The Trouble with polls

 

Polls this year seem to be at odds with observations.  Let me offer some insights about polls based on my experience as a campaign manager.  First, only about 1/3 of people will respond to a poll.  That’s not a problem as long as the poll manages to get a good sampling.  But what if one partisan group refuses much more often than the other? Then getting a good sample is very very hard.  That seems to be the case of R’s who don’t like pollsters.  Perhaps that is the perception that polls are conducted with bias for biased news organizations—to give the answer that media group wants.  Whatever the case, it is playing havoc with polls today.

            Secondly, in the past pollsters used land line telephones. With a copy of the county voting records in hand, a pollster could ask his questions and then assign a likelihood of voting based on that person’s voting history.  A respondent who had voted twice in the last 4 elections got a .5 weighting factor compared with someone who always votes.  But how can the people be polled accurately if they have cell phones or won’t identify themselves?  If no person or address corresponds to the person interviewed, the pollster cannot judge likelihood. 

            What happens if the respondent lies?  Trafalgar Group seems to have circumvented this somewhat by asking how that person sees their neighbors voting.  This can be telling if the voter is leery about answering correctly.

            Turnout is a huge factor in elections.  Only about 2/3 of all registered voters vote in a general election.  If one side turns out more of its base, that, not an overarching popularity may win. Base voters usually don’t split their vote and so there is more of a coattail effect in recent elections with a lot of partisanship.

            To check their prediction, pollsters often conduct polls of people right after they have voted.  But here, the trouble seems to be that many folks are particularly elusive about sharing who they voted for right after casting a vote—voting in secret is a well-guarded right!  These Exit Polls have been wrong in about half of recent elections.

            There are indirect polling measures that can give a predictive picture.  If an incumbent is at 47% approval or above, they are likely to be re-elected.  But if approval falls below, to 43 or 44%, they are likely to lose; 50% they are sure to win.  Evidently about 3% of people will say, “I don’t approve of this guy, but I’ll still vote for him.”  An incumbent is a known quantity.  If he polls low approval or in the low 40s against a challenger, that says that even though there may be a lot of undecideds, the one thing they have decided is that they aren’t going to vote for the guy they are saddled with now. In the old days the Right Track/Wrong Track polling told a similar story but has become unreliable.  Evangelicals will answer that the country is on the wrong track even if their guy is the incumbent.  They are reacting to the general moral malaise, not the candidate.

            Independents are somewhat enigmatic.  They vote only half the time as partisans.  They consist of many differing types of attitudes.  At least 20% are “hidden partisans”—people who hide the fact that they are, say, Democrats, in the midst of their Republican friends to claim superiority in open-mindedness.  Or their father is a harsh man who wouldn’t stomach them registering with the other party. Other Indies have a weird stew of issues.  Some have a premier issue that neither party addresses, like cockfighting or feral hogs.  Some have R ideas about fiscal budget balancing with D ideas about social issues.  They are people without a party.  Some are anti-political and simply refuse to listen until the last minute.  Hence Independents don’t vote all the time.

            So this year?  Clearly there are a lot of Trump voters who won’t talk to a pollster and maybe not even their overbearing friends.  Trump needs to get his approvals up over 47%, however.  Turnout?  That’s a big unknown. 

            One final note that belies the usual media message that the polls are tightening as we approach election day.  It’s B.S.  By October 1 all but 3% of people have determined who they are going to vote for.  By one week away, it’s only 1% of people still truly undecided.  The reason the media says piously that polls are tightening is to to squeeze more ad money out of the candidates.  They make an enormous haul during election season and they are pumping everybody dry, to make a few more bucks, and pollster play along with the game by making it seem like polls are tighter.