provide you with mine) and,
2. You are a regular (at least once weekly) contributor to a political, economics or science blog with an Alexa traffic global ranking of 50,000 or lower.
The reason for the latter requirement is because I want to be able to shame/humiliate you if you back out of the challenge or refuse to pay, as I'd assume you'd do the same with me.
The rules of the challenge are as follows:
1. For each day that the high temperature in your hometown is at least 1 degree Fahrenheit above average, as listed by Weather Underground, you owe me $25. For each day that it is at least 1 degree Fahrenheit below average, I owe you $25.
2. The challenge proceeds in monthly intervals, with the first month being August. At the end of each month, we'll tally up the winning and losing days and the loser writes the winner a check for the balance.
3. The challenge automatically rolls over to the next month until/unless: (i) one party informs the other by the 20th of the previous month that he would like to discontinue the challenge (that is, if you want to discontinue the challenge for September, you'd have to tell me this by August 20th), or (ii) the losing party has failed to pay the winning party in a timely fashion, in which case the challenge may be canceled at the sole discretion of the winning party.
Someone tried to take Silver up on the bet. See here. His major change to Silver's terms were that the temperature be based on Minneapolis since that's where the whole thing started. Silver declined saying:
But can you use somewhere other than your hometown? No, that would be a problem. I'm fully aware that this is a somewhat "bad" bet for me for the next 30-60 days in about 10% of the country - an area which happens to include Minneapolis. After that, basically the whole
country is expected to have average-to-above-average temperatures through the middle of 2011.
If someone from MSP or Chicago or Milwaukee happened to want to take the bet -- that's fine, I'm a man of my word (although I'd hope that person would not be a total weenie and would continue the bet for longer than 30 days). But if I allow people to cherry-pick their location, I'm just printing money for them, and it sort of defeats the purpose, which is to illustrate that *on average* the country is getting (slightly) warmer.
So what should we make of this? First, Silver should have stood by his original statement and accepted Minneapolis. He seems to have backed off from his original assertion that things are normal in MSP after checking the data more closely. More important is Silver's use of days above and below average temperature as a measure. Ryan Underdown mentions this is passing and points out that average temperatures have been down. I'd like to expand on this subject.
The problem is that average temperatures do not conform to what we think of as averages. When someone says "average", we think "typical". This works well when the data is distributed evenly but poorly when the data is grouped.
Take average height. If I refer to average height for men then I am describing a typical man. The same for women. The problem is that the average height for men is several inches higher than the average height for women. If I refer to the average height for humans I would be using a number that does not describe the typical person.
To illustrate with a very small sample set - I am 6'1". My wife and daughter are 5'3". Our average height is 5'6", even though none of us is that height. What's more, the sample set is skewed towards the lower end. If I were to bet that the next person to walk into the house would be below the average I would win 2/3s of the time. (If my daughter still lived at home.)
To apply this to Summer weather - temperatures are not evenly distributed. There are sunny days and rainy days. A typical sunny day in July in Ohio (where I have personal experience) will have a high in the upper 80s or the low 90s. A typical rainy day will have a high in the 70s. There are more sunny days than rainy days in July but there are enough rainy days to drag down the average a bit. The average high for Ohio in July is 85 but, as I said earlier, the typical high on a sunny day is slightly higher.
This means that in an average year the number of days above average is going to be higher than the number of days below average. That means that Silver has an advantage in his bet which has nothing to do with global warming. It is just a side effect of the statistical set he chose to use.
If Silver wanted an honest bet then he would bet on the average temperature for the month. That includes enough sunny and rainy days to bring the monthly average in line with the historical averages.
You can see this in the various links I have already given. By figuring the number of days above or below average, Minneapolis is having a normal Summer but when you look at the average temperatures it is cooler than normal.
None of this proves anything about global warming or long-term climate trends. The weather for a single city measured a month at a time is meaningless when compared to global trends. Silver should know this which makes the point of his entire wager dishonest. The fact that his use of daily average temperatures gives him a statistic advantage makes things worse.
No comments:
Post a Comment