Late, very late, Saturday night I get a nervous text from a friend on a thread that generally talks smack about each others’ sports teams:
Jared: “Kev - what do you make of the Iowa poll? Are there chinks we should be worried about?”
Since I had been busy being a dad for most of the day and had been off line, I asked him to let me look into it.
Yep, there it was, the Des Moines Register’s latest cover screaming, “HARRIS LEADS TRUMP - 47-44.”
Of course, the first thing I always do when looking at a poll with an outlier finding is to immediately understand the sampling weight the pollsters used. But since the Register didn’t bother to publish it, I nearly immediately dismissed it. It doesn’t matter if they do a poll and find a certain outcome. If you do a poll that isn’t reflective of the population of likely voters, then the predictive power of said poll is meaningless.
For example, if they had polled 800 inner-city black women exclusively, 47-44 for Harris would be a death sentence for her.
But........