Elimination of Investigator Bias

Videotaping may seem extreme practice but I’ve found in my career that very few people are alert to the interpretation of body language, facial expressions and other visual clues of their colleagues and customers. Studies can be videotaped using digital rotoscoping, an animation technique that captures all the facial features, body movements, gestures, and other signals+ of emotions in a very lifelike representation, but respecting the ethical concerns that come with recording investigators, patients and animal testing. Videos can be used for automated digital analysis of customer responses and/or interpretation of responses across multiple investigators, allowing us to more accurately measure and manage both investigator and customer bias. They provide permanent records in the event we later uncover unexpected results: did we see any indication of that issue in our earlier studies? Recollection and/or written records are inadequate. Videotapes, both the audio and visual portions, can provide a powerful guard against study bias.

We won’t throw out the baby with the bathwater. Just because we discard statistical analysis doesn’t mean we ignore the many insights on bias developed over the centuries by the discipline of statistics, for example how to avoid selection bias, investigator bias, etc. We preserve intact those practices that mechanically reduce bias in experimental results. And we have special firewalls in place to fend off the Trial Lawyers. Although it is somewhat flawed by its uncritical acceptance of the value of comparator trials, the article by Lindner (subscription required) should be one of the first sources consulted for its extensive list of potential investigator biases of which we should be aware, and for which we should develop direct methods of control (i.e., in contrast to the indirect methods of hiding behind a veil of numbers).

Again, if analysis of video tapes is a powerful means of avoiding investigator bias, where would you look to recruit individuals in order to improve your odds of catching subtle visual clues or signals?