This morning's email had a link to an interesting study by one of the foremost experts in Web usability, Jakob Nielsen. The headline: "Talking-Head Video is Boring Online." You can read it yourself here: http://www.useit.com/alertbox/video.html .
The bottom line: his eye tracking study showed that even in the short video Nielsen tested (24 seconds) the test subject's eyes often wandered the page looking at and reading other stuff as the talking-head droned on, suggesting that she was quickly bored with the video. To extrapolate to my own reactions to video on the Web: the Internet is about lots of content speedily delivered and video is inherently slow and ponderous.
But what are the potential implications for surveys? Two come to mind.
First, there would seem to be a caution here for those who dream that some day Web surveys will be done with "virtual interviewers," that is, video of a human interviewer asking the questions. This certainly has been talked about with the added argument that you can control potential interviewer effects by matching interviewer characteristics (like gender, age, and race) to the respondent. None of this has been extensively tested and it surely needs to be before it's put to use, but this study suggests that this may be one more thing people do in Web surveys simply because they can, not necessarily because it adds anything to the respondent experience or improves data quality.
Second, there may be a more general caution about using video of any sort in Web surveys. Do respondents watch it? For how long? How closely? What kind of multi-processing might they be doing? If nothing else, this study would seem to suggest that keeping it short is a good idea.
This is a topic on which I expect we will hear much more over the next few years.