From: J. Andrew Rogers (andrew@ceruleansystems.com)
Date: Sat Jun 04 2005 - 11:36:28 MDT
Mark Waser wrote:
> The 2004 exit poll statistics from Pennsylvania, Ohio, and Florida tend to
> indicate otherwise with all three being several sigmas out of whack with the
> "vote results."
Only with the most ridiculously naive analysis of the polling internals, suggesting that the
polls were "analyzed" in support of a desired outcome rather than for any predictive value.
That this was the case was noted online within an hour of those polling numbers being
released, and reasonable analysis at that time gave numbers that matched the final results
just fine. The outcome was fairly certain to the quants before the election even started and
only self-deluded political groupies thought otherwise. My own publicly stated quant
analysis in early 2003 for the popular vote (51.5 +/- 1.0 for Bush IIRC) was surprisingly
accurate. The strongly Democrat leaning audience thought the prediction was mad without
any regard for how I arrived at it because they found the result uncomfortable. Denial about
reality is not healthy; much better see reality as it is and work from that assumption.
Conspiracy theories are fun, but reality is calling and I have work to do. This is off-topic
anyway.
j. andrew rogers
This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:22:56 MST