Reply #92: You conveniently ignore the analysis which refutes you

[WillE]

Your pre-election challenge to find a PA pre-election poll to match the exits was a perfect illustration of why no one believes you anymore.

1) Besides being a ridiculous challenge, it was blatant cherry-picking.
2) But I fooled you. I gave you more TIA information than you asked for.
3) It showed that the unadjusted exits had a lower margin.
4) It showed that the final pre-election trend was to Kerry at 50-45.
5) A reasonable allocation was applied to the undecided 4%.
6) Final pre-election LV polls understated Kerry's projected share in that high turnout election of 22 million new voters. The larger RV poll superset would surely have indicated a bigger margin for Kerry everywhere, not just in PA. In 2004 the pollsters underestimated turnout.
So a 51-44 or 52-43 split was very feasible.
7) But undecided voters need to be allocated.
8) It was a 53-46 projection using the 50-45 LV poll(conservative)
9) It was 54-45 or 55-44 in the RV. Kerry did 2% better in RVs (see the pre-election RV/LV graph).

THIS SIMPLE EXERCISE IN BASIC LOGIC GAVE AN EXACT MATCH TO THE EXITS.
NOTHING IMPLAUSIBLE HERE. JUST BASIC ARITHMETIC ON THE BACK OF AN ENVELOPE.
BUT YOU PREFER NOT TO CONSIDER IT.

YOU WOULD RATHER WAX POETIC ABOUT "SURPRISE".
YOU NEVER FAIL TO DISAPPOINT. NO SURPRISE.
YOU NEVER CONSIDER A RATIONAL ANALYSIS INCOMPLETE PRE-ELECTION AND EXIT POLL DATA.

You prefer to get into interminable pissing contests - even when you are drowning.
Could you name one analyst or political scientist who agrees with you at this point?
How come we have never heard of even one "expert" lauding your work?
The graphs debunked 16 myths that you helped propagate.

Your bag of tricks have included
Swing vs. red-shift
False recall
Slow drifting fog
Surprise
Retrospective bandwagon effect
Lousy exit polls
Cherry-picked pre-election polls
and on and on...see the pattern, Mr. Other?
It's all faith-based mumbo jumbo.

On the Other Hand, you never consider factual data such as
Approval ratings
Undecided voters
RV vs. LV polls
Net Uncounted votes
Correlation analysis
Late votes
Mortality
Voter Turnout
New voters
Returning voters
Third-party voters
Phantom returning voters
Sensitivity analysis of alternative scenarios
Expected value
Correlation analysis
Monte Carlo simulation
Probability analysis
Recursive True Vote Models
Aggregate state exit polls
Forcing exit polls to match the recorded vote
....
In other words...you never do an analysis.
But I am repeat myself.

Not many tricks are left in that bag of yours, eh?