This is not debatable. Ted is right. We've measured it, too. We can show you the differences in panel interviews vs. IVR vs. live-caller. We didn't abandon it to piss mediates off. We abandoned it b/c it was the right decision.
People want to know what I mean re: New Jersey.
Very simple.
In some NJ precincts, you really need certain people on your side to GOTV. If they are not themselves crooks, but you are a crook who wore a wire to take out their mentor, you might not get their 100%.
Feel me?
Very simple.
In some NJ precincts, you really need certain people on your side to GOTV. If they are not themselves crooks, but you are a crook who wore a wire to take out their mentor, you might not get their 100%.
Feel me?
It's also a matter of who is more likely to actually vote. All crossed with voter files. Online panel was more likely to vote and, while that looks relatively small, even those small adjustments can have an enormous impact on outcome. Tiny modeling errors could blow it all open. 
What you should find MOST interesting from the above graphics, is the gap in the percentage of people from phone files who didn't vote vs. online respondents. Who was telling the truth more often?
And understand... most are not checking vote history. They are not.
And understand... most are not checking vote history. They are not.
