In his book, Code Red: Computerized Election Theft In the New American Century (post 2014 edition), Dr. Jonathan Simon points out and debunks the rationalizations (see below) used to explain "Red Shift" discrepancies:
- votecounts that don't match exit polls, and
- Republican candidate sweeps or narrower-than-expected losses accompanied by wide-margin victories of liberal ballot measures
But in the end, he says, it is not a matter of logic and reason (p.35):
Simon analyses and disposes of these rationales (p. 113ff, looking at the 2014 "rout"):
Finally, there is religion. I don't mean here belief in a deity but rather a secular religion of equal intensity. To illustrate I'd like to recount an experience I had at a national conference on media reform in 2007. At a panel session I publicly asked Cornell Belcher, chief pollster for the Democratic National Committee, a question about poll-votecount disparities and the red shift. Having stated flat out that the red shift cold not be attributable to any problem with vote counting (ie., election theft), he then added this curious observation, which I give from memory: "You know, it's odd but we have the same problem with our own internal polling: in important races, when our polls show our guy [the Democratic candidate] up by 10%, we've learned that we need to regard the race as a dead-heat toss-up."
Well. What exactly are we to make of that? Internal polls are designed not for political salesmanship but for maximum accuracy; they inform the party where support is needed, not needed, or likely to be wasted -- where vital campaign dollars should and should not be spent. When such internal polls are consistently "off" in the neighborhood of 10%, all sorts of alarm bells should be ringing and ringing loud. It takes a religious belief in the sanctity of an entirely unseen process, and everything we so desperately want that process to stand for, to ignore those bells -- to not even hear them. After I pointed this out, Cornell Belcher then restated flat out that the 10% disparity between his internal polls and election outcomes could not possibly be caused by election rigging. It felt like something out of Inherit The Wind, or perhaps Elmer Gantry: "Brothers! Sisters! Do ye believe?!"
- Low Turnout -- There is no reliable direct measure of who shows up, simply the assertion "Dems lost because Dems are fringe voters and don't show up." Then why the exit poll and initiative/candidate discrepancies?
- Voter Suppression -- Exit polls measure the voters who actually show up, not those who are disenfranchised by suppression.
- Dark Money -- Analyses of 2014 financing show a Rep advantage of only 3%, and major $$ advantages did not correlate with victory in many key races (i.e., law of diminishing returns).
- Gerrymandering -- It has no effect on Senatorial and Gubernatorial elections, which were rife with red shift, and can't explain the exit-poll/vote-count and initiatives discrepancies.
- Skewed polls -- i.e., "Dems lose because pollsters oversample Dems who then don't show up" -- circular reasoning based on an assumption of votecount accuracy. So polls are now being designed to "better match" votecounts by (a) using the "Likely Voter Cutoff Model" for selecting the sample, disproportionately eliminating Dems (p. 250ff), and (b) adjusting exit polls to match votecounts after the fact (i.e., assuming the polls are wrong) and then using the resulting "demographics" to weight the next election's sample.