Quote:
Originally Posted by monkeyman420
I worry about pollsters not polling rural districts, or polling the areas where Trump won by only a small margin where previously it would have historically gone democrat or at least leaned democrat.
|
As an FYI, most pollsters make available their demographic and crosstab info available. We can see racial, educational, gender, etc splits. I don't think they're avoiding rural areas, based on that info that is available.
I'd also ask the follow up: why would they ignore rural areas or focus on areas where Trump won by a small margin that historically went Dem? Understanding that their goal is to get the right result, rather than some odd goal of distorting results and being wrong, why would they do something they know full well would cause sample bias and put them in position to get it wrong? Remember, this is a business.
Quote:
Originally Posted by monkeyman420
My skepticism is more rooted in feelings than facts, and I have no issue acknowledging that. I really really hope the outcome matches the polling.
|
That okay, too. A lot of us have fears of another Trump term. There's some PTSD from folks who mistakenly understood 2016 polls to indicate a done deal for HRC. So I get it. And I think some skepticism is healthy, and we should all vote no matter what. Vote like we're down 6 in the 4th quarter.
What can ease that skepticism is A) understand polling and what it tells us, but more importantly what it doesn't - polls on September aren't predictive, they're snapshots; B) compare polling in this race to that of the recent elections - Presidential and midterm - and see if the gap and support levels are different; and C) watch how polling moves over time - is the race stable, are news events causing shifts in voter preference, are there an increasing or decreasing volume of undecided or 3P voters.
And, as always, look at aggregates. This will help account for single-firm polling errors. A bonus is if they rate pollsters, like 538 does, and detail which direction a certain pollster has historically had a rate of bias towards so that you can revert to the mean.