by Thomas Streeter
June 21, 2000
|
|
In another life -- long before I started pushing bits down a wire -- I packed up and went to graduate school to study audience research. I was interested in Nielsen’s accounting for channel surfing in both their diary and audimeter (and, later, PeopleMeter) services. The short answer was that ratings really didn't account for channel surfing. Their methodologies were structurally incapable of accounting for the rapid or simultaneous channel viewing made possible with a wireless remote, though not technically incapable on the metered side of things, but their operational definitions didn't allow for frequent channel changing. For example, "viewing" a program was defined as watching five minutes or more of a program in a given 15 minute period. When pressed, Nielsen personnel explained, the five minutes had to be contiguous. You couldn't watch one minute here and another minute there and eventually have enough minutes to add up to five. This was true of both how they recorded the meter information and how they instructed their diary respondents--if a diary respondent even bothered to ask about such a thing.
At the time, there was some compelling empirical evidence being collected that people "multitasked" when they were watching TV, i.e., they monitored more than one program at a time and processed the information. I became frustrated when there was a lack of interest on the industry side to what I was doing. The response was sort of, "Oh, that's interesting, but it's beside the point."
If accurately representing the size, composition, and behavior of audiences was beside the point, what was the point?
I started looking at how people in the industry understood ratings, going back to the early days of radio, essentially asking the question "When ratings are the answer, what are the questions?" Formally, I was examining audience measurement as a lay epistemic system, which can be more or less defined as studying ratings as an everyday means to acquire knowledge. What do people think they know after they look at a ratings book that they didn't know before? What questions would knowing an audience's size and composition answer for someone needing to make a decision? It turns out that asking what people know after seeing a ratings book was the wrong question. The right one was "What didn't they know?"
Media scholar Ien Ang observed that our common understanding of an audience is of a group of people who gather in a particular place at a particular time to experience some event together. If people begin clapping or booing, there is a social calculus taking place to guide us in either joining in or opting out. Most importantly, it's easy to tell who's a member of the audience and who's not.
Media audiences are largely disconnected from one another. I might throw a brick at my TV during a show while someone else may record the program for posterity. I'm not aware of the other members of the audience and they are not aware of me. I might sleep in front of my TV and someone else might watch intently. Who is "in the audience" for a program (and who is not) depends entirely on three major interests in the equation: the content provider, the advertiser, and the measurement organization.
On the web, sampling problems, caching, deep-linking (and the like), are really just details and not the largest issues with audience measurement. The real issues are like 900-lb marshmallow problems, since the deeper you go, the stickier they get. What does it mean to "watch" TV, "use" a web page, or "view" a stream? How do you take what is an independent series of individual behaviors and describe them only in collective terms that mean something or one thing? We don't want multiple audiences, we want THE audience.
The "problem" of ratings isn't really in the collection methods, but within the organizational use. Ratings are interesting, organizationally, as they seem to have low credibility (loosely defined as believability) but high epistemic authority (defined as their influence on the formation of knowledge). I've never met industry professionals claiming that ratings numbers were more than vaguely accurate. They did believe, however, that their jobs depended on these metrics.
Longtime researcher Leo Bogart once said that audience measurement isn't an instrument of knowledge as much as it is an instrument of power. Ratings numbers are powerful because they are convenient shorthand for knowledge--nice and empirical, and they look good in spreadsheets. They are useful as a social convention, being both the means by which decisions are made as well as how they are evaluated.
Why isn't there more than one ratings group in each industry segment? Because two conflicting information sources don't reduce ambiguity nearly as well as one source of information--even if the one is spectacularly wrong. It's OK to be wrong, by the way, as long as everyone is using the same wrong information.
In the end, I chucked the whole audience research thing and got into streaming media. When asked, I'm usually opposed to positioning streaming media as a new "mass" medium, though I don't usually explain why. Well, this is why: If streaming media is put in the same category of advertiser-supported mass media as newspapers, radio, and television, it will inevitably fall into the ratings trap--it structurally has to. The notion of audience has no meaning in those media beyond surrogate (or actual) circulation figures.
What's the alternative? Avoiding depressing op-ed pieces like this comes to mind. Or find a way to sell yourself on the relationship between your audience members and the content you provide. Show how you deliver your message to some group of living, breathing humans, not some mythical "audience".
And argue like crazy that anything else is beside the point.
