Media diaries do not generally allow for the collection of data regarding context, such as other activities done at the same time.
Experience sampling methods provide a look at typical activities by sampling at random times of the day, often by paging participants and asking them to fill in particular information in a booklet of self-report forms. This method can capture internal aspects of the experience (motivation, mood, cognition), as well as external information (location, context, etc.). It is somewhat less burdensome than other methods, but it is useful only with participants who can write coherently about their experiences (i.e., not most children under 10—a key target for media research). It also does not allow researchers to determine the total amount of time spent on activities of interest.
Direct observation (or video observation) is an ethnographic technique for investigating social phenomena in natural settings. It provides very accurate and rich information about such issues as television viewing habits, but it is extremely time- and labor-intensive and consequently expensive—which limits the sampling that is possible. Moreover, although observation eliminates bias in answering questions, the presence of the observer or camera may have significant influence on participants’ media choices and behaviors.
Electronic monitoring is a commercial method designed to collect the data that are used to generate ratings for television programming. Two somewhat different technologies for this purpose are owned by two companies, Nielson and Arbitron, which have occasionally made them available to researchers for a fee. In the Nielson method, each viewer in the household is assigned a button on a device supplied by the company and is asked to push it when they begin and end a viewing session. Thus all programming played on that channel before the button is pushed off is counted as “viewed” by that family member, although he or she may have wandered off or begun a different activity. The Arbitron company uses a device called the Portable People Meter, which can detect codes that are embedded in video and audio programming and thus record the viewing and listening patterns of the people carrying them. Both these methods produce the same kind of data that a media diary yields; that is, they do not provide information about the context in which the media use occurs or its relation to other activities in which the participant engages.
Having called attention to the limitations of each of these methods of measuring the quantity of media consumption, Vandewater raised the question of how much precision is really needed to answer important questions about media exposure. The reliability and validity of these measures vary. Diary estimates, for example, are highly correlated with observational measures (.80 to .85), whereas global time estimates show only moderate correlation with observational data (.20) or diary data (.40). However, despite the variation, the big picture result from all of these sources is that the average level of television viewing for adolescents is 2.5 to 3 hours per day. This figure may be sufficiently precise for many purposes, while a much greater degree of detail may be needed for predictive research, or for exploring error variance. What is of greater concern is the introduction of imprecise measures as an independent variable in another study, such as the effect of parenting restrictions on media habits. In such cases, low reliability of the media measures can bias the estimate of the main effect.
Dimitri Christakis commented that no more studies demonstrating how much TV children watch are needed. Instead, he argued, the focus should be on studies that