Chapter 5
Perspectives from Other Kinds of Epidemiological Research
The field of media research is relatively young and its methods and contributions are not widely understood or appreciated beyond its borders. Social epidemiologist Michael Oakes brought the perspective of an outsider to an analysis of the extant research on the effects of media consumption on children (Oakes, 2006). The discussion began with Mark Becker’s point that the prevailing research model for media studies has been based on research methods typical of psychology. He suggested that other statistical or epidemiological models may be more effective for some types of media research analyses—most especially for developing the kinds of causal links that are needed to support policy and public health measures that could influence behavior and improve outcomes for children.
The focus of Oakes’s presentation was to compare the research designs, standard measures, and modeling frameworks used in media research with those used in other social epidemiology and prevention studies. He began by stressing the high quality of much of the media research he has reviewed. He was favorably impressed by the volume of compelling studies and reiterated the point that the evidence for links between exposure to violence in the media and aggression is as strong as that for links between cigarette smoking and lung cancer. From his perspective, the measures of media exposure are effective and are likely to improve, although he recommended an intensified focus on the coming challenges of new technologies and changing ways of interacting with media, as well as marketing innovations.
Several gaps in the research base were evident as well. One notable lack was in what Oakes called the study of effect modification by social subgroup. In this category he included examination of media effects by social subgroup and by other social-environmental contexts besides race and class, as well as efforts to investigate the effects of social interactions on individual and group outcomes. While differences between boys and girls and across age categories are frequently explored, unexamined differences related to wealth, region of the country, and other factors related to socialization may have profound influences on how children are affected by media. Other contextual effects insufficiently addressed in media literature include school settings, neighborhoods, family structures, and social and cultural interactions; all these factors may have important influences on the way children process media exposure and may in turn be influenced by media themselves. Other types of social subgroups may be categorized by the preferences of children and youth for certain types of media content or technology (including design features as well as technical source). Significant differences may reside in how light or heavy media users respond to certain program themes or stimuli.
Traditional regression models make it difficult to isolate selected contextual effects, Oakes explained, since media exposure is frequently pervasive. This challenge is similar to studying the effects of smog or water quality on different populations; the level
of exposure cannot be determined with precision. Multilevel regression models are increasingly being used in epidemiology and public health studies to examine contextual influences in a variety of areas. He used John Snow’s discovery that removing the handle of a pump reduced the incidence of cholera in the neighborhood (by limiting access to tainted water) to emphasize the importance of exploring all potential factors. This finding led to efforts to disentangle more proximal causes for disease—which in turn led to the breakthrough discovery of germ theory and the effects of bacteria in sewage.
Oakes also found little innovation in media research methodologies in their approaches to inferring cause in observational studies. The axiom that correlation does not imply cause remains true, but the potential of surveys and correlational studies to illuminate cause, even in the absence of an explanatory theory or mechanism, is dismissed too readily, he argued. Other research fields are making significant strides in finding innovative ways to infer causation. Counterfactual causality, originally identified by philosopher David Hume as the “but for” condition—the case in which situation x would not be true but for intervention y—has been used in contemporary research to account for alternate outcomes that result from differing exposure to particular influences. This line of analysis compares scenarios in which two conditions with similar features produce very different outcomes as a result of the presence or absence of intervening factors.
This model helps to compensate for the basic challenge of causal inferences—that it is never possible to observe the exposure of one individual to two different sets of influences in the same period of time. A substitute for the unobservable condition—the individual exposed to influences other than those to which he or she was in fact exposed—must be found. Bias creeps in when the substitutes are not good controls because they are not sufficiently similar to the research subjects. Randomized groups of sufficient size provide one means of ensuring that the control group is similar to the targeted group, but in observational studies, in which no intervention is done, there is no opportunity for randomization.
To approximate the benefits of randomization when it is not possible, a variety of new statistical techniques can be used to analyze the data from observational studies to identify causal relationships. One such method, developed by Rosenbaum and Rubin (1983), is called propensity score analysis. This approach begins by developing a statistical model to predict exposure independent of the outcome in question. In a study of the impact of exposure to media on children’s diets, for example, logistic regression could be used to estimate the probability—called the propensity score—that a child is exposed to food advertising on TV as a function of parental monitoring and other factors. The probability of exposure, known as the propensity score, is then used as a control variable in an analysis of the relationship between diet and food advertising. Under certain assumptions, it has been shown that the relationship between the outcome (diet) and exposure (advertising) that can still be observed after controlling for other factors using the propensity score can be attributed to food advertising. Like any other statistical method, however, the validity of the inference depends on the assumptions, and propensity score analysis results can be biased if the assumptions are not correct (Rubin, 1997).
Oakes stressed that propensity score analysis is a relatively new approach that may be particularly well suited to the kinds of questions that media researchers tackle.
Not only epidemiologists, but also economists, political scientists, and other researchers who are especially concerned with public policy, have been struggling to find new ways to make inferences about people’s behaviors and the choices they make. Instrumental variable analysis—another method of accounting for the unmeasured confounding variable used in econometrics—might also have an application in media research. Another approach is group randomized trials, in which the effects of policies are assessed using random samples of families, schools, neighborhoods, and even towns.
Oakes noted that longitudinal research designs, although they are the prototype for epidemiological research, may not be as helpful in media studies as other approaches. Longitudinal designs, he explained, are very useful for identifying random, unintentional exposures or the natural occurrence and progression of a disease. But when individual choices affect outcomes, the background characteristics and preferences that influence those choices are too difficult to disentangle. Moreover, factors that change over time, such as the nature of media technology or content, and the cumulative effects of ongoing exposure, have changing effects on outcomes, which can affect the findings in a longitudinal study.
The general challenge underlying most epidemiological research, and media research in particular, is that research subjects are not Robinson Crusoe, affected only by the conditions on a small island. Individuals respond to certain stimuli on the basis of their prior experiences as well as current conditions, and they are affected as well by the presence or absence of others. Individuals’ choices and experiences are dependent on an infinite number of choices that others have made, as well as infinite other ways in which they are influenced by the individuals and groups that surround them. The more basic question is the perennial one about how social environment influences biological, psychological, and social processes, and vice versa.
Thus, Oakes argued, the most valuable approach may be to use field experiments to investigate the potential impact of interventions. While models that posit causality are needed to support the choices of interventions to evaluate, conclusively resolving the relative contributions of individual versus social factors (nature versus nurture arguments) may not be necessary to achieve public health objectives.