Link Found to Poll
First, I want to give a hat tip to Tracy Warner at the Journal Gazette for providing the link to the poll.
After seeing the poll it probably is valid. Here is how the Indy Star said the poll was conducted.
"The poll is based on telephone interviews with 501 Indiana residents age 18 and older. Interviewers contacted households using randomly selected telephone numbers. The sample was drawn in such a way that every household equipped with a land-line telephone had an equal chance of being contacted. The poll was adjusted by age and race to reflect Indiana's population age 18 and older."
The only concern I have about this poll is the lack of sample distribution information. It sounds like a ramdomized dialing system was used so they may not have the collected that data.
If the poll took a disproportional sample from any given area it would skew they numbers. Many campaigns use polls in this way as a tool to get people who vote via peer pressure. This is why I say don't trust any poll unless you can see how it works.
After seeing the poll it probably is valid. Here is how the Indy Star said the poll was conducted.
"The poll is based on telephone interviews with 501 Indiana residents age 18 and older. Interviewers contacted households using randomly selected telephone numbers. The sample was drawn in such a way that every household equipped with a land-line telephone had an equal chance of being contacted. The poll was adjusted by age and race to reflect Indiana's population age 18 and older."
The only concern I have about this poll is the lack of sample distribution information. It sounds like a ramdomized dialing system was used so they may not have the collected that data.
If the poll took a disproportional sample from any given area it would skew they numbers. Many campaigns use polls in this way as a tool to get people who vote via peer pressure. This is why I say don't trust any poll unless you can see how it works.
posted by Indiana Pundit at 8:30 AM
4 Comments:
Statistically your concern is not valid. Every landline had an equal probability of being dialed, but you still seem concerned that it might've taken a disproportionate amount of data from one specific area.
This is unlikely, and more to the point, it would have to be EXTREMELY disproportionate to produce the lopsided polling numbers presented.
Jeff, read this post again. I didn't say it was disproportinate. I said the poll was probably valid.
Without information one can only guess. The only way I can personally confirm what is reported is to examine everything.
I have seen polls produced from skewed data collection processes. Therefore I am skeptical until I am able to dig into the process of a poll.
But they clearly state that the calling process was random - equal probability for every number.
Are you suggesting they are lying? Or possibly your suggestion is that they are incompetent and inadvertently caused the polling to be non-random? These seem like the only two conclusions that can be drawn from your statement.
I must reiterate that when you weigh how easily the governor was elected vs these poll results, I just don't see how the data collection process could be skewed enough to account for such a large discrepancy...
Again, I am not suggesting they are lying nor incompetent, nor anything else. I'm saying I can't verify their results.
Post a Comment
<< Home