![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Gallup-ing to the right?
Oct. 26th, 2004 11:28 amMoveOn.org recently placed this advertisement on Gallup's polling techniques in the New York Times.
From my eight years of experience in marketing and public opinion research companies, I know that sampling techniques alone (i.e. how the list of survey respondents is compiled, or "who is asked to do the survey") can account for wide swings in results for any type of polling question. The MoveOn advert raises some good questions about whether or not Gallup's techniques are on the mark, but provides no specific detail as to what is the exact nature of the problems. The gold standard would be to run frequencies on the original call centre's survey data file. From that, along with an examination of the survey sample itself and how it was selected, we could determine if there were any methodological errors that might cause us to make erroneous conclusions.
The advert's point about the relationship between the polling company and the poll sponsor was interesting. Certainly, extreme care must be taken to ensure that methodological corners are not cut for promotional purposes or otherwise. But if the polling company is a legitimate one whose operations are managed by competent statisticians and research analysts, then chances are that the research being carried out is methodologically sound.
On a side note, it would be nice if polls were carried out by completely objective third parties (ideally by universities or other public centres of learning that include statistics experts on staff) and if their results were always peer-reviewed prior to publication. This method works extremely well in the medical and scientific fields, and it increases the credibility of the research by untold proportions. But of course, who would pay for that?
From my eight years of experience in marketing and public opinion research companies, I know that sampling techniques alone (i.e. how the list of survey respondents is compiled, or "who is asked to do the survey") can account for wide swings in results for any type of polling question. The MoveOn advert raises some good questions about whether or not Gallup's techniques are on the mark, but provides no specific detail as to what is the exact nature of the problems. The gold standard would be to run frequencies on the original call centre's survey data file. From that, along with an examination of the survey sample itself and how it was selected, we could determine if there were any methodological errors that might cause us to make erroneous conclusions.
The advert's point about the relationship between the polling company and the poll sponsor was interesting. Certainly, extreme care must be taken to ensure that methodological corners are not cut for promotional purposes or otherwise. But if the polling company is a legitimate one whose operations are managed by competent statisticians and research analysts, then chances are that the research being carried out is methodologically sound.
On a side note, it would be nice if polls were carried out by completely objective third parties (ideally by universities or other public centres of learning that include statistics experts on staff) and if their results were always peer-reviewed prior to publication. This method works extremely well in the medical and scientific fields, and it increases the credibility of the research by untold proportions. But of course, who would pay for that?
![[livejournal.com profile]](https://www.dreamwidth.org/img/external/lj-userinfo.gif){kind=small}
_imran_