Online polls became a punch line last week when Donald Trump cited numerous opt-in, online polls, without analytics to show that he won the first debate. Telephone and online polling that came out after the debate, from CNN and YouGov, both with analytics, showed Clinton winning the debate. This was affirmed by subsequent telephone and online polling throughout the week that both showed Clinton as the winner and demonstrated the impact with a larger lead in the horse-race. While it is great to have a random and representative respondents, the opt-in method or the online mode of the polls Trump cite is not necessarily the problem, it is the complete lack of analytics. The polls that Trump cites are raw data; no polling company ever reports raw data. With MSN, we are taking the same type of raw data and making it useful.

While a slight majority of respondents to the MSN post-debate poll stated that Trump was the winner, when we ran that data through our model we got 80 percent saying Clinton won with just 12 percent saying Trump had won. More important, we saw the impact of that in our daily tracking poll. We saw a distinct rise in vote share for Clinton, couple with a similarly large drop in vote share for undecided or other. Trump’s vote share is relatively flat.