Skip to main content
ABC News
Special Election Warrants Caution for Forecasters — and for Republicans

Two new polls of Tuesday’s special election in New York’s upstate 26th Congressional District show similar results — and both suggest that the Democrat, Kathy Hochul, is the slight favorite in the traditionally Republican-leaning district.

The first poll, from Siena College, shows Ms. Hochul with 42 percent of the vote, the Republican Jane Corwin with 38 percent, and Jack Davis — an unorthodox candidate who is running on the Tea Party ballot line but who has been affiliated with both major parties in the past — with 12 percent. The second, from Public Policy Polling, has Ms. Hochul with a 44-to-38 advantage over Ms. Corwin, and Mr. Davis at 13 percent.

I cautioned two weeks ago against reading too much into this election. Some of that was because I am inherently skeptical about the predictive power of any one special election. And some of it was because of contingencies specific to this election — most notably, the presence of Mr. Davis, who at one time had about 25 percent of the vote.

With Mr. Davis’s share declining, I’ll relax my skepticism somewhat — perhaps the election requires a flashing yellow light rather than a stop sign. But first, let’s dig a little deeper into the polling.

In some cases, like a Senate or governors’ race held on a traditional November date, a roughly 5-point lead in the polls would be quite significant and would imply a 90 percent or greater chance of the leading candidate winning. This is not one of those cases, however.

First, House polling in general is fairly poor, and it is liable to be especially so in a special election (where turnout is lower and therefore more difficult to model) and one with a third-party candidate in the race.

Second, Ms. Corwin has a chance — although is by no means certain — to pick up some additional votes if some of Mr. Davis’s supporters conclude that their candidate isn’t viable. In the Siena poll, only 43 percent of Mr. Davis’s supporters said they were certain to vote for him, and just 22 percent said they expected him to win. Particularly given that these polls are likely to be widely reported upon in the district (this is one of those times when polls can influence voting behavior), Mr. Davis’s support may erode further.

The good news for Ms. Corwin is that the plurality of Mr. Davis’s voters — about 45 percent — identify as Republican. The bad news is that her favorability rating among those voters is very low — just 16 percent, as compared with 37 percent for Ms. Hochul. I’m not sure which of those factors will win out, but it does contribute volatility to the outcome — and when you’re trailing in a race, volatility is a good thing.

Third, the turnout models in the polls imply an electorate that looks more like 2006 or 2008 than 2010. The Siena poll has a sample that is 41 percent Republican and 35 percent Democrat, while the Public Policy Polling survey has 39 percent Republicans and 36 percent Democrats. In contrast, the actual voter registration in the district is 39 percent Republican and 33 percent Democrat.

Some of you may be wondering what the problem is, since the turnout estimates replicate the voter registration in the district pretty well. It may well be that there isn’t a problem at all.

Nevertheless, even in good Democratic turnout years, the percentage of Democrats in the actual electorate rarely exceeds their registration statistics by more than a percentage point or two. By contrast, when the political climate is bad for them, they can underperform them by several points. If turnout looks more like 2010, in other words, Ms. Corwin probably has no worse than even odds of pulling out the victory, even if she is something of an underachieving candidate. I can imagine any number of reasons that turnout will not resemble 2010 — or could be quite strong for the Democrats instead — but nevertheless, this is one of the ways by which she could win.

If I were asked to set odds on the race, I would probably make Ms. Hochul something like a 2-to-1 or perhaps 5-to-2 favorite. A relatively wide range of outcomes are possible, from a double-digit win for Ms. Hochul to a win for Ms. Corwin by several points.

As to the interpretation of the results, one thing I’d remind the readers of is that the margins matter as much as the victor: if Ms. Hochul wins by a single vote, that tells us almost exactly the same thing as if she loses by a single vote. Also, Ms. Hochul’s share of the vote matters: the lower Mr. Davis’s vote goes, the more we can read into the results. As I noted two weeks ago, if Ms. Hochul finishes with a vote share in the mid-to-high 40s, that would be consistent with how Democrats performed in the district in the strong Democratic years of 2006 and 2008 and is a result that Democrats could be pleased with.

There is also some evidence that Republican plans to significantly alter Medicare, which has been the focal point of Ms. Hochul’s campaign, may indeed have made some difference in the race. In the Siena poll, voters were asked to identify their most important issue. Of the 21 percent who picked Medicare, some 80 percent said they planned to vote for Ms. Hochul (excluding undecided voters).

What’s tricky about this is that it isn’t straightforward to determine whether voters are prepared to vote for Ms. Hochul because of the Medicare issue — or rather, whether they were going to vote for her for some other reason, but emphasize Medicare to pollsters because she has also. Correlation may not equal causation.

Nevertheless, of those voters who identified Medicare as their top issue, just 50 percent are Democrats, and an additional 24 percent are independents. Since Ms. Hochul is winning 80 percent of those votes instead, that implies that she is in fact picking up some support from independents and moderate Republicans (of which there are many in this district) on the issue.

Suppose, for example, that under ordinary circumstances, Ms. Hochul would win about 60 percent of these voters (almost all of the Democrats, plus a significant share of the independents). If she in fact wins 80 percent instead — overperforming among them by 20 points — that would be worth about 4 percentage points to her over all since they represent about one-fifth of the electorate (while taking roughly the same fraction of votes away from Ms. Corwin).

If the results end up something like Ms. Hochul 46 percent, Ms. Corwin 44 percent and Mr. Davis 9 percent, that will reflect a fairly big departure from typical outcomes in the district. What will be more difficult to determine is whether they represent any sort of leading indicator. Medicare is likely to be more important in this district than in most others because of its preponderance of older voters, and Ms. Corwin has committed at least one major gaffe in the race unrelated to the issues themselves. At the same time, it would be naive to suggest that the Medicare issue has had no impact at all.

Once there have been several special elections, they may begin to have some predictive power — but reading too much into the results of any one of them is dangerous. Nevertheless — since Democrats are going to hammer away at the Medicare issue almost no matter what — the burden of interpretation will be with Republicans if they have a disappointing evening. They’ll face a choice between toeing the party line on the issue of entitlements and allowing their candidates more wiggle room. Because of the issue’s salience in the presidential primaries, it’s a choice that they’ll have to make on an accelerated timetable without the luxury of waiting for more data to come in.

Nate Silver founded and was the editor in chief of FiveThirtyEight.