2016 Postmortem
Related: About this forumLet's have another lesson in polling, shall we?
First, as stated before, the goal of polling is to create an accurate representation of a population group: frequently (but not always) the population who will vote in the upcoming election.
There are generally two population sets: REGISTERED VOTERS (RV), who are anybody who is registered to vote and LIKELY VOTERS (LV) a subset screened for those who indicate an intent to vote. Beyond that, the population set will usually be subdivided further to reflect accurate representations of age, race, and Party identification.
Now, it has been suggested by supporters of Bernie Sanders that polling which indicates ahead of Sanders in all national polls, and in polls for all States except NH (where currently polling appears split) is inaccurate because it is not picking up a mass of "disaffected voters" who don't participate in the political process. Here's the problem with that argument:
1) If they are registered to vote, then they are eligible to be picked up by an RV polling sample. Yes, there are issues with phone accessibility or willingness to participate, but SOME element of the group should be reflected in polling
2) if they are NOT registered to vote, they won't be picked up in either sample, but the issue is moot because, if no concerted effort is underway to register them (some deadlines have already passed), they won't have an impact on the outcome. I'll observe that the Sanders campaign has talked about how many contributors and volunteers they have, but I haven't seen any numbers on new voters.
Response to brooklynite (Original post)
Cheese Sandwich This message was self-deleted by its author.
Android3.14
(5,402 posts)The dirty secret of polling is that anyone can determine the criteria by which you choose your population from which the pollsters draw their responses.
Your oversimplification is misleading.
I am looking forward with hope to the caucuses.