Statistics Don't Lie, Even in Minnesota

Al Franken’s unusually large vote gain in the Minnesota Senate race since the Nov. 4 election has generated a lot of anger. Just correcting “typos” in how the votes were recorded has made the difference -- reducing incumbent Sen. Norm Coleman’s lead from 725 to 206 votes when Minnesota counties certified the votes on Monday, Nov. 10. Correcting these typos supposedly produced a net swing of 459 votes to Franken and took 60 votes from Coleman.

Last week, using the vote totals from Sunday, Nov. 9, I pointed out that Franken’s net gain was huge -- “new votes for Franken from all the precincts is greater than adding together all the changes for all the precincts in the entire state for the presidential, congressional, and state house races combined.” One “precinct’s corrections accounted for a significantly larger net swing in votes between the parties than occurred for all the precincts in the entire state for the presidential, congressional, or state house races.”

Minnesota’s leading newspaper, the Star Tribune, ran a story by Eric Ringham evaluating my numbers. Glenn Howatt, their editor for computer-assisted reporting, was quoted as saying, “[Lott’s] numbers are simply wrong.” He went on to say, “I don’t know what statistical calculation Lott was using, but Obama clearly got more from the corrections.”

Franken’s campaign has pointed to Ringham’s piece to defend the “typo” correction process. Other pieces at the Star Tribune attacked the “GOP bobos, starting with Fox News and some conservative columnists like John Lott.” Media outlets such as Salon.com have reiterated the newspaper's claims.

The Star Tribune argued that Obama’s net gain was 1,121 votes -- much bigger than Franken’s, and thus evidence that nothing unusual had occurred in Franken’s race.

So what accounts for the different claims? While my work looked at the impact of correcting typos, the Star Tribune primarily relied on something quite different -- two precincts didn’t report the presidential votes for their districts until Nov. 5, the morning after the election.

For example, precinct Richfield W-3 P-01 went from 0 to 908 votes for Obama. For McCain, it went from 0 to 474. This was not an incorrect number that was entered and later corrected, but simply the result of no numbers having been entered for either Obama or McCain or any other presidential candidate until Wednesday morning.

In precinct Alexandria W-1 P-2, the Obama vote count went from 0 to 107. That precinct also had shown zero votes for Obama, McCain and all the other presidential candidates. So, again, it is clear that the precinct had not yet reported its totals for the presidential race.

What is the difference between these two precincts and all the other 4,128 precincts in the state? These two precincts reported their numbers a couple of hours late.

FOX News contacted the auditor of Douglass county, where the Alexandria precinct is located. The auditor said that they had tried to send the presidential results, but there had been a network malfunction and so the numbers were not sent until Wednesday morning. The auditor thought this change was very different from those that other counties had in correcting typos.

Looking at the precincts that had entered in a number and then changed it between Wednesday, Nov. 5, and Monday, Nov. 10, Obama picked up 106 votes. Adding together the changes for Richfield (908), Alexandria (107), and the corrected typos (106) yields 1,121 votes, exactly the number that the Star Tribune reported.

The fact that correcting typos increased Franken’s count by 459 votes (not counting Coleman’s lost votes) and Obama’s by 106 doesn’t prove fraud. Indeed, the Star Tribune might still be right in its claims that election officials made mistakes because they were tired.

But my point was a simple one: Why did the “typo” corrections increase Franken’s total so much more than any other candidate’s? Indeed, so much more than all the other races for the presidency, Congress, and statehouse combined. The Star Tribune’s response was to deny the claim was true.

Unfortunately, neither Eric Ringham at the Star Tribune nor Joe Conason at Salon.com have agreed to run corrections for their pieces. Presumably, the people at the Star Tribune just hadn’t looked at their data set to see that no numbers had been recorded in those two precincts before they ran their numbers. Yet, if these newspapermen really couldn’t figure out why our numbers were so different, why didn’t they try contacting me before they published their articles?

With Democrats officially picking up Alaska’s Senate seat Tuesday, the anger will only increase. Democrats are only two seats away from a filibuster-proof Senate, which will enable them to completely ignore Republicans in Congress. And, with races in Georgia and Minnesota still undecided, they have a shot at winning those two seats.

Yet, however heated the race becomes, the most important thing is that the media get the facts right.

*John Lott is the author of Freedomnomics and a senior research scientist at the University of Maryland.