Iowa’s U.S. Senate campaign has been stuck in a holding pattern for most of the summer. Seven straight opinion polls showed either a tied race between Bruce Braley and Joni Ernst, or one candidate ahead by 1-2 percentage points, well within the margin of error. For weeks, I’ve seen negative ads against both candidates almost every day on television, with a positive spot occasionally sprinkled in. I keep hearing the same anti-Braley or anti-Ernst ads again and again on radio too. Since no major external event has occurred to change the dynamic of the race, I was expecting to see more statistically tied polls at least until the first of three debates to which the candidates have agreed.

Instead, last week Loras College released a poll showing Braley ahead by 45.3 percent to 40.5 percent. Braley had better favorability ratings than Ernst.

Today Quinnipiac released a poll showing Ernst ahead by 50 percent to 44 percent. Ernst had better favorability numbers, led among independents, and had a much bigger lead among men than Braley’s lead among women.

The Loras poll of 1,200 likely voters had a margin of error of plus or minus 2.82 percent. The Q-poll of 1,167 likely voters had a margin of error of plus or minus 2.9 percent. At least one of these polls is way off. Neither Loras nor Quinnipiac have polled in Iowa before this election cycle, so we don’t have a track record to judge them by. For what it’s worth, the available evidence hasn’t convinced me that either Braley or Ernst has a significant lead, and here’s why.

UPDATE: Fox News is out with their latest Iowa poll: Braley and Ernst are at 41 percent each. Notably, the sample includes 36 percent self-identified Democrats, 34 percent Republicans, and 25 percent independent/other.

One important thing to remember about margins of error: they only have meaning if the pollster sampled a representative subset of the electorate.

Polling involves recruiting a random sample and recording their answers to the poll questions. The results are usually reported as precise values, which give us an estimate of the population’s views. But the sample is only a subset of the population, and that estimate will have some amount of error.

The margin of error lets us estimate a range, within which we can be reasonably confident the population’s views actually fall. The sample values, our best estimate, are in the middle of that range, but the range extends above and below that point by the margin of error. In other words, we estimate that the population’s real support for any given polling response are within one margin of error above or below the percentage response in the poll’s sample.

Assuming perfect sampling methods, the real state of public opinion lies within the confidence interval about 95 percent of the time. But even with perfect sampling methods, about one out of every twenty polls would be way off, just by chance.

If we believe in the Loras poll’s sampling, there’s a 95 percent chance that Ernst’s support lies between 37.7 percent and 43.3 percent. If we believe in the Quinnipiac poll’s sampling, there’s a 95 percent chance that Ernst’s support lies between 47 percent and 53 percent. At least one of the surveys is way off.

There were a few methodological differences. The Loras poll used live interviewers and included many more landlines (80 percent) than cell phones (20 percent). The Q-poll sampled about the same number of respondents by cell phone and landline.

The polls made different predictions about Iowa’s likely voter universe too. In the Loras poll, 32.6 percent of respondents identified as Democrats, 30.2 percent as Republicans, and 33.8 percent as no-party voters. In the Quinnipiac poll, 26 percent of the sample were Democrats, 28 percent Republicans, 41 percent were independents, and 5 percent were “other” or did not answer that question.

I don’t know whether more Democrats or Republicans will cast ballots in Iowa this year. I don’t know whether Braley or Ernst has better favorability ratings now. I don’t know whether either candidate has a significant lead among no-party voters. I don’t know whether Braley’s lead among women will be enough to off-set Ernst’s lead among men.

But I do know one thing: no-party voters will not cast a plurality of the ballots in Iowa’s midterm election, let alone anything close to 41 percent of the ballots (as predicted by the Quinnipiac poll).

A plurality of registered voters in Iowa have no party affiliation. However, participation in midterm elections compared to presidential elections drops off much more sharply for independents than for partisans. No-party voters comprised only about a quarter of the electorate in the last three Iowa midterm elections (click through for statewide statistical reports on turnout in 2010, 2006, and 2002).

Until several polls show a clear lead for either Braley or Ernst, I am going to consider both of these polls possible outliers. But even if a trend emerges in the polling, it’s important to remember that survey samples involve guesswork by the pollster on the likely voter universe.

Any comments about the U.S. Senate race are welcome in this thread.