30
Sep 04

the trust issue

Proven By SciencePost a comment • 247 views

the trust issue

[ps i don’t think i made a remotely clear point in three tries in the comments to barry’s post below! (memo to self: don’t post when you got up at 6am fact-checking footnotes), and i think i’ll leave it for another, more clear-headed day]

plus also anyway: the real fatality in the upcoming us elections isn’t trust in science (probably holding its own for now) , or in the media (oops) or in politicians (haha) but in OPINION POLLING!! The heartblood of our mass information retrieval system!! oh no!!

gallup in particular is taking lumps for invalid modelling – ie preselecting its sample to reflect hope or prejudiced guesswork re the spread of voter support* ===> of course gallup’s critics are highly partisan themselves (ie if the model’s wrong, how do they know it’s not wrong in the OTHER DIRECTION) but the deep point stands: small-sample polls are a guide “all other things being equal”, ie when the underlying pattern is fairly stable and shifts are provisional shallow swings —> but if the situation on the ground really is radically unusual (= huge surges in new voter registration for or against a given candidate; or = very stable heartland voters on either side plus untypically volatile undecideds, say)

*(one telling datum: a huge fall in the number of folks who actually give anwers in phone-polls, as opposed to not picking up, or hanging up w/o giving an answer, from c.80% ten years ago to c.25% now [figures from memory] ==> anyway it means the people staying on the phone are self-selected as choosing to register an opinion publicly, if not necessarily truthfully: what this says abt their politics i don’t know, however)

Add your comment

(Register to guarantee your comments don't get marked as spam.)


Required

Required (Your email address will not be published)

Top of page