Gaining access to the web’s ‘walled gardens’

Comley suggested in 2007 that the US might have already passed ‘peak panel’. Whenever it occured, Eggers says the effects will be felt keenly in 2011, with capacity constraints prompting researchers to look elsewhere for respondents.

“We will increasingly find respondents out in those environments where they most enjoy spending their time,” says Eggers. “It could be Flickr or Facebook – but 2011 will be a year when many other interesting internet sites will get our attention.”

For researchers this new reality may take a little adjusting to, as the role of respondent gatekeeper shifts from panel owner to website owner. Many of the most popular sites on the web, the social networks, are ‘walled gardens’ – which even the web’s inventor Tim Berners-Lee recently criticised for limiting the free flow of information.

Vaulting the walls is not an option, so researchers are going to have to negotiate with site owners to gain access to these vast pools of potential respondents – and site owners will be cautious about what content is presented to their members for fear of damaging their cherished relationships. In the brief time that business networking site Linkedin made its member base available for surveys, there were frequent reports of how prescriptive the site was about what could and could not be done with its sample.

Eggers prefers to the see the upside of this scenario: that researchers will be forced to create enjoyable, engaging surveys that treat respondents “delicately and with respect”. “It’s forcing a long-term protection of the relationship rather than short-term ‘churn and burn’,” he says.

“The history of market research has typically been one where we have overused whichever channel is available – mail, telephone, email,” says Eggers. “The interesting thing about the internet is that it is going to be increasingly difficult for market researchers to overuse it.”

With multiple potential sample sources out there, researchers need not worry about running out of respondents any time soon, though as we’ve seen with research panels themselves, each website or social network will have their own built-in biases as regards the demographic and psychographic make-up of their members or visitors.

This makes it difficult to produce samples representative of any population, though sample blending can help – but first researchers need to understand those biases. Eggers says GMI is using respected population surveys – such as the General Social Survey in the US – to act as a benchmark against which to compare the profile of a particular sample source.

As far as blending goes, once you know the bias of the various possible sample sources, you can counterbalance it to produce a set of results that should be projectable to the population you are looking to study.

GMI has already profiled six of the largest research panels in the US in this way, with similar work ongoing in Canada and the UK, and Germany, Australia and France to follow.