Earlier this week, Daily Kos political director David Nir joined Al Jazeera’s program, Start Here, to discuss polling, statistical bias, and the lessons we’ve learned from the mistakes made in 2016. In 2016, a plethora of polls predicted a Clinton victory; but after her loss, the shortcomings of the polls—especially state polls—became apparent. For instance, college graduates were overrepresented in polls, resulting in Clinton voters being overpolled and Trump voters being underpolled.
Why do we hear so much about why polls haven’t gotten it right? Focusing on the general strength of presidential polls, David says:
“When polls do fail, they get outsized attention—and those failures are spectacular … when [that happens], they lead all the news.”
He also notes that prior to 2016, education levels were not necessarily a predictor of how people would vote, but things have shifted.
Just two years later, polls for the 2018 midterm elections turned out to be much more accurate, correctly predicting a major blue wave that would ensure a Democratic majority in the House. As David notes: “While 2016 was in so many ways a miss, in the 2018 midterm elections, polls functioned very well. They forecast a Democratic landslide, and that was exactly what we got.”
The most important takeaway is that polls capture the pulse of public opinion, but they need to be weighted properly to make sure a sample is representative. David explains: “If you think that 60% of voters are going to be women, then you want to make sure that if you talk to a thousand people that 600 of those, roughly, are women.”
You can watch the full video below.
from Daily Kos https://ift.tt/2SoTTH7
No comments:
Post a Comment