Quote (low-ki @ Sep 15 2016 10:06am)
new daily LA times poll
trump 47.2
clinton 41.3
I don't know why people keep referencing this, this is actually a rare example of when it's worse to have bad information than no information at all. For starters it's not a poll: it's a combination of a panel (which contacts the same respondents continuously) and a poll, but it combines the worst aspects of both and the best aspects of neither.
The worst part about it is that their original sample that they're now locked into (absent adding more respondents to it over time) is hilariously biased towards Trump: it has way too many Trump supporters, and for whatever reason they use a 1-100 scale of enthusiasm on top of that, which further skews the numbers towards Trump when votes don't actually count like that in this country. It's really a massive failure in the undertaking of measuring public opinion, which is a real shame because I was under the impression it's conducted by some of the same people that managed the IBD tracking panel in 2012 (an actual panel), which was at least moderately successful.