You may remember that Rasmussen poll from one month ago that momentarily had everyone in a tizzy when it showed Frank Lautenberg, in the immediate post-primary frenzy, leading Dick Zimmer by only 1 point, 45-44. Rasmussen’s back with another shot at it:
Rasmussen (7/7, likely voters) (6/9 in parentheses):
Frank Lautenberg (D-inc): 49 (45)
Dick Zimmer (R): 36 (44)
(MoE: ±4.5%)
This is very much in line with the June 26 Fairleigh Dickinson poll (Lautenberg 45, Zimmer 28) and the June 11 Quinnipiac poll (Lautenberg 47, Zimmer 38). So what happened on June 9? Well, outliers happen. Even if the best pollster in the world does everything absolutely right — gets a good sample, eliminates all bias from the instrument — 1 out of 20 times, it’s still going to be completely wrong, i.e. outside-the-margin-of-error wrong. That’s what the 95% confidence interval, a caveat buried in the fine print of every poll report, is all about. Happens to everyone, even Rasmussen.
Remember, however, that the confidence interval only deals with the likelihood of RANDOM error. There are all sorts of systematic errors in polling. Telephone polls often exclude cellphone users. Household surveys (telephone or in person) typically only take one person from each household (not strictly a random procedure). Introductory questions vary. So does the phrasing of the questions. Then there is the issue of minor candidates, are they mentioned or not? Do we only take registered voters? Months ahead of time that tends to under count young people and people who have just moved. Do we use a “likely voter” model? Everyone has a different one. Should we adjust for party affiliation? What is the “real” party affiliation of the public at this point.
The point is that we focus far too much on the confidence interval issue.
Texas, North Carolina, Kentucky Senate, the latest Missouri Gubernatorial race.
http://www.rasmussenreports.co…
Obama – 43
McCain 43
Fav/Unfav
Obama – 54/43
McCain – 59/38
Great news!
NJ is always deceptively close, it’s like Alaska for the Democrats, an overwhelmingly Democratic state the Rs think they can win but never do.
An odd poll, with Lautenburg surging and Obama slipping.
Have there been puma sightings in New Jersey? My guess is that the energy issue is a factor in this. Lautenburg will
return to the Senate, but the one and (only way) Obama loses New Jersey (and the election) is the energy issue. It really is the Democrats’ Achilles heel in this election.