Aw, c’mon people. Just tell us what you’re thinking. Please?
We go to all this trouble to build an entire industry based on figuring out what’s going on in your noggins. We create complex data tools to parse the survey results by every conceivable demographic and geographic slice imaginable, all to explain who you’re voting for or what you’re likely to buy. People pay good money to know that.
And then it all goes wrong.
One of America’s favorite post-election sports is to figure out which industries and interest groups will be winners or losers when the new president/governor/mayor takes office. While opinions differ on who benefits, this year there’s been rare unanimity on the one industry that decidedly took a hit — public opinion research.
Never miss a local story.
Even those in the industry are conceding the point that this wasn’t one of their better performances, even as they cite the usual caveats — polls only capture attitudes at one specific instant among one group of people, it’s still within the margin of error, it’s an inexact science (and more art than science), they call a lot more right than they miss (no one noticed that the Washington governor’s race was as uncompetitive as the polls suggested), upsets and surprises are bound to happen, etc.
What’s concerning to them, and those who make decisions based on what polls turn up, is that this wasn’t an isolated incident. Whether it was the 2014 congressional election, or the Kentucky governor’s race, or internationally the Brexit vote, the polling business has had some high-profile stumbles.
All right, fine, that’s been going on since the days of “Dewey Defeats Truman.” More worrisome is that the recent misses reveal some longer-term challenges to the entire category of polling, market research and opinion surveying that may prove an even bigger headache in the business realm than it’s been in the political world.
Some of the problem is structural. The death of the landline telephone makes it harder to find interview subjects in sufficient numbers to build a reliable sample. Other approaches, such as online surveys, are easy to game.
A lot of it is attitudinal. Many people don’t have the time or patience for a survey, or they find being asked their opinions intrusive and an invasion of privacy. Some interview subjects don’t want to give an answer that might depict them in an unpopular light, even if it’s an anonymous survey. Others are suspicious of just how impartial the poll or its sponsors are (the practice by campaigns of trying to plant criticisms of opponents in voters’ minds through the wording of questions hasn’t helped). Still others lie just for sport.
The reliance on polling has led to some lazy and sloppy practices for some businesses, the media especially. Less emphasis on polling and more on field reporting would help this industry avoid some occasions of embarrassment when reality refuses to conform to predictions.
But other businesses don’t have an option of just walking away from market research. A lot of money is riding on the decision to design and make a product or service, and to build the accompanying supply, sales and distribution chains to get it to the buyer. If it turns out that market research was wrong about the propensity for someone to buy what that business is selling, that’s an expensive mistake to make. Get it right, and you get the string of products that made Apple the success it is. Get it wrong, and you get New Coke.
Getting the kind of research that makes those expensive bets less of a gamble is tough, which is why so many companies are experimenting with so many techniques to get insight into consumer behavior. Sometimes that still involves asking them. It’s rare to get a receipt in a store or restaurant these days that doesn’t have some sort of enticement to participate in a survey.
Increasingly, though, the emphasis is on finding out what you’re actually doing instead of what you’re saying. One of the most valued commodities that companies have is your data — what you bought, what websites you visited. An entire subindustry has sprung up to mine that data for useful predictions on where you’ll spend your money next.
Political polling may need to go that route if voters are hard to reach and unreliable when you do. No predictive system is going to be completely accurate in politics any more than it is in consumer-product marketing or gambling. But the industry can spend time either changing its methodology or explaining why, in the horse race of an election, the betting favorite finished out of the money and out of office.
Bill Virgin is editor and publisher of Washington Manufacturing Alert and Pacific Northwest Rail News. He can be reached at email@example.com.