The other day we noted a SurveyUSA poll that put State Auditor Claire McCaskill nine points ahead of her opponent, Senator Jim Talent, 51% to 42%.
That same day the Talent campaign advertised an October 12th Rasmussen poll that put Talent ahead by one point, 45% to 44%.
One possible reason for the wide discrepancy has to do with weighting by party ID; Rasmussen does while SurveyUSA doesn't. A similar discrepancy was exhibited in Ohio.
Rasmussen weights its results based on party affiliation trends they track month-to-month. The detailed info on the latest Rasmussen poll is locked behind a subscription wall, but an October 12th report on the Foley affair's effect on party affiliation conducted by Rasmussen found that 37.5% of Americans say they are Democrats while 32.4% say they are Republicans. It is likely that Rasmussed used comparable numbers in weighting their data.
Some pollsters say the weighted method is a more accurate predictor because party affiliation is more stable over time, while others say the un-weighted method gives a better snapshot of public opinion and that the weighted method will suppress the impact of real, if short-term, events such as the Foley scandal.
Keeping that in mind, the latest number is the second time Talent has had the lead in the Rasmussen polls since August. The two candidates have been within three points of each other since March.
McCaskill was ahead by one point in August and September in the SurveyUSA polls. The jump in October may likely have been driven by the events of recent weeks and may be short-lived. It's worth noting that in the October SurveyUSA poll, 37% of respondents identified themselves as Democrats while 32% said Republican, close to the Rasmussen numbers; September was 40/32 and August was 35/35.
An excellent source for the aggregate poll data is available at Pollster.com. A summary of both the last five and ten polls has McCaskill ahead by 2 points.
Right a Wrong. Submit any tips or story ideas by using our anonymous email form. Confidentiality is guaranteed.