There’s a fascinating interview on Tom Anderson’s blog about Next Gen Research and biometric research technology. Biometric feedback capture for market research purposes has come a long way since I first came across it in the 90’s, and the things biometric researchers can learn about individuals border on uncanny. Ryan Brown notes in the interview, however, that biological indicators are often “useful for detecting individual differences,” but projecting these results across a sample population is complex and still not directly linked to actions (like purchase behavior.)
The Internet has transformed our business, as it has yours, and the surge in social media usage presents market researchers with an even more exciting opportunity to observe behavior (or, more accurately, another window into reported behaviors.) There are loads of startups in the analytics space tapping into social media monitoring, data mining and other sorts of datastream wrangling–but not so many trying to crack the real nut of Internet research, next generation sampling. One can mine Twitter for sentiment a dozen different ways, but in many respects sampling has not really been able to improve upon old–but incredibly effective–methods like door-to-door address-based sampling, and indeed that method of research still works best in many parts of the world. Interestingly, as the ubiquity of mobile devices enables multiple touch points for survey respondents, their existence has also created the mobile-phone only respondent, the bane of telephone/RDD samplers everywhere. In response, door-to-door in-person sampling has experienced a bit of a renaissance, though it seems on the surface to be a defiantly low-tech solution.
All of which leads me to ask a truly agenda-free question, since Edison employs any and all methods mentioned above in the service of providing our clients with actionable data. As research methods from biometrics to social media monitoring have enabled–and improved–the study of samples, have we gotten better at predicting the behaviors of populations? This may sound like a provocative question, but no provocation is intended–merely intellectual curiosity. There are loads of case studies, for instance, on the efficacy of social media monitoring, but most of them boil down to the deft application of tactics, not to predicting user behavior. As next-gen research methods continue to evolve, we the diagnosticians need to continue to diagnose our own methods–physician, heal thyself–and conduct meta-studies in the real-world utility and predictive power of the insights new research techniques enable. Otherwise, we risk learning more about the trees at the expense of the forest.