NOVA Green
Oregon Progressive
Atlas Icon
Posts: 11,450
|
|
« on: November 14, 2016, 05:05:42 PM » |
|
It is interesting how few polls in the last week or so of the election actually had Michigan as a tie or Trump leaning, or even showed the race as remarkably close.
You had this 538 unranked pollster showing the race as a tie, a SurveyMonkey (C +2), Trafalgar Group (T+2), and a Google Consumer Surveys showing C+1. These are all relatively questionable polling firms
Meanwhile the EpicMarist (A-) showed it as C+4 with a large number of "others" and "undecideds".
Ipsos was arguably the closest of the higher quality polling firms (538 A-) showing C+1 with relatively few undecided and 3rd party voters.
I guess the question is did some of the sketchier polling outfits, and Strategic Decisions:
1.) simply do a better job of forecasting the composition of the electorate, or 2.) did polls overestimate Clinton LV % numbers, 3.) Or did undecided & 3rd party voters break heavily Trump in the last few days of the election which would not have been captured in polling numbers? 4.) Or do we just fall back on the cliche that Michigan polling is just historically flawed and we should completely ignore all polls coming out of Michigan?
|