In an interesting new experiment, the New York Times has partnered with Siena College to poll almost 100 House races, with the results getting updated on the Times website live as each interview is concluded. That means, of course, that while a poll is in the field, the early numbers won't be useful and in fact should be disregarded, since they'll reflect sample sizes that are too small; even the Times itself grays out early results and warns, "Don't take this poll seriously until we reach at least 150 people." We recommend waiting until these polls reach 300 respondents, which is generally considered the minimum acceptable number for any poll.
Fortunately, we already have full results for the first three contests that Siena has polled (you can see polls that are in progress here). All show extremely tight races in GOP-held seats:
- CA-48: Harley Rouda (D): 45, Dana Rohrabacher (R-inc): 45
- IL-06: Sean Casten (D): 44, Peter Roskam (R-inc): 45
- IL-12: Brendan Kelly (D): 43, Mike Bost (R-inc): 44
In a move that's rare but shouldn't be, the Times also cautions at the top of each write-up, "But remember: It's just one poll." That's always been our mantra: A poll is just a single piece of data, and it should never be taken as gospel but rather read in conjunction with all other available information, including other polls.
So what do other polls of these races say? Well, there haven't been many, but in July, Monmouth found a similarly close battle in California's 48th, with Rouda ahead 46-43. By contrast, in Illinois' 6th, another July poll for a Republican firm, Victory Research, which said it doesn't have a client in the race, gave a wider lead to Roskam, putting him up 44-37 on Casten. And in Illinois 12th, we haven't seen any surveys since April, when Public Policy Polling found Bost up 44-39 on behalf of a Democratic super PAC. (In fact, if you scroll down toward the bottom of each poll, you'll see the Times has included summaries of these other polls.)
In all three cases, the Times concludes, "Our poll result is about what was expected," and we agree: Everything we know suggests these races should be competitive, and indeed they are. One thing to note, though, is that while the horserace is neck and neck in each of these districts, Trump’s approval rating varies quite a bit. He puts up a lousy 40-54 score in California’s 48th and a brutal 36-57 in Illinois' 6th, both of which he lost in 2016. However, in Illinois’ 12th, which he carried 55-40 two years ago, he’s predictably doing better, with a 48-46 rating.
Numbers like these show we can win back the House. Give $3 to help Democrats do just that!