Saturday, April 08, 2006

Is This How We Do Recounts?



From the 2004 election recount effort in Ohio:

Election workers in each county are supposed to count 3 percent of the ballots by hand and by machine, randomly choosing precincts for that count.

If the hand and machine counts match, the other 97 percent of the votes are recounted by machine. If the numbers don't match, workers repeat the effort. If they still don't match exactly, the workers must complete the recount by hand, a tedious process that could take weeks and cost hundreds of thousands of dollars.

But the fix was in at the Cuyahoga elections board, Baxter charges.

Days before the Dec. 16 recount, workers opened the ballots and hand-counted enough votes to identify precincts where the machine count matched.

"If it didn't balance, they excluded those precincts," Baxter said.

"The preselection process was done outside of any witnesses, without anyone's knowledge except for [people at] the Board of Elections."

On the official recount day, employees pretended to pick precincts randomly, Baxter says.

If this is true I'm...stunned into silence. Me.

To clarify. The point of the recount was to check if the machine counts were correct by counting a random sample of votes by hand. Differences between this hand count and the machine count indicate that the machine counts might be wrong. The counters are given a second chance to do this random check. If even then the counts differ all votes must be recounted by hand. To omit those precincts that showed differences between hand and machine counts would guarantee that we will never find if the original machine counts were wrong....

What's the word to describe all this? Criminal?
----

Or think of this example. You are in charge of the large canned goods department of a supermarket and a customer argues that she got sick from a bad can of salmon. The sign of a bad can is dents. You go to the salmon cans and randomly check a few. Sure enough, they have dents. So you randomly check a few more, and they have dents, too. Then you take out the cans with dents you found and state that the rest are all good.