Polls and Meaning
Those who conduct public opinion polls get understandably prickly when their work is criticized; folks who don’t like what polls say dismiss them as inaccurate, while those who do like the results proudly announce them.
The folks at Fox pointed out recently that some of the narrative about the 2016 presidential election polling is wrong, and it reminds us to examine things more deeply whenever polls are released.
The narrative is that public opinion polling got the 2016 election wrong, since polls showed Hillary Clinton ahead. The fact is that her lead going into election day was always within the margin of error, something that was rarely reported. And the polls were national assessments of how people would vote, and that’s fortunately not how we elect a president…we do it state by state, leading to the electoral college. So those national polls actually were accurate, since Clinton (thanks to California alone) won the popular vote.
But the deeper examination part is that national opinion polls are worthless if you want to predict who will win an election. They are fine for promoting a candidate who is ahead for bandwagon effect, and I suppose they’re fine for getting supporters of a candidate who is behind to rally and grow the faithful. But since we elect presidents via the electoral college, show me polls of each state…of not just registered voters but likely voters. And make sure they are polls with a margin of error of four percent or less, so they are of sufficient sample size to be relevant. And remember, margin of error is plus or minus whatever number, for each side…so a three point margin of error means you have to add three points to Candidate A while taking away three points from Candidate B…meaning as much as a six point swing.
None of that makes for easy headlines, however. “Scoreboard” is easier, but if the way you are keeping score doesn’t matter…easier is not only not better, it leads to false impressions.
