Here we are about 18 months before the off-year election and no one has seriously addressed the elephant in the room, namely, the crisis in polling. Trying to conduct telephone-based polls, even with cell phones, is going the way of the dinosaurs. There is a reason every major market research firm has turned to Internet panels for data collection. There is a reason that the major news outlets like NBC and The New York Times are partnered with an Internet platform to conduct their voter surveys.
The 2016 election required pollsters to do more modeling and weighting than any year since polling became an accepted part of campaigns. There is an echoing mantra I keep hearing since the 2016 election -- THE POLLS WERE WRONG. My biggest concern is that academic types are now producing analysis indicating response rates to telephone surveys are so low the data is being skewed.
If only left-handed people were interviewed during a survey, wouldn't a campaign question whether the sample is representative of the electorate? Of course. So, why wouldn't we use the same logic when a poll only focuses on voters who still use landlines or who will answer an unknown call on their cell phone?
Changing the way pollsters have operated for 30 years is hard, frightening, and bold. But, if we don't do it, that big elephant will be sitting in our living room in November 2018.