Measuring media us is vital to advertisers that base the value of purchased time or space in terms of its reach. And thus also vital to media outlets who depend on advertising for some or all of its revenues. And the need for good, reliable and valid measures is critical in a world of expanding competition, both within and among traditional and emerging newer media outlets, and increasing fractionating and targeting of media audiences.
Historically, the need for some audience use measures prompted the development of multiple systems of media use measures based on sampling (a full census has been too expensive). Sampling theory tells us that there will always be some imprecision in media use measures based on samples - and even higher levels of imprecision if the sample isn't perfectly random. Deviation from perfectly random samples brings to the measure the added potential of sampling bias, which can skew numbers. Reliability is more of an issue of consistency, and can be improved by standardizing both the sampling and measurement procedures over time, or using more accurate means of collecting data (as in the use of a meter instead of relying on recall). But the most problematic has always been the issue of validity - whether the measure is actually measuring what it is supposed to.
The issue of validity has always been at the heart of audience measures, and is tied to the question of just what is the purpose of the measure. Broadcast ratings are a good illustration of the issue. Ratings are supposed to measure the size of the audience consuming broadcast programming, the number watching or listening to a particular signal at a particular time. But their stated purpose is to indicate the size (and make-up) of the audience exposed to advertising sports inserted into the program. But the measures are based on whether an audience member was watching or listening to a certain proportion of a broadcast program during a specified period of time, and not whether they were exposed to a specific advertising slot. When an audience member's only access to a program was through a specific channel at a specific time, the discrepancy of the two goals (audience for program & audience for ad) is likely to be fairly small. Moreover, it was viewed as being largely consistent - thus the media and advertising industries came to an agreement on measurement process. They both understood the issues (imprecision, reliability, and validity), and saw reliability (consistency) as the most important issue, agreed to accept the measures as reasonably valid, and accepted (while often ignoring or forgetting) the inherent imprecision.
As media markets fragmented, new channels emerged, and competition arose, media faced the problem of their audience usage shrinking, at least according to the older measures. Television broadcasters, for example, were impacted by cable's importation of distant signals, then by the expansion of cable channels, the rise of videocassettes, videogames, DVRs and VOD options, and ever-expanding Internet-based video options. All of these increasingly competed for the time and attention of traditional TV audiences, shrinking and fragmenting them. As traditional ratings for programs continued to drop (threatening the ability to capture advertising revenues), the broadcast industry tried to recapture some viewing by offering the program at different times, on different channels, and through trying to capture use through alternative media (including viewing via DVR, VOD, or IPTV), and then sought to redefine audience use measures to include such use.
If the goal is to measure exposure to the programming content, redefining measures in this way can be considered as a more valid measure based on current audience media use behaviors. (It may also negatively impact on sampling precision and bias, especially if different samples and methods are employed for the various media, and reliability, at least in the comparability to older measures). However, unless the embedded advertising is also delivered with the program over each option, the new measures will be increasingly less valid as a measure of exposure to specific advertising messages, and thus less useful to advertisers. With the growth of alternative delivery systems for media content, the gap between the measures of content exposure and the goal of estimating audience size for specific advertising slots is likely to increase.
In the US television industry, the increased use of audience size guarantees in advertising sales is indicative of the concern that advertisers already have as to the value of existing audience measures as predictive. As the gap between the competing goals increases, look for the advertising industry to increasingly discount the value of audience estimates, and for it to seek alternative ways of measuring exposure to, and impact of, advertising messages and their value to advertisers. (see this post for TV, this post for newspapers)
As John Osborn recently blogged,
"ratings were always just a surrogate for the number of potential impressions an ad could garner within the program, helping set pricing for advertisers in a world of imperfect measurement... Without a better understanding of DVR usage, advertisers will pay more and more to run spots - especially on successful, high rated programs with the highest commercial loads - all while significant numbers of consumers are increasingly avoiding ads. Can you imagine advertisers opting in to such a business model if it hadn't already been running this way for years? "With a predicted $10 billion already committed to "upfront" TV deals - based on weak predictions of audience size drawn from increasingly problematic measures - the stakes for advertisers are reaching a critical point. Media needs to develop better, and more valid, audience use metrics to justify the value of its advertising slots, and the financial contributions they expect advertisers to contribute.
Prompted by: "TV Audience Ratings: Less and Less Relevant for Advertisers," TVBoard