Do live-interview polls and nonlive polls garner the same results? Not according to recent Fivethirtyeight.com analysis. It revealed that when only live polls are considered, Hillary Clinton has an 86% chance of winning with 354.1 Electoral votes and a +7.1 popular vote margin. However, Donald Trump has performed much better in nonlive polls – those conducted online or by interactive voice response (IVR or “robopolls”). Using nonlive polls in their model drops Mrs. Clinton’s chances to 70.5%, with 315 Electoral votes and a +4.8 popular vote margin.
Looking at RealClearPolitics.com “Toss Up” states, live polls give Mrs. Clinton a higher margin in all but one state (AZ). The differences between the Clinton victory margins in live polls and nonlive polls is significant in WI (nonlive polls produce -8 point smaller margin), CO (-5.4), MO (-5.3), GA (-5), NC (-4.5), and PA (-3.8), and smaller in VA (-3), IA (-2.5), NV (-2.5), ME’s 2nd Cong. District (-2.3), NH (-2.1), OH (-2), FL (-1.5), and AZ (+1.6).
Which method is correct? Fivethirtyeight puts more stock in the traditional live-interview polls for 2016, as nonlive polls still risk including nonvoters in their Internet samples and robopolls are prohibited from calling cellphones, limiting them to landlines. Still, now that outlets like NBC and New York Times have partnered with Internet polls SurveyMonkey and YouGov, nonlive polls will likely be common. It will be important to watch which type of polls are more accurate in Nov. and how both will evolve and adapt in future election cycles.
