Pointing out the flaws in worthless polls would be a full-time job, if anyone were willing to undertake it. Today’s example comes from CNN and ORC. News organizations report on polls like this one as though they meant something; this one will be popular with media organizations because it shows President Obama with a six-point lead over Mitt Romney among likely voters. But a cursory review of the poll data shows that if it means anything at all, it is good news for Romney.
To begin with, the CNN/ORC poll follows on the heels of another survey by the same organizations two weeks earlier. The findings–based, obviously, on a small sample–suggest that four percent of voters changed their minds about the Democratic Party over the space of two weeks:
That is, of course, ridiculous. What happened, rather, is that the second sample contained a lot more Democrats than the first one.
If the CNN/ORC poll contains any significant finding, it is this: independents favored Romney over Obama by 14%. That’s right, 14%:
If Romney carries independents by anything like that margin, the election won’t be close. The pollsters found that virtually all Democrats are voting for Obama and virtually all Republicans are voting for Romney, so someone with quicker math skills than mine can pretty quickly figure out how much they had to over-sample Democrats to come out with Obama winning by six points.
Really, this whole exercise is getting tiresome. Yet one question remains: where is the poll that over-samples Republicans?
Image courtesy of Shutterstock.
UPDATE: A commenter calculates:
to get to those totals the following breakdown works…
D – 38.
I – 36.
R – 26.
that’s D+12, right…