George,

The polls were way off in the Michigan Republican primary, but since all the candidates were white men the media took no notice. The average of all polls predicted that Romney would get 29% of the vote. He got 39%.

If we were to assume that the poll results were so far off because many of those polled were “lying”, the argument might go something like this:

Romney was a native son. Many intended to vote for him for that reason. But voting for a candidate just because he is a native son is not a substantial reason. When the pollster called, these “native son” types thought to themselves: “I don’t want this pollster to think I am superficial.” So they said they would vote for someone else even though they really intended to vote for Romney.

Respondents to polls do not contort their answer in this convoluted way.

They do not “lie” or try to deceive. I have read through thousands of open-ended materials from surveys and, believe me, people are happy to spill their guts to a pollster who they will never see again. Indeed, it is one of the few times that they can reveal their true feelings. It is from friends, co-workers and even family that people hide their true attitudes.

The “Romney effect” that I have just facetiously presented will never be brought up by the media or believed by anyone reading this. But the “Wilder effect” – which is just as imaginary – continues to be believed. Racism is still present in our society, but it does not manifest itself in answering polls. See my analysis of the “Wilder effect” at http://george.loper.org/~george/archives/2008/Jan/935.html

David RePass (Electronic mail, January 16, 2008)