Online Sampling Again
Representivity ain't what it used to be

Mixed mode, but with a twist

I was in London last week at the MRS Annual Conference where there were a number of interesting sessions, including an excellent workshop on social media measurement by Ray Poynter from NewMR that I believe Ray is turning into a webinar. One of my favorites among the other presentations was one by Neil Griffiths from Blinc. I liked two things about it.

First, the work it reported on was inspired by one of my favorite books—Consilience by E.O. Wilson. The book is primarily about a unified theory of science that explains everything from physics to psychology and even religion, but for this presentation the focus was on the scientific meaning of the term “consilience” itself: “the principle that evidence from independent, unrelated sources can converge to strong conclusions.”

Second, Neil described something that I hope we are going to see more of in the not-to-distant future: use of four different methods to study the same research question. The application at hand was ad testing and the four methods used were dial testing, facial coding, implicit measurement, and predictive markets.

Those might not be the four that I would choose and I am sure that we all can point to shortcomings in any one of them, but it is hard to argue against a result that to one degree or another is confirmed by all four.

A month or so ago at IIeX Europe Eric Salama from Kantar gave a talk in which he argued that MR companies must move from “or” to “and.” I’m not sure that he was making the same point as Neil was making, but it struck the same chord. It was great, especially for a survey geek, to live in a time when the first step in solving any client problem was to do a survey or a bunch of focus groups. But those days are rapidly disappearing in the rear view mirror. At the same time, there is no one thing on the horizon to replace them, despite what you may have heard.