re: #104 goddamnedfrank

I’m looking at the new Pew polling breakdown, their likely voter breakdown not only seems to under sample Hispanics by 2% compared to their own LV model, it also list the actual likely Hispanic vote as - -, as in no breakdown, at all. It’s a single, glaring omission in an otherwise thorough looking report. In the same report alongside this “data” they list the previous poll’s Hispanic breakdown as 77% to 22% but give no breakdown for this poll. It’s weird, on the surface it looks pretty unprofessional, like they just dropped the actual Hispanic numbers from consideration, lost them or refused to publish them or include them in the final calculation. I know that Obama didn’t help himself in the debate, but I’m seeing some bending over backwards with methodology to reinforce that point.

I wonder if it’s just that, for some reason, this poll sample didn’t include enough Hispanic voters to make an actual breakdown feasible. (i.e. if you only had 10 and they all were Obama voters, it would give you 100%, but you wouldn’t want to report that 100% of Latinos will vote for Obama, because that would make you look like an idiot).

In which case, it would still be shoddy polling, just not intentionally or maliciously so, as it looks like you’re thinking it may be.