Posts Tagged ‘polling’

Things are not always as simple as they seem in surveys

The New Hampshire Primary was a cautionary marketing tale. It shows us why surveys (polls) can be very misleading.


The polls missed the Clinton victory by a mile yesterday, yet they were right on target with the McCain victory. Why?

No one knows exactly what happened for sure yet, but several possibilities illustrate some of the pitfalls in marketing research.

First of all, there is the assumption that you can believe people when you ask them what they have decided to do in matters of simple choice like whether they have a preference for Coke or Pepsi, or which candidate they favor. In most cases, this is a reasonable assumption, and most of the time, polls are accurate.

Other times, what looks like a simple question is not. It turns out that in the minds of New Hampshire voters, McCain versus Romney was one of these simple questions. Though polls got it right, both on preference and amount.

On Clinton vs. Obama, not so simple, on many grounds.

Here are some of the possibilities that are yet to be investigated and quite possibly never definitively determined.

(1) What if people’s minds are not made up? Not a problem, at least as far as taking a snapshot. While that would make prediction difficult, with these people would have turned up as a large number of undecideds. This is not what happened. So, the pundits say, the people didn’t make up their minds at the last moment. But what if they were undecided, but didn’t know it? What if they thought they favored one candidate over the other, but this was a weakly held preference and they were easily swayed by last-minute remarks that they heard on the radio or from their friends, on the way to the voting booth?

So, when people hold an opinion, the strength of the opinion is just as important as the opinion itself. Sometimes opinions can be held so weakly that they might as well not be an opinion at all. But that’s not the way it’s experienced by the person.  In the absence of a challenge, it’s often experienced as an opinion that is pretty firmly held.  So it is of no use to ask the person in the survey, “on a scale of 1-10, how strongly do you hold that opinion?” It’s also equally nonsensical to ask people what it would take to change their minds.

So, the first possibility is that they changed their minds at the last moment, perhaps even in the voting booth, but did not know until that moment that their previous choice was weakly held.

(2) People may have in fact held a very strong beliefs, but changed their minds quickly and decisively when they saw, for instance, Hillary Clinton cry the day before the voting. This may have been too close to the voting for the polls to have picked up. But, I don’t think so because it wasn’t picked up in the exit polls either.

(3) Here is the most intriguing one for me: What if racism isn’t dead? Well, of course racism isn’t dead. It didn’t evaporate just because Obama it isn’t running as a black candidate. In that case, many people (it would only have to be about 10% to account for the data) might not want to tell a pollster that he or she was not voting for Obama, even though they thought that Obama was the best candidate. They might be feeling guilty, or they might expect disapproval, or they might just experience a vague sense of unease about Obama that makes them feel vaguely uncomfortable. Or, they may be worried about being perceived as racists even though they are not. So, when asked, they blurt out “Obama,” and maybe even mean it at the moment. But in the privacy of the voting booth, that vague sense of unease — which, I suspect, is the main way that racism is experienced among people who are just mildly racist, especially those who are ashamed of it or are unaware of it— rears its annoying little head and causes a private little finger twitch that never gets reported.

This is a well-known, and well-documented effect called the Bradley Effect, or the Wilder Effect, where blacks often poll with stronger support than they ultimately show in the polling booth. Even exit polls often say that the politically correct person won, but when the votes are counted, the politically correct is not the politically erect.

It will be interesting to see what happens in the primaries versus the caucuses. For instance, in the Iowa caucus, people had to stand up in front of their peers openly and declare their allegiance, unlike in the New Hampshire secret ballot.

So the marketing lessons here are:

  • “Take a survey [poll]” isn’t always the answer.
  • Simple questions are often not.
  • Asking questions gets answers, but not necessarily the truth.
  • People don’t always know what they believe or how strongly they believe it.
  • People often have beliefs that they don’t know they have.
  • People can’t even predict their own behavior.
  • People often say things based upon what they think you want to hear.
  • Distrust after-the-fact explanations from pundits, including me.

What I am really saying here is things are not as simple as they seem. If you have a product with any degree of controversy you are navigating a mine field when you try to assess public opinion. Even professionals don’t always know what they are doing.