The special election train pulls into Arizona on Tuesday, when the votes will be counted from the state's 8th Congressional District.
Republican Debbie Lesko, by several metrics, is favored to win over Democrat Hiral Tipirneni in the race to replace Republican Trent Franks, who vacated the seat following allegations of sexual harassment.
But the result could be closer than Republicans would like.
President Donald Trump carried the district in 2016 by 21 percentage points.
Early voting results look good for Lesko. Over 150,000 ballots have been returned, which is about 80% of the number of ballots cast for governor in the district in the last midterm election. Of the ballots returned, 49% have been from Republicans, compared with 28% from Democrats.
That 21-percentage-point advantage for Republicans is actually slightly greater than their 17-point edge among all active registered voters in the district. Democrats could close the gap on the day of the election, but it seems unlikely at this time that Democrats will greatly outperform the voter registration baseline of the district.
This makes Arizona 8 unlike many other special elections since Trump took office. In those, Democrats benefited from low turnout among Republican voters.
To win, Tipirneni will have to convert a large percentage of Republican-leaning independent and Republican voters to overcome the Republican lean of the district, many of whom have no ties to the Democratic Party. (This is different from the special election in Pennsylvania's 18th Congressional District, where Democrats held a registration advantage that doesn't exist for them in Arizona.)
We don't have any "gold standard" polls in the district. The polls we do have either didn't use live interviews, didn't call cellphones or were put out by a campaign. Historically, polls like these are less predictive of the final result.
A poll earlier this month from Emerson College showed Lesko and Tipirneni within a point of each other. This tracked closely with a poll put out by Tipirneni's campaign, although campaigns typically release only the polling most favorable to them.
It shouldn't be too surprising that there is a wide spread in the polling. Polling a special election where you're not exactly sure who is going to turn out to vote is difficult, so turnout models among pollsters can differ greatly.
Historically, this has meant that polling in House special elections isn't all that predictive. If you were to construct a 95% confidence interval from House polling in the final three weeks of special elections since 2004, it would be +/- 13 percentage points.
The polls we do have, however, appear to indicate Tipirneni is outperforming the partisan baseline in the district based on the last two presidential elections. In a neutral environment, we'd expect Lesko to win on average by about 25 points. With the limited data we have at this time, that doesn't seem likely.
If the polling is anywhere close to accurate and Tipirneni does overpeform, this would mark the ninth congressional election this cycle in which the Democrat did better than the partisan baseline. We'll know for sure when results come in Tuesday night.