A theory about WPE and SEM for researchers and statisticians

[Sancho]

If I were a Smart Election Manipulator (SEM-pun intended) and I knew the typical sample size (ie power to detect effects) of the exit pollsters, I would set the manipulation (flips, drops, and fixes) so that the error would never indicate an effect outside of the normal (pun intended) error of polls and cover up the evidence of fraud and abuse! Sometimes, the SEM would be too small to change the election, but would always be hard to detect.

Of course the CPA's (Clueless Pollster Associations) would notice strange things and point to theories about the unexpected noise in the results (interviewer gender, reluctant responder, winner enhancement, etc.), but NEVER FIX THE PROBLEM of SME after years of confusion (2000, 2004, 2006) because there is never an absolutely identifiable problem within all the noise. In order to be "credible", the CPA's would simply change the results to match the expected results sans manipulation because they are chasing ghosts and can't figure out why the errors keep on appearing even when the methodology changes to fix the last sampling problem. They even tell the SEM's the plan for measuring before the election so that fixing the SEM's target is pretty easy and puts a smile on the face of FKR's (Friends of Karl Rove).

So, how does the CPA get smart and not remain part of the the world of CPA's? They can target some areas of sampling in SPECIFIC UNANNOUNCED places and times so that they eliminate SEM in that sample and catch the SEM'S! Do they want to do that? Let's poll the CPA's and see! Why not?

How do the WCT's (Wacko Conspiracy Theorists) use existing data to catch the SEM's if they exist? They focus on the unit of analysis (pay attention to a particular subset of data) so that the effect of SEM within that subset is less than the typical noise so that the manipulation is visible. This is made more difficult since the effects are hidden by not having access to precinct level exit poll data and confusing alteration of the final results to match an expected target. Meanwhile, the CPA's keep arguing that the WCT's are wacko's instead of joining forces to catch the SEM's!

The WCT's may prove successful in the end,

but my theory is that the CPA's will never find a good reason for the errors or a fix for the lack of accuracy, but that when a transparent election process is in place, the exit polls will magically have less error and need less adjustments at midnight!




:think:

[bleever]

1. Very interesting. Cross-posting this in the Election Reform forum

will probably get it more views.

(FKR's: :rofl: )