Monday, September 21, 2009

Not to besmirch the Vermersch...

Beyond the interesting findings in Light and Wakeman's article (2001) and their implications for designing web interfaces, what intrigued me was the Vermersch interviewing technique. A Google search revealed an earlier paper that had been produced (Light, 1999) that discusses in greater depth their application of this technique for the same study on website user perceptions and text entry fields. Heretofore having only been used in France and applied to non-HCI activities from exploring the performance of professional sports players to helping to remove 'mental blocks' in remedial students, this 'explicitation interviewing' technique was determined to be the best approach for gathering data at the depth and granularity needed for this study.

To someone with a memory that is increasingly disappointing in both the long- and short-term arenas, it seems questionable that the interviewer would be able to gather any meaningful data by asking a user to detail the last time they visited a site where they were asked to enter personal information. Granted, this was back in 2001 when even a 9 to 11 split of female to male participants was considered too heavily weighted towards women and non-representative of general web users, so web usage was perhaps both limited to a more narrow segment of the population as well as not so frequent amongst those who were users. Certainly, I remember the splash page was only starting to wane around that time, with many sites' web presence being seen as an extension of their paper-based brochures. No input from users necessary. There were definitely a lot fewer forms--I think. Still, it seems akin to asking me to recall the last time I did the laundry, in painful detail. It wouldn't help that I am supposed to focus on one single instance of an activity I probably performed with little self-awareness. I wouldn't trust myself not to accidentally pull in recollections from previous times so that my account soon becomes a conglomeration of multiple experiences to better flesh out the narrative for my listener. This would surely distort the data.

So I wondered why the experiment wasn't more controlled, why the site visits didn't occur just before the interview, and videotaped to the benefit of both interviewer and interviewee. In their pilot studies they had adopted this approach, having the participant visit a designated site, then verbalize afterwards all their thoughts as they were reading and interacting with the pages. As it turned out, this led to a deluge of irrelevant thoughts about how the user felt about participating in the study, and how it contrasted against the manner in which they would normally choose to use a website. Since motivation underlies one's perception of an activity, and motivation was being toyed with for the sake of the experiment, this would have fundamentally undermined the aim of the study in determining nuances in user perceptions. Furthermore, if the interviewee had been shown a videotape of themselves as they were visiting a given site and told to verbalize their thoughts as they were doing this or that, instead of achieving the evocative gaze into space where they are able to re-live the activity in their minds and thereby focus more deeply on their thoughts rather than peripheral concerns, they would be examining themselves, using the video to post-process their thoughts and rationalize their actions. Not acceptable.

Besides relying on interviewer skill to avoid leading questions with more care than warranted by the usual user testing (rather than simply ‘what did you see?’ the properly posed question would be 'what did you see, or hear, or think, or whatever?’), the Vermersch technique also adopts some concepts from neuro-linguistic programming. Given three modalities--visual, audible, and kinesthetic--the interviewer would use the dominant modality as a basis for breaking down into sub-modalities and further exploration to help 'pin down' memories. Also, by making the use of the social tendency to echo body language, if the interviewee is making too much eye contact with the interviewer, the interviewer may staring off at a wall in a subtle effort to induce 'the gaze' in their interviewee. 

Also interesting and not discussed in the 2001 paper was the idea of the interviewer-subject 'contract.' The participant's agreement to this contract is essential to the success of the interview, and the significance of this seems not to be underestimated. Although the contents of this contract are not revealed in any detail, general remarks about it indicated that it would include a willingness to full cooperation in openly answering questions as they were asked. It was mentioned that there were times when it became necessary to remind the participant of their agreement to this contract in order to guide an interview along. I imagine this contract would include specific instructions against fabricating details. Again, though, there was no explicit mention of issues of deliberate dishonesty or inaccuracy on the part of the interviewee other than that a few of them admitted to having some mischievous thoughts and deliberately giving false information on the forms they had filled out during their website visit. This was apparently a good thing--evidence of both the interviewee's web savviness--they had wanted an intermediate user-base--as well as testimony to the effectiveness of the technique in revealing deep (and dark) thoughts. Still, the interviewer in the study seemed to take everything they were told at face value.

Other than the questionable integrity of interviewee accounts, the explicitation interviewing technique seems to be a good one that can be called upon given the aims of a study, and more likely to be used in conjunction with other approaches. Thinking back to a project from last semester in which we followed test subjects around a grocery store then conducted post-observation interviews, this technique could have been employed to deal with an isolated few minutes of the observation period to reveal thoughts that could not have been anticipated. Instead of relying on my own generalized--and somewhat leading--questions, I may have gleaned some real insights. Analysis of these inner thoughts may have led to designing more helpful features in our application--or better yet, isolated the truly essential ones. 

~ Jenny Wang

No comments: