Quantification and the Vote

Pollsters face a myriad of problems: so many people not using landlines, so many people (less than 10%) actually answering the phone for pollsters, that you’d have to think one of these years–this year above all, given that many of those that voted (especially the younger voters that upped their vote in the US) are those least likely to answer the phone they don’t have (the old landline) or respond to a robocall—the pollsters would have it wrong, perhaps not by a lot, but by a few points to make a difference. Perhaps that’s what Romney let himself believe, but after much work when I was doing a Public Policy M.A. and an undergrad on statistical models, it’s still amazing, despite the above seeming major problems, all the models got them right. Except the Republican “wishing makes it happen” model, which keeps failing them on a quite regular basis.