Black Mamba
Member
I am certain Gravis' methodology is wrong because the filter they use to establish likeliness to vote is whether someone has voted in that caucus before. If you haven't voted in a caucus before, you are not considered likely to vote. This flies in the face of ~30% of Iowan caucus-goers in '08 being first timers. The +18 is an outlier. For the record, if you include +18 in that average, you get Clinton +6.4. Now, the chances of an accurate poll getting +18 from a population value of +6.4 with a sample size of say 500 is less than 0.01%. Gravis is immensely unlikely to be an outlier. Either it is wrong, or the others are wrong, and I think we have sufficient reason to suppose Gravis is wrong.
That's fine. I'm not opposed to eliminating polls with improper methodology, I'm against throwing out outliers which are different. Your post didn't differentiate the two at the time so without any other information, I'm not throwing it out.
I am only against people putting a single poll or pollster on a pedastol and ignoring other pollsters. This isn't a Hillary vs Bernie thing at all. Just want people to understand not to overlook the aggregate because they like a pollster and to also realize primary/caucus polling hasn't yet proven to be super reliable.