What a week!... Some folks warned months ago that Donald Trump would be elected President WHEN the Chicago Cubs won a World Series (meaning, at the time, when Hell froze over!). Anyway, a quick request if anyone cares to take me up on it:
There's been a lot of discussion/criticism of prior polling since the election outcome... often proclaiming the polls surprisingly inaccurate. In fact though, as best I can tell (and others have noted), IF you take into account margin-of-error, the polls were pretty darn on-the-mark!... it was THEIR INTERPRETATION by press pundits and others that was off-base and poorly-understood.
In any event, I'm curious to hear, from anyone involved who can speak to it, just how, representative, randomized national/state polls are even conducted these days when such a reduced percentage of the population have landlines (which I presume were the staple for most polling in the past). Just what are the basic procedures now for attaining a random, but representative sample?
If you can answer that in a few lines in the comments great, OR, if someone would like to write something up as a 'guest post' here about the mechanics and difficulties of current polling methods, let me know what you have in mind [sheckyr at a gmail account]; I'd be interested.
----------------------------------------------------In any event, I'm curious to hear, from anyone involved who can speak to it, just how, representative, randomized national/state polls are even conducted these days when such a reduced percentage of the population have landlines (which I presume were the staple for most polling in the past). Just what are the basic procedures now for attaining a random, but representative sample?
If you can answer that in a few lines in the comments great, OR, if someone would like to write something up as a 'guest post' here about the mechanics and difficulties of current polling methods, let me know what you have in mind [sheckyr at a gmail account]; I'd be interested.
Meanwhile, this week's math-mix:
1) Looks like the folks over at "DataGenetics" may have had a little too much time on their hands one day ;-):
http://datagenetics.com/blog/november12016/index.html
2) A post on pseudoscience and the 'garden of forking paths' from Scott Alexander for the tinfoil-hat crowd:
3) Mathologer videos are always worthwhile, and the latest is no exception:
4) The gender bias in math journals:
5) Simon Singh reviews Alex Bellos' latest:
...and if you like 'hat' problems Alex had this good Monday workout for you:
...and still more examples from his book:
6) Ben Orlin's latest:
7) John McGowan considers Facebook, algorithms, echo chambers, fake news, and the election:
http://math-blog.com/did-facebooks-news-feed-algorithm-elect-donald-trump/
8) And the latest from Evelyn Lamb, more on Ramsey Theory:
https://blogs.scientificamerican.com/roots-of-unity/moores-law-and-ramsey-numbers/
http://math-blog.com/did-facebooks-news-feed-algorithm-elect-donald-trump/
8) And the latest from Evelyn Lamb, more on Ramsey Theory:
https://blogs.scientificamerican.com/roots-of-unity/moores-law-and-ramsey-numbers/
Potpourri BONUS! (extra NON-mathematical links of interest):
1) Yanis Varoufakis on capitalism and democracy, from TED Radio Hour:
2) "Squirrel Cop" was one of those classic (and hysterical) "This American Life" episodes (from 2013). They re-ran it last weekend:
https://www.thisamericanlife.org/radio-archives/episode/510/fiasco?act=3
------------------------------------------
Seems clear that current methods/options for polling are especially vulnerable to a variety of problems. In fact I’m even more impressed now with someone like Nate Silver’s ability to ‘massage’ or tweak meta-poll data to attain the level of accuracy he does!
------------------------------------------
ADDENDUM: No one has yet offered to do a guest post, but a reader did send me a few pertinent links about election polling, the two most helpful being these:
No comments:
Post a Comment