Environmental lead (Pb) and crime
At Mother Jones, Kevin Drum has an in-depth article on the hypothesis that environmental lead is a likely root cause of the increase in cri...
Polling geeks in the news
In today’s San Francisco Chronicle. I’m wedged in there between Mark Blumenthal and Nate Silver.
Topics:
I’m surprised they didn’t mention Andy Tanenbaum.
“There’s an issue here – not all polls are created equal. Some have more rigorous methodology than others. You lose something by simply offering aggregations,” said Frankovic.”
That assertion is exactly wrong. The evidence is in the meta-margin graphs, which should show spikes if that was true. What you see is the reverse. There may be polls out there with bad results, but averaged out with other polls there is no difference between bad results and the natural margin of error you get with any poll.
The reason is that even bad polls have signal even if they have noise. Aggregating polls merely sums up both signal and noise into a larger sample set. The signal gets accumulated with the signal from the other polls, while the noise gets washed out.
If you want to see this in action, turn on your cellphone.
re: the SF Chronicle
This is the main newspaper I ead, and without it, I might not have visited this web site. I checked you out this morning, and I really like the way you organize and analyze the polling information. As an avid Obama supporter, I am sometimes a bit dismayed when I read that the polls are close, and I appreciate your presentatioin of the information, which has helped to calm my fears somewhat. Great work!
I hadn’t been to this site until I read that article in the chronicle. Congrats! and thanks for all the work you do for poll junkies like me.
Could you make a post or point me toward, something that talks about the allocation of the undecideds. As a “poll of polls” can tell us the state of the race, so too can it tell us the state of the NUMBER of undecideds. Everything Ive studied show they break toward the person behind. Any thought to that, and if history shows that, why?
Good on you, Sam. Dunno. Intuitively, lol, I feel far more comfortable with methodology.
Thanks as ever…
This may seem obvious and or inane, but does it make any sense to combine in a statistically valid fashion electoral vote projections, across all the poll analyses. (RCP, 538, polster.com, princeton, etc.) ?
Jefflz,
I can’t see how you would reasonably combine poll analyses. They are all based on the same set of polls, so one analysis doesn’t have information that others don’t. Things like assumptions (and goals) are not consistent across analyses so any combination would not be self consistent.
Michael S, when interviewed I did mention Andrew Tanenbaum and his creation, electoral-vote.com. As far as I can tell he pioneered this peculiar hobby. But reporters have to choose an angle. Perhaps simple data aggregation (as opposed to the meta-analysis/projection approach) didn’t seem interesting enough.
jefflz, there are meta-meta-analysis sites. One is 3bluedudes.com.
You might like to peruse this compendium of similar sites.
Sam, I just realized what you should do, is show in some way the odds of winning the different likely battleground states. It would be really cool to know the most likely electoral scenario, and would be a real draw for the site!
Sorry for the off topic question, Sam, but could you comment on the latest video from the Defenders of Wildlife group on polar bears? I have given them quite a bit of support to get the aerial hunting of wolves video on TV at your suggestion (and hope that it has made a difference). Now they are asking about money for this new video and I am wondering if I am better off to continue to give money for the wolf one since it is so effective.
Vicki, my feeling about the Defenders of Wildlife Action Fund is that they were quite influential in driving perceptions of Governor Palin – but that those perceptions are now driven. The polar bear business seems repetitive. Nothing against bears.
As I’ve stated before, I don’t think there is much that either side can do to influence the Presidential race. There could be a surprise, but it’s highly unlikely. Your most influential investment in the future will be affecting the composition of next year’s Senate. Therefore my interest in the races I list at ActBlue (and for Republicans, the NRSC).