Keynote talks

Thursday night’s American Association for Public Opinion Research keynote speakers were both interesting. Nielsen research head honcho Paul Donato saw survey researchers going back to the future and then heading in new direction. Donato described how in-person surveys gave way to phone surveys, which costs and then missed coverage with all-cell phone households gave way to Internet surveys, which now also seem imperiled because of rather extreme duplication (same people on multiple “Internet panels”) – especially with apparent “professional panelists” – and missed coverage – with the “digital divide.” Donato said we may be going back to in-person surveys (he may have mentioned regular mail surveys in here somewhere).

But Donato also highlighted new digital research strategies (which he called "listening,
as opposed to "asking," which is what he said surveys do). Nielsen (the company that historically monitors household TV viewing) has apparently started tracking and “coding” (categorizing and counting) text in Internet chat rooms and on blogs and is assessing volume and content. He also linked viewing of CNBC business reports with consumer spending two weeks later. If a lot of people are watching CNBC, they’re worried about the economy – and that means they’ll spend less two weeks later. Another Another fascinating example Donato gave was what he called “electronic ethnography” – sending subjects cell phones and texting them every hour during the day to ask what they’re doing and asking them to take pictures of what’s in the refrigerators and cupboards every day. (Yet another example was linking social networking Web sites with Geographic Information Systems – so friends could monitor their friends’ whereabouts – like on Yahoo mapping – and researchers were monitoring both: both what friends were saying and doing and where they were at.)

Ken Prewitt, director of the Census Bureau during the Clinton Administration, not so implicitly criticized Donato for focusing on commercial issues and for not talking about areas in which universal coverage is important. For democracy and fair policy-making to work, the United States needs something like the census that covers everyone. Prewitt battled the Republican Congress back in the late 1990s about whether Census 2000 would employ sampling – and lost. (About this, Prewitt quipped: When you look out the window to see if it’s raining – do you insist on looking out all of the windows before deciding to go out with an umbrella?) Right off the top, Prewitt suggested, because of concerns about the census and immigration control, without sampling, the census is likely to miss up to half of Latinos. And when so much of government resources (from Congressional seats to Community Development Block Grant funds) is allocated on the basis of census information, this is fundamentally unfair. Ultimately – since sampling seems unlikely to win out now, too – the government may have to abandon the household as the unit of census enumeration (partly because cell phones and e-mail addresses aren’t intrinsically linked with geographically based households; and partly because household surveys – in-person, phone, or regular mail – are so expensive) and may need to turn more to administrative records (gathered for some other purpose, like Social Security records – which at this point aren’t very accurate except for certain information that is key for the program they’re gathered for) and perhaps for the kind of digital information that Donato called for.

-- Perry