Reply #70: Sounds good but not true... [View All]

[anaxarchos]

Think about it. Exit polls are unique in two ways:

1) They carry a huge amount of demographic data with them (the survey questions). Overcoming one skew produces another. You have to match in a general way the remaining demographics (age, income, education, religion) or your precinct sticks out like a sore thumb. The problem you point to is easily detectable.

2) They have a historical element. They are "corrected" election to election to "compensate" for what you describe. In order to be a factor, all gun-shop owners (and many many more to be significant) would have to decide to not answer ALL AT ONCE (i.e. in this election only).

It is also not true that these types of issues change election to election (WPE). The opposite is true. They are remarkably stable. Otherwise, exit-polls couldn't even come close with such a small sample.