At 03:14 PM 12/5/2002, Paul Bernhardt wrote:
Anyway, we then compared the actual outcomes to these crudely adjusted polls. We found the outcome was within the margin of error in about 16 of the about 20 races. (I'm not at my office writing this, so I don't have the data in front of me. If y'all are interested, I can get the details and post them.)
i think we have to consider the timing of the polls ... i don't know the time frames of the data you examined but, i would imagine that if we use poll figures a day or two before ... very close to election day ... that they would be fairly close ... since, not too many people REALLY wait until they get into the booth to make up their minds
but, if these poll figures were say a month before ... a real sort of "prediction" model ... that accuracy would be much worse because of the undecided voters ... both undecided about even voting ... or WHOM to vote for ...
a day or two before the election ... people have made up their minds A) whether to go vote or not, and B) who they will vote for
on a totally different matter, what are people's views on even allowing polls to be publicly displayed ... within X amount of time of the election day? i know it would be nearly impossible to do much about this but ... the goal should be to encourage as many people to go vote as possible and ... i can't think (generally speaking) that poll figures close to the election help that situation (except perhaps in some very very tight races)
even though it might be hard to document in any clear way the impact polls have ON how people vote ... we know that polls do affect how some % of people vote ...
perhaps it is time for us to think about whether we should allow that to continue ...
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
