My friend Ian Moore recently sent me a copy of an letter that he sent to Senator Norm Coleman. Ian is a keen thinker, and his letter to the Senator is sharply reasoned.

Ian writes:

“I sent this letter to Senator Coleman via his website, and I thought I’d like to share it because it outlines some details that I think are important having to do with measurement and how they relate to this recent election. Maybe you would like to pass it on, or even resend it to Coleman, maybe you would like to ignore it. Anyway, I think everybody needs to think about this right now.”

Hello Senator Coleman:

I am writing to explain a little bit about measurement uncertainty from a scientific standpoint, and to point out how it relates to the recent senate race you participated in.

I am a senior in physics at the University of Minnesota. The U of M is an institution you should be proud of, and one of the things I have learned there is measurement uncertainty.

The election uses votes to measure the desire of the state population, and like any measurement, there is uncertainty. Now, an initial measurement cannot possibly have the precision or the accuracy necessary to be within 0.04 percent (the 0.04 is from when the difference was as high as 700 and some votes) and have that margin be greater than the uncertainty.

An analogy that comes to mind would be to take a common meter stick off the shelf and measure something, and get a result that was to within 0.4 mm. The problem is that this meter stick would have worn ends. And the tic marks would be not quite in the right place from the factory. Even the temperature of the room will have an effect, etc. That measurement just could not be trusted because the probable uncertainty is greater than the quoted margin of error of the measurement.

In the case of this election, the probable uncertainty in the measurement is far greater than the relative difference in votes. The law says the probable uncertainty is 0.5 percent, and even that is probably low. But it is at least 10 times the margin obtained in the vote.

In physics or any time, when a measurement is made (often with a process that is even more direct than an election), a number that is within a couple percent of the theoretical value can be difficult to obtain. And to get any better, one would have to take the measurement a large number of times and obtain a average and a standard deviation. This election was performed once (for obvious reasons).

It only makes sense that in this situation, since we only took one measurement, that we examine our meter stick closely to make sure to get as accurate a result as possible.

I’m sure you would want the recount to proceed if such a statistically insignificant amount was declaring you behind instead of ahead. I’m sure we all would.

This is not about negativity, or an unwillingness to step aside. It is simply that the margin is too small to be trusted, and to expect an opponent to step aside based on that margin is unreasonable and shows bad sportsmanship.

I have one more thing to mention. I have read that some people in your camp are suspicious of the fact that the gap is closing in a systematic manner. Are you or your people casting doubt on our election officials? Because if you are, then you are doubting the entire process that got you a lead to begin with. Let them do their job. You can’t pick and choose which measurements to trust, whether you are a physicist or a politician.

Thank you very much for your time and your attention, a proud Minnesota resident,

- Ian Moore

Used by permission — thanks, Ian!

See also:

Measurement uncertainty @ Wikipedia.

United States Senate election in Minnesota, 2008 @ Wikipedia