After all my analysis, I have to conclude that I have no idea.
The prevailing theory is that Severson's work with minority communities helped him close the gap in his race, but the data does not seem to back that up.
I based my analysis on comparing Severson's results to those of two other statewide candidates- Jeff Johnson and Randy Gilbert. Johnson because he was the "top of the ticket," and therefore should theoretically get more votes than down ticket races and Gilbert because the Auditor's race saw a similar "drop-off" rate to the SOS race.
Here are a few high level observations:
- There were about 1700 more votes in the Auditor's race than the SOS race. Total "drop-off" (meaning people who voted in race for Governor but not in other constitutional races) was about 3% in both races. There were about 60,000 fewer votes total for Auditor/SOS than Governor.
- Despite the drop-off, Severson only trailed Johnson by 227 votes statewide, while he received over 100,000 more votes than Gilbert.
- Severson received more votes than Johnson in the 4 "outstate" congressional districts (1,6,7,8) while Johnson received more votes than Severson in the "metro" districts of 2,3,4, and 5.
These results stay pretty consistent when you break down the data a bit more.
Severson received more votes than Johnson in 59 counties. Johnson carried more votes in 28.
Of the 19 counties where Severson's margin over Johnson was more than 100 votes, all were outstate: Stearns, Benton, Blue Earth, Clay, St. Louis, Crow Wing, Carlton, Houston, Beltrami, Itasca, Nicollet, Steele, Winona, Cass, Waseca, Pennington, Polk, Freeborn, Lyon.
Johnson's margin over Severson exceeded 100 votes in 15 counties, many of which are metro, or border on the metro area: Todd, Sherburne, Chisago, Fillmore, Isanti, Olmsted, Anoka, Mower, Carver, Becker, Wright, Washington, Dakota, Ramsey, Hennepin.
So again on a broad theme, Severson did better outstate, and Johnson did better in the metro.
When you dig into the precinct level results, the results get kind of interesting. Gilbert actually outperformed Severson and Johnson in several precincts in St. Louis county, but nowhere else.
But again, the precincts that favor Severson tend to be outstate, while the precincts that favor Johnson tend to be in the metro. There are a handful of outliers to that rule, but there doesn't seem to be any pattern.
Overall I will restate that I'm not sure what my analysis ultimately proved, but if the theory about Severson's minority outreach efforts were to be true, I would have expected to see several metro precincts where Severson outperformed the ticket. That just wasn't the case.
It seems that Severson's strength came mostly from outstate areas, while Johnson's came from the metro.
Here's the spreadsheet I used for my analysis, in case you want to dig in yourself.