[lbo-talk] more polling stuff

Miles Jackson cqmv at pdx.edu
Tue Oct 19 12:42:46 PDT 2004


On Tue, 19 Oct 2004, Doug Henwood wrote:


> Miles Jackson wrote:
>
> >Let's be fair: we need to compare polls closer to election day to
> >the actual outcome to assess the accuracy of the polls. Cue Doug?
>
>
> Most had Bush ahead, by an average of 2 points. Ralph's final numbers
> were almost twice as high as they actual vote.
>

Given the typical sampling error of 3 or 4 points, all of the poll results Doug provided are reasonably accurate estimates of the true population parameters here (% of votes for each candidate). Chalk one up for probability sampling, and color me impressed.

Miles



More information about the lbo-talk mailing list