## Archive for November 8, 2017

### The Virginia Governor’s Race and MP.6

Hotly contested. Particularly nasty. Ugly. Uncivilized. Controversial. Disgusting. Dirty.

Those are just some of the words and phrases used to describe this year’s race for governor in my home state of Virginia. It’s part of the reason that we had the highest voter turnout for a gubernatorial election in 20 years. It’s also why every citizen in the Old Dominion was anxiously awaiting the results.

I was no different. At 8:30 p.m. EST, I strolled on over to POLITICO, where I was presented with more information that I could handle:

It was surprising that Northam had a 5.8% lead, since some recent polls suggested that his lead had dwindled to as little as 3.3%. It was surprising that only 90 minutes after the polls had closed, many news organizations were already declaring Northam the winner. But it was outright astonishing that POLITICO was displaying the percent of precints reporting as:

72.68121590023384%

WTF?

On Feburary 13, 2011, in a post titled Statistically Speaking…, I presented the following joke:

69.8724% of all statistics reflect an unjustified level of precision.

Three years later, in a post titled Sound Smart with Math Words, I presented another version of that same joke, though this time the percentage was expressed to the millionths:

An unprecedented 69.846743% of all statistics reflect an unjustified level of precision.

Did I think the additional precision would make it more obvious that the sentence was actually a joke? Or did I just think it would make it funnier? I’m not sure.

But I do know that it would never have occurred to me to take the level of precision to 14 — count them, 14! — digits of accuracy beyond the decimal point.

But POLITICO thought it was necessary.

That’s right. They calculated and displayed the percent of precincts reporting **to the hundred-trillionths place**. Hundred. Trillionths.

That’s like stating the winning time for the Tour de France to the nearest millisecond.

Or estimating the weight of the Earth to the nearest gram.

In fairness to POLITICO, though, the percentages that they were reporting not only reflected an unjustified level of precision. They were also wrong.

According to the Virginia Department of Elections, there are 2,567 precincts in the commonwealth. If 1,865 precincts had submitted results, the percent of precincts reporting could have been displayed as:

72.6529%

If 1,866 precincts had turned in their results, the percent of precincts reporting could have been displayed as:

72.6918%

But there is no number of precincts for which the percent could have been reported as:

72.68121590023384%

So, either POLITICO was using an incorrect denominator, or their algorithm was incorrectly calculating percentages.

Oh, well. At least they got the election results correct. (I hope.)

In their defense, they did finally make a correction. When I checked the results at 9:24 p.m. EST, this is what was presented:

The percentage of precincts was displayed as a more reasonable 97.74%. From this, I can surmise that 2,509 precincts had reported their results (since 2,509 / 2,567 = 0.9774) and that POLITICO had finally found someone who was nimble with a slide rule.