Why it pays to get social in research communities

Online research communities have for some time been talked about as a way to reverse the pattern of declining response rates and consumer apathy in the research industry by engaging more personally with people and providing a way to unearth insights more effectively. But relatively little research has been conducted in industry or academia in to the best way to set up and run such communities or to establish the extent to which they are able to drive up participation levels.

Against this backdrop, the research and development team of InSites Consulting, in collaboration with researchers at the University of Maastricht, decided to put communities to the test. We wanted to examine two things. Firstly how community structure can influence participation levels within the community itself, and secondly how communities can affect participants’ response rates for other, ‘traditional’ surveys to which they are invited by the research agency.

We ran a parallel test by setting up two international online research communities, and monitoring them over 12 months from October 2008. The two communities discussed the same range of topics (including technology, services and FMCG products) but there was one key difference. Community A included a social forum and promoted interaction between community members, while Community B was strictly focused on research discussions, meaning that members could not post their own topic threads, only reply to topics posted by the community moderators.

Three sub-samples were drawn from InSites’ proprietary research panel, aligned on age, gender, nationality, marital status, education, length of membership and propensity to accept research invitations. One group took part in Community A, another in Community B, and the third was used as our control, not being involved in a community and so more representative of ‘passive’ panel members.

For one year we gathered data on members’ participation in communities A and B as well as the participation rates of all three groups in normal surveys (outside the community) to which they were still invited.

What we found
We discovered that one of the benefits of running online research communities is that participants share a huge amount of detailed information within such a setting, not only information directly related to research objectives but broader, more personal and often more emotional information. This can be attributed in part to the qualitative and ongoing nature of community discussions and to the presence of a wider range of participants – real people as well as researchers.

Looking at the average number of total posts in the two community settings, a high degree of social interaction appears to drive up participation levels considerably. While members of community B, which didn’t encourage interaction, posted an average of 136 messages a month, members of community A, in which social interaction was encouraged, posted a monthly average of 215 messages – 58% higher. Social interaction appears to be the driving force behind higher community participation.

We also wanted to ascertain the broader impact of hosting research communities by looking into members’ external survey completion rates. As can be seen in the chart, we again uncovered strong evidence of the beneficial influence of social interactions within communities.

We also noticed a time-related effect. Members of both communities show an increased level of survey participation right after the community launch in comparison with the ‘normal’ panel members in the control group. This suggests that launching a community generates a honeymoon period in which interest and commitment levels are heightened. This effect motivates respondents to engage not only in community discussions but in other traditional surveys to which they are invited by the agency. From the standpoint of a research agency, a community is beneficial in the short run irrespective of whether it involves social interaction or not.

The long-term view
However, after the honeymoon period, the completion rates of the different groups diverged according to community type. Members of community B participated significantly less in surveys and completion rates decreased steadily over time, from 73% at the starting point to 43% by the fourth quarter. Indeed, this response rate is significantly below that of the control group, which is not part of any community. Evidently, the lack of social interaction in Community B serves to demotivate participants in terms of external survey completion rates.

By contrast, a community that promotes high degrees of social interaction has a fundamentally different impact on members’ participation rates. Over time, an ‘enrichment effect’ can be observed, with the survey completion rate of members in this group rising to 95% in the fourth quarter, compared with falls to 43% in Community B and 60% in the control group.

The results support our hypothesis that this social aspect engenders an emotional attachment to the community – and by extension the research agency behind it – which encourages respondents to participate in more research than they otherwise would.

Social success
Our study concludes that social interaction among research respondents has greater benefits which spill over into the wider performance of participants in non-community research. It is not merely hosting research communities but promoting social interaction within them that drives success.

To facilitate and promote social interaction among respondents community moderators can encourage research discussions alongside discussions of current (non-research related) events, encourage a higher degree of personalisation by asking members to disclose more personal information, and extend the community beyond the online setting by facilitating occasional offline events. Although not always feasible, previous studies have shown that such shared customer experiences strengthen a community and boost participation levels.

While more remains to be learned about the dynamics and longer-term impacts of communities, this study illustrates the potential for agencies to improve engagement levels by putting in place communities that enable people to connect with one another, above and beyond the research objectives.

By treating people as people and not merely as respondents, allowing them to express their thoughts, to make connections and to engage in genuine conversation, communities promise agencies the ability to create a virtuous circle of engagement.

Stephan Ludwig is a researcher at the University of Maastricht. Paul Nola is a director in the UK office of InSites Consulting.

The data set for the study was made up of 1,234 respondents, of which 331 were members of Community A, 403 were members of Community B and 500 were in the control group.