Why Are the Election Polls So Different?

This US presidential political season there seems to be extra disparity in the polling data coming out. Personally, and this is just a theory, I think the US presidential race polls are swinging wildly in short time frames because the 2 major candidates are so odious that a number of people are vacillating on who they like less on any given day. That is strictly opinion but there are real reasons why the polls show disparity.

First and foremost the various polls use different criteria to determine “likely voters”. Using anything from self-reporting to 7 question indexes pollsters frame how likely someone is to vote. Another issue is cell phones. Most polls include cell phone users (Gallup, NY Times , NBC etc) some do not (Rasmussen, Public Policy).

The big outlier this year has been the LA Times Poll which consistently favors Trump by 4-7 points it seems. It uses a methodology that asks respondents on a scale of 1-100 how likely they are to vote for a particular candidate and puts that data into a formula to determine probable voting. There is weighting in the poll that many think skews to Trump. Maybe you should just average all the polls? The folks at 538 don’t think that idea is too helpful.  They more reason that you should search for bias and note it in your estimation of accuracy.  If you want to dig deeper go here for more in depth study of the polls.

Just remember polls can be wildly wrong, ask the Brexit  poll followers about that.

UPDATE 10.27.16:   And just when we thought there were  innocent explanations for poll differences Wiki Leaks had to come out and expose the DNC folks of encouraging “oversampling” to skew polls. Its a simple proposition, you oversample demographics that skew one way and the resultant poll skews that way as well. ABC has been tagged as using obvious oversampling to come up with Hillary leading by 10-12% in recent polls.