Monday, February 13, 2012

Sad, Really.

Gotta love Ohio. Not sure poll workers anywhere are any better, on average.

But the fact that they aren't good could be...well, a thing. Like this court case.

In at least one instance, a poll worker appeared to be unable to distinguish between even and odd numbers. (Hr’g Tr. (Hampton) 2-202 to 2-205.) When asked whether the house number 798 was even or odd, the poll worker responded:
A. Odd.
Q. And why do you think that’s odd? I’m sorry. Why do you think her address is an odd address?

A. Because it begins with an odd number.

Q. It starts with an odd number?

A. Yes. Nine is an odd number. Eight’s even.
. . .
Q. . . . So on Election Day, if somebody came in with an address 798 and you had two ranges to choose from, you would choose the odd for them?

A. Yes.

Q. Okay. And is that how you did it for all the ballots that you looked up on Election Day?
A. To determine if they were even – yes.Q. To determine if they were even or odd, you looked at the first digit of the address?

A. No. I looked at the whole address.

Q. And you chose however many – if there were more odds than even numbers, it would be an odd address?
A. Yes.

Nod to Chateau, who KNOWS from odd...


Hasdrubal said...

Interesting, that shouldn't introduce a bias, should it? If the categorization is just to fill out pools randomly, it should work effectively the same as even/odd, just with different people in the pools. Now, if that interacts with how other people/districts do it, that's another thing, but in general shouldn't there be a roughly equal balance between "even/odd" and "greatest number of digits even/odd?"

Or would Benford's Law lead to a bias in favor of odds?

Anonymous said...

I wonder what happens when there are the same number of odds and evens?

Mungowitz said...

Interesting. The question in the court case is to throw out these ballots, and that clearly seems wrong, since they were in the right district, just the wrong precincts within the district.

DOuglas2 said...

I've lived in three Ohio cities that followed a house-numbering scheme that was uniform to the greater metropolitan area. As a result house numbers with a leading "1" and 3 or 4 other digits make up a majorty of most districts within those cities.

So, from my small sampe, there would be a bias.