Poll Questions and Goldfish

Harman has a good pick up on a CNN poll. And I just happened to be talking to someone yesterday who works on polling, as I did once upon a time. (I was at SUNY Stony Brook when it was setting up its political polling program, and actually worked the phones myself. I still have scars from jamming my pencil in my head after asking the so-called “knowledge” questions: do you know who the vice president is? Do you know if you have a pulse? that sort of thing…)

This picture has something to do with goldfish, I promise...

Anyway, I know I saw something on this once, and I’m wondering if anyone’s seen an updated version of a fake question poll, which involves not just asking questions (called controls) for seeing if people are paying attention or understand the issue, but in actually conducting an entire poll on the most ludicrous thing and seeing how many yeses you would get. I saw an article recently on an experiment involving measurements of gold fish brain wave responses to pictures. Now theoreticallythe measurements should always have been zero given that all the gold fish were dead before the experiment began. Or so the scientists say. In any case, they got a number of false positives showing some measure of response by the dead fish to photos. (Which may explain something about the ratings for Fox News…) And so, it’s a reminder not to take too much away from minor changes in measurements when conducting research focused on narrow results.

Anyway, as to the polling, I wonder how many false positives one could get seeing if people are in favor of eating elves, or how many find the taste of camera film heavenly, or some such. Maybe 2%? Maybe .05%, which would still be millions of people…