Reply #90: "Weighted" and "corrected." Good questions.

[IndyPriest]

There are people around here smarter than I am about this stuff. But from what I have read, "weighted" refers to the fact that exit polls are built to INCLUDE population distributions: age, race, gender, party ID, etc. Supposedly, good exit pollsters know what those distributions are supposed to be based on census data, past voting patterns, etc., and they purposefully aim at getting their data to match those distributions in order to get as accurate a snapshot of the vote patterns as possible. This "weighting" is one of the reasons exit polls have a much smaller margin or error than "survey polls." Some of the questions about this year's exit polling is about whether they were "weighted" correctly.

"Corrected" is a whole different horse. One part of the process of analyzing the exit polls is to compare them to "actual" votes and, as an election day goes on, begin to feed the actual votes into the exit poll numbers to "correct" for any (supposed) difficiencies in the exit polling. (I put actual in "" because, as we know from this election, trying to discover what the "actual" votes were is a nightmare.) But the theory is that if an election is clean, comparing actual votes to exit polls is supposed to "confirm" the accuracy of the voting by "correcting" for any remaining problems that may have arisen because of "incorrect" weighting.

Does that help?