CBS Pollster Says No Blue Wave

Anthony Salvanto, a top pollster who has been heading up CBS’s election polling unit for several years, has just written a book, “Where Did You Get That Number?” He reviews the mistakes made in 2016 and lessons learned. One lesson in his view is that panel tracking polls that call back the same voters during the course of the election are better than discrete random samples. Maybe, but for this election, he does have a targeting technique that is worth watching. His panel is made of voters in the 60 House districts that are most likely to be in play in November.

Anthony Salvanto | Twitter

From his latest poll, he believes Republicans could lose 20 or 22 seats, but that there is a very good chance they will hold the House. That conclusion is getting him very good coverage in the pro-Republican media, such as The New York Post and RealClearPolitics.com. But, one weakness in his panel methodology is that to the extent the election is driven by the usual voters, indeed Republicans may be able to hold on, but if new voters are involved, Republicans are in trouble. Panels are prone to miss changing voter turnout patterns. The two national panel polls in the 2016 election, which had Trump ahead, in fact, missed the popular vote result by a wide margin. The surprise of President Trump’s 2016 election win was a product of failed state-level polls as Salvanto points out, not a difference between panel and discrete random national surveys.

Although there is still a debate among pollsters and pundits on the likely results of the 2018 midterms, the weight of opinion, as of August 25, is Democrats by 23 seats or more.