Comment by JumpCrisscross
1 day ago
> What if you're suspicious of all polling regardless of whether it agrees with your preferences or not?
I’d still call that statistical illiteracy. Polling, as a cohort, contains information. It’s dispersed across polls and concentrated among quality pollsters.
It’s never definitive. But someone concluding that all polling is useless because the statistics are hard is sort of analogous to someone rejecting cosmology because we haven’t actually been to Andromeda.
> what good are any of them?
If I want to know, today, who will be in power tomorrow and what policies they could pass that would be popular, polling is useful. If I want to know what issues I can build a coalition around, and which to abandon because the people most passionate about them cannot bother to vote, polling is helpful.
> rather than anything about the true preferences of the population
They’re telling you how people think when they communicate and act. What is in their heads is unknowable. At the end of the day, I care how they will vote (and if they will vote) and if they will call (or are even capable of calling) they’re elected if pissed off or enthralled. Everything else is philosophical.
At the end of the day, whether by poll or advert, information is introduced to a population in a biased form because it’s promulgated by biased actors. Knowing which way that bias is trending and resonating is useful.
> I’d still call that statistical illiteracy
It am suspicious of polling because I have a decent understanding of statistics. That is the opposite of statistical illiteracy.
> But someone concluding that all polling is useless because the statistics are hard is sort of analogous to someone rejecting cosmology because we haven’t actually been to Andromeda.
That isn't the argument being made. Nobody said it is "useless". I said I was "suspicious of polling organisations". Polling can be and has been used to manipulate public sentiment.
Therefore it is prudent to be suspicious of any polling.
> If I want to know, today, who will be in power tomorrow and what policies they could pass that would be popular, polling is useful. If I want to know what issues I can build a coalition around, and which to abandon because the people most passionate about them cannot bother to vote, polling is helpful.
That's fair in the context of, you're a political operative who is trying to enact specific policies as your occupation and you therefore have the time to go through and carefully inspect numerous polls to derive a well-rounded understanding. But that's also quite disconnected from how polls are typically used in the public discourse.
Ordinary people don't have time to do that, so instead political operatives will commission a poll to get the result they want, or find one from a reputable pollster who unintentionally made a phrasing error in their favor, or just cherry pick like this: https://xkcd.com/882/
And then use the result to try to convince people that the public is actually on their side and it would be ineffective or costly to oppose them. Which, unless you have the time to go carefully read a hundred different polls to see whether the result is legitimate, means that the sensible strategy is to give polls no weight.
Or to put it another way, on any politically contentious issue there will always be at least one poll saying X and another saying not-X, which means that in the absence of a more thorough analysis that exceeds the resource availability of most members of the public (and even many legislators), neither has any information content because the probability of a poll existing with that result was already ~100%.