Here is another interesting piece of news announced during the AAPOR conference: Automated pollster SurveyUSA plans to introduce a new service later this year called "SurveyDNA." The subscription-only service will allow subscribers to "take apart" SurveyUSA polls and re-weight and re-tabulate the results as they see fit.
For now, this press release on the SurveyUSA web site has the only available details on the new service, which promises the following features.
SurveyDNA subscribers will be able to:
- Examine SurveyUSA’s unweighted data and see how SurveyUSA weighting changed the data.
- Re-weight SurveyUSA poll results in real-time, using state-of-the-art interactive rheostats.
- Create their own geographic sub-regions for analysis.
- Create their own subscriber-defined time-series for scrutiny.
- Re-define who is and who is not included as a likely voter in a SurveyUSA pre-election poll.
- Re-graph, using custom colors, fonts and shapes, SurveyDNA revelations.
I emailed SurveyUSA’s Jay Leve for more details, and he says that while they have yet to finalize pricing and are not yet ready to release a demo to the public, they are actively seeking beta testers at this address: betatester@surveyusa.com
Many survey research organizations routinely deposit their respondent level data in academic archives (such as the Roper Center or the Interuniversity Consortium for Political and Social Research –ICPSR) or make it available directly (such as the Pew Research Center) months after their initial release so that scholars can slice, dice and re-weight the results as they see fit. What will make this announced SurveyUSA service unique, assuming it lives up to its promise, is that users will get an immediate (e.g. "real-time") ability to manipulate and re-tabulate respondent level data using web-based software rather than having to use their own statistical package.
This interesting, however, in your opinion how reputable is SurveyUSA. It seems as though the automated system would tend to bias the pool of respondents. Although I see what they report regarding their prowess, I’m still a bit skeptical. Do you know of any independent tests of their methods?
Dave,
You might want to start with my comments on their methodology here:
http://www.mysterypollster.com/main/2005/11/polling_the_pro_1.html
…and in my POQ article last fall – various links here:
http://www.mysterypollster.com/main/2005/12/mp_in_poq.html
See also Jay Leve’s interview with the Hotline:
http://www.mysterypollster.com/main/2005/02/the_hotlines_su.html
Neat. I sent them a message trying to get into the beta test.
Mr. Blumenthal, are you going to beta test?