Posted on 09/25/2006 10:44:47 AM PDT by Redmen4ever
Calling between the hours of 3 and 9 p.m. Eastern time, on every day but Friday and Saturday, Steiger's telephone interviewers made about 2,900 calls and spoke to 2,100 people before finding 603 willing to spend five minutes answering questions.
Of that 603, 63.5 percent are women and 36.5 percent are men.
(Excerpt) Read more at courierpress.com ...
1st. The professor's team matched phone numbers to registered voters (since Congressional Districts do not align well with phone numbers, which obviates Random Digit Dialing). Since only about 60 percent of people are in the phone book nowadays, you've got to wonder if this induces a bias (and which way).
(I am surprised the professor said he wasn't concerned about identifying Likely voters from Registered Voters, since you can simply look to see if a Registered Voter has voted in the recent past, in the process of matching Registered Voters to phone numbers.)
2. Of people who are registered to vote and are listed in the phone book, only about 40 percent will agree to a 5 minute interview. In a year in which Democrats are energized and Republicans dispondent, you've got to wonder if this induces yet more bias.
3. Of those interviewed, almost two-thirds are women. Since there is a pronounced tendency of women to vote Democrat and men to vote Republican, you've got to be suspicious that the sample is indeed biased.
(Many poll-takers "weight" their samples, so as to reflect known demographic features of the population, which can help with one form of bias, but might not help with the bias correlated with willingness to cooperate with the survey. Again, I don't know why the professor didn't weight his sample according to demographics, since his sample was so obviously skewed.)
4. Lead-in questions can very much effect responses, and I don't know what the 20 questions were prior to popping the question regarding preference in the Congressional race. If the professor is a liberal or left-winger radical, questions that appear to be unbiased to him might actually skew the results. However, I have no knowledge concerning this.
As to how much bias is in the sample, I am thinking maybe 6 points (based on the male-female ratio in the sample), so that Hofstettler is behind but not by the 15 points indicated in the poll.
Since he won, two years ago, 53 to 45 to 2 for the Libertarian, I am thinking he should be able to pull this one out, but this will require a strong GOTV effort.
Last weekend, I received a call from a pollster. I was surprised. I never was called about a poll before in my life. However, the cable guy arrived to upgrade my tv service. So, I told the pollster to try again some other time.
Should I have taken in part in the poll? Perhaps, I should have. But, I wasn't in the mood.
In insight into the polling process.
Polling data can be valuable, but the reputation of the pollster is the only way to know if it's reliable or not.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.