Old Timer
I ain’t no mathematician (and obviously no grammarian either). But statistics were a part of my life for a long time. Take track and field. In the “good old days”, before all of the fancy cameras (the ones that take a picture of “time over the finish line”, instead of space – see below) and automatic start timers, we used devices called stopwatches. If there were eight racers, there were eight or more people designated as timers. They all watched the starter. When they saw smoke from his gun, they pushed the button to start their watch. Then they focused on the finish line, and when the torso of the runner they were timing crossed, they pushed the button again.
Really good timers would be consistent, both from race to race and with each other. And so with experienced timers, two people could time the same place, and get the same time, within a tenth of a second. And there was a rule in the book: if two timers timed the same place, the slower time counted. A tenth of a second was a good as humans could get. Even when electric watches came that timed to the hundredth of a second, we (by the book) always rounded up to the next slowest tenth. So a time in the 100 meters of 10.90 was 10.9, but 10.91 was 11.0. We recognized that no one was good enough to catch the exact hundredth.
Wiggle Room
(By the way, now with fully automatic timing the same rule applies, but to a different order of magnitude. The “machine” normally places times to the hundredth of a second. 10.90 is 10.90, and 10.91 is slower. But, if two athletes run 10.90, the machine can go to the thousandth. 10.895 is faster than 10.897. But if two athletes run 10.895, then it’s a tie. Even the fully automatic start and video cameras have limits.)
In track time and measuring there is always a margin of error. Discus is measured to the lesser inch; shot, high jump, pole vault and long jump to the lesser quarter-inch. We recognize that there is a margin of error in measurement, just like we recognize there is one in timing. (With improvements in laser measuring, that may someday get “tighter”, like timing did. Tape measures are already being replaced by laser sights and reflecting pins).
Ok, I’ve beat the margin of error concept to death. But the same thing is true in lots of other forms of measuring, and it’s easy to be fooled. So to quote the British Rockers The Who, we “Won’t get fooled again!!”
NBC Poll
NBC News announced the results of their newest poll Sunday. This new survey has Kamala Harris at 49%, and Donald Trump at 44%. Steve Kornacki, the NBC polling “guru” was almost giddy in describing it, showing the big shift from July. Those results had then-candidate Biden with 43% and Trump at 45%. But here’s the problem, that damned margin of error thing. The mathematical margin of error in this poll is +/- 3.1%. That means that Harris could be as low as 45.9%, and Trump could be as high as 47.1%. Or it means that Harris could be at 52.1%, and Trump 41.9%. Or anywhere in between.
And the July poll was the same way. So, given the margin of error, what does this poll really tell us? It’s close. That’s all. And perhaps, it looks like Harris is doing better than Biden, but it’s all within the margin of error, so there’s no sure thing.
Listen, Americans all want a winner and a loser. It’s great to see the Bengals and the Chiefs, Burrow and Mahomes, battle it out on the football field. We marvel at the accuracy of their passing, and the agility of their runs. But, in the end, we really remember one thing. The Chiefs won the game (Dammit, wait until next time!!).
We want to evaluate polls like they were scores. So we ignore “margins of error” and just say “we’re winning” or “they’re losing”. And, in the end polls might give us some sense of what’s going on, but they are very, very, deceptive. We are fooled because we want a score. Don’t get fooled again!!
Pictures of Time
And, by the way, even if the results are beyond the margins of error: polls are a lot like those photo-finish pictures you see from the track meet. While it looks like a “u” shape with all the runners, it’s actually a picture of the exact same finish line, taken over time as each runner crosses the line. The faster runners are in front, because they finished first. If you look closely at the picture you’ll see the thin, red finish line, over and over again as each runner crossed. The space between the lines – is the time interval between the first runner and the next, all the way to eighth.
How much did Noah Lyles win the Olympic Gold Medal by? The slim difference between the first and second red lines, 9.784 to 9.789, 5 thousandths of a second. The photo-finish picture is of a moment (or about ½ a second) in time. If they lined up and did it again, it might be completely different.
In the same way, polls, even beyond the margins, are simply a snapshot of a moment in the political race. The only finish line is on election day, 41 days away. That’s the only time it counts.
Let me paraphrase President Obama:
“Don’t cheer, VOTE.”
