Why Al Franken Will be Minnesota's Next Senator

No one should be surprised if Democrat Al Franken is elected to the U.S. Senate in the Minnesota vote count or canvass, which is not completed until Nov. 18, and that a statewide hand recount, an election contest in Minnesota courts, and the United States Senate will all confirm this result.

Here's why. Minnesota only uses paper ballots, no punch cards and no touch screen. There is a complete paper trail. Most of the ballots are counted in their precincts by optical scanners. Some rural counties use a large machine to count in the clerk's office after the polls close. In both precinct and central counting, the machine reads only marked ovals, but the test audits during the canvass (and the subsequent recount) are by hand and include all discernable marks and a recount is mandatory for close races.

Minnesota's system is pretty close to the ideal. The precinct counter warns voters if it reads more than one vote for a single office (known as an overvote). This is usually human error, and the voter has a chance to correct it (or to choose to reinsert his or her ballot, in which case the machine will count the unaffected races). In some areas, precinct scanners also warn voters if they fail to vote for every office, but most Minnesota counties do not use this setting. Because it allows voters to correct many human errors, the precinct counters are more reliable than paper alone.

There is also a strong tradition in Minnesota that every vote counts. Any discernable mark is counted as evidence of voter intent. Minnesota courts don't blame voters if election officials made a mistake in keeping the ballots in less than ideally secure locations; they require evidence of actual fraud. And Minnesota ballots are pretty straightforward, on a single page without the design flaws that confused so many voters in the 2006 congressional election in Sarasota, Fla. There were thousands of undervotes in that close contest, mostly from Democratic-majority areas.

The largest source of error is going to be human error -- marks that the machine cannot be expected to recognize. The voter crosses out one oval and fills in another. The machine sees an overvote, but the hand canvass counts the vote correctly. When the voter circles the candidate, or uses a check mark, or doesn't fill in the entire oval, the machine sees no vote at all, but again, the hand canvass registers the vote.

Since 2000, seven out of seven peer-reviewed academic studies confirm that Democrats tend to make more mistakes than Republicans. The populations that are, in academic jargon, "vulnerable to error" are low-income, low-educated and minority, all of which disproportionately vote Democratic.

Press reports from Minnesota emphasize that there were 24,100 more votes for president than for senator in last week’s election, and note that these come from "Obama counties." Further analysis will likely show that most come from Democratic precincts. This is consistent with the fact that low-income and minority voters tend to live in the same areas.

This is not the correct way to measure the unintentional undervote. The undervote is larger than the dropoff between president and senator, since some voters may not have votes counted for either office. When the Senate vote is subtracted from the total number of ballots cast, the undervote rises to 34,916. One analysis looks at this undervote by precinct and then allocates it between Norm Coleman and Franken according to the precinct vote, resulting in a net Franken gain of 1,769.

But that overestimates the error, because the vast majority of the undervote is an intentional abstention from the Senate race. Here, historical data is no guide, since how many voters deliberately abstain depends on the particular race. But the last polls before Election Day showed an unusually low percentage of undecided, most of them liberals, so the level of abstentions should also be low. If just 8 percent of the undervote results from defective marks, Franken moves ahead.