MN-Gov: The biggest question hovering over the Democratic primary to succeed retiring Gov. Mark Dayton is what Attorney General Lori Swanson will do. Swanson has been quiet about her interest for months, but the Minneapolis Star-Tribune writes that there's a "widespread belief she will be a candidate." Swanson's office is still playing coy, with a spokesperson saying she’s "at an important juncture in fulfilling her responsibilities as attorney general and cannot at this time divert her focus to commenting on the governor’s race."
The paper also says that if Swanson gets in, she'll likely skip the party's endorsement convention. As we've written before, both major parties will hold nominating conventions of activists the weekend of June 1. Many candidates will, in local parlance, "abide" by the party endorsement process and drop out instead of proceeding to the August primary if they aren't chosen. And while a party’s endorsement carries weight with plenty of voters, candidates can still win their primary without it.
Indeed, in 2010, Dayton announced he would not abide by the party endorsement and would proceed to the primary no matter whom the convention picked. Convention delegates gave their backing to state House Speaker Margaret Anderson Kelliher, and a number of defeated candidates, most notably Minneapolis Mayor R.T. Rybak, ended their campaigns then and there. Nevertheless, a few months later, Dayton beat Kelliher in the primary by a tight 41-40 margin.
The Star-Tribune also commissioned Mason-Dixon to conduct a new statewide poll, featuring early looks at both primaries. However, we aren’t writing them up because both polls sampled too few voters. The Democratic primary poll included 298 respondents while the Republican portion included just 218. Daily Kos Elections requires a poll to include a minimum of 300 respondents in order to make it to the Digest.
The reason for this cutoff is that the smaller the sample size, the less accurate a poll tends to be. Small sample sizes are also are more prone to produce fluctuations where there aren’t any, creating statistical noise that pundits invariably insist has to mean something, even when it really doesn’t. This seminal chart that compares Gallup’s up-and-down 2012 presidential polling with the Obama campaign’s much more sophisticated—and much more stable—polling illustrates the problem well.
On the down-ballot level, the University of New Hampshire’s habit of releasing polls of House races with small sample sizes likewise highlights this issue. UNH's results often shift wildly from one poll to the next, and while the changes often get hyped (the Democrat is surging! ... oh no, she's plummeting now!), they’re often simply the result of random error. As Daily Kos Elections' Daniel Donner has explained, this problem could largely be eliminated simply by polling more people in each poll.
We've seen very little polling in Minnesota, so we won't see large swings from one candidate to another just yet. However, because Mason-Dixon sampled so few voters for each primary, especially on the GOP side, it's tough to know where each candidate actually stands: Are they truly ahead, or are they just benefiting from random error? Hopefully, we'll see some larger sample sizes and more polls here soon to get a better feel for where these races stand.