Atheism doesn't poll well. That begs the question of....... why?

[Act_of_Reparation]

18. The poll has two serious experimental flaws...

...anyone with any training in social or behavioral sciences should be able to quickly identify.

First, the poll is asking people to self-report their attitudes. When asked, people will invariably present their ideal selves, their selves as they see them, or how they would like to be seen by others. This confounds the accuracy of the test.

Ideally, you'd test people implicit biases, the attitudes they aren't consciously aware of. Racism tests, for example, that ask people to self-report their racial attitudes invariably show a miraculous dearth of racism among respondents. Racism tests that prod at peoples' racial attitudes without them knowing it tend to present different results.

This isn't a jab at Gallup. Their polls are well-designed and generally useful. However, there are limitations to their usefulness that should be recognized when discussing these topics.

Second, there is the context of the question. It asks if voters would vote to elect an atheist candidate nominated by their party. The question assumes the party has put forth an atheist candidate, and that candidate is running against the opposition party. Because party identification is so strong, this is virtually a meaningless question. Of course party-affiliated voters are going to to vote for their party over the opposition.

The problem here is that the parties don't nominate atheist candidates. A more meaningful question would ask if affiliated voters would cast their votes for an atheist versus a person of faith in the primary election. It would help demonstrate what voters would do once the party identification variable-- a strongly confounding variable--was removed from the equation.