Recently Nate Silver asked us why our polls don’t bounce around much. In our polling, Hillary Clinton had a small lead in September that expanded to five or six points after the first presidential debate on Sept. 26. Since then a lot has happened – sex tapes, charges of election rigging, WikiLeaks – but our numbers have budged only slightly. Over the past three weeks, our election model and polling for The Economist has shown a consistent lead for Clinton over Donald Trump of three to five percentage points. In contrast, some other polls have shown wide swings. For example, the ABC/Washington Post poll had a Clinton lead of two points on Sept. 22, rising to 12 points on Oct. 22-23, and falling back to a single point earlier this week.
We believe that most of the bounces seen in surveys this year represent sampling noise that can be reduced or eliminated by adopting by better statistical methodology. We risk a repetition of 2012 where polling swings were largely statistical mirages. The convention and first debate bounces in 2012 were mostly the consequence of transitory variations in response rates. Fewer voters were changing their minds than were changing their inclination to respond to surveys.