Frank Bokhorst writes:
>As Dennis pointed out in an earlier mail, many
>of the issues were not statistical. A better term
>is perhaps >methodological", but so be it. One
>cannot always deal with these matters in isolation
>when "real life" examples are discussed, and the
>value of such examples is actually the integration
>of information.
It would be valuable to record all of the issues of interest to
Statisticians. Here's a quick list off the top of my head. I'm not looking
to re-open these issues for debate. I just want to summarize what I think
were the statistical concepts brought to our attention by this election.
Outlier detection: was the number of votes in Palm Beach County for Buchanan
unusual. If so, by how much?
Outlier detection again: Was the rates of undervoting and double voting in
Palm Beach county unusually high compared to other counties and/or compared
to previous years?
Validity: will a machine recount or a human recount provide a "better"
number of the actual vote totals? Better in what sense?
Bias: is there evidence of bias on the part of the vote recounters?
Estimation: With a partial manual recount in Dade County, can you estimate
what the results of a full recount would be? Based on the recounts in
several counties, can you estimate what the results of a statewide manual
recount would be?
Estimation again: What is the rate of undervoting and double voting using
punched card ballots? Is it higher than the rates for other ballot
mechanisms?
Survey design: Was the butterfly ballot confusing? Why or why not?
Data quality: What caused the two premature calls for Florida in the media
on election night? Are exit polls a useful source of information?
Data quality again: What is the impact of having different standards for
counting votes in different counties?
Protocol deviations: were the rules of the election changed after the
election in a manner to unfairly favor one candidate?
Sample versus census: Is the concept of sampling error meaningful in the
context of an election? What types of non-sampling errors are present in the
vote count?
Clearly some issues, like whether the U.S. should keep the electoral college
are not statistical in nature (though the discussions were still
interesting).
If there were other important statistical issues raised by this election,
let me know so I can add to this list.
Steve Simon, [EMAIL PROTECTED], Standard Disclaimer.
STATS: STeve's Attempt to Teach Statistics. http://www.cmh.edu/stats
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
