Quasi-Experimental Research on Television and Behavior: Natural and Field Experiments

Quasi-Experimental Research on Television and Behavior: Natural and Field Experiments

Methodological discussions regarding research on the effects of television often contrast the advantages of laboratory experiments for making causal inferences with the advantages of more naturalistic or ecologically valid field studies. Laboratory experiments are designed to answer the question “Can television affect viewers?” whereas field studies are designed to answer the question “Does television affect viewers as they use it over time in day-to-day life?” Most field studies are correlational, however, and permit us only to determine whether television use is related to some other behavior, such as school achievement or aggression, not whether there is a causal relationship. This chapter focuses on two special and relatively unusual types of quasi-experimental field studies that have the potential to answer both the can and does questions. That is, if well designed, they enable us to [Page 110]make causal inferences about the effects of television as they occur naturally.

In natural experiments, researchers take advantage of a naturally occurring change in the availability of television reception to assess the impact of this change on viewers. For example, a preexisting group with access to television may be compared with another preexisting group that has less or no television reception. Or, a preexisting group may be studied before television reception first becomes available and then again after some period of use, usually a relatively longer term period of months or years. The group experiencing change in the availability of television may be contrasted with other similar groups whose exposure does not change over the same interval. In field experiments, researchers also often study preexisting groups, but in this case they assign different groups (e.g., preschool classes) to different television viewing “diets” (e.g., prosocial programs vs. cartoons vs. no television). The groups may be observed first during a baseline period, to establish their similarity, and then again after the exposure period. The many methodological problems, issues, and considerations confronting researchers interested in conducting natural and field quasi-experimental studies are the focus of this chapter.

In their now classic monograph, Campbell and Stanley (1966) explained the distinction between true and quasi experiments and described a series of quasi-experimental designs along with their strengths and weaknesses. That monograph formed the basis of both a chapter and a book by Cook and Campbell (1976, 1979) on methodological issues involving quasi-experiments conducted in field settings. Their work has since been revised and updated in a chapter by Cook, Campbell, and Peracchio (1990). That chapter provides the jumping-off point for this chapter on methodological issues involved in natural and field experiments on television and behavior.

According to Cook et al. (1990), experiments are “any experimenter-controlled or naturally occurring event with rapid onset (a “treatment”) whose possible consequences are to be empirically assessed” (p. 492). Two major categories of experiments exist. In randomized experiments, participants are assigned to treatments at random, whereas “quasi-experiments primarily depend on self-selection or administrative decisions to determine who is to be exposed to a treatment” (p. 492). As Cook et al. pointed out, “It is generally easier [Page 111]to implement quasi-experiments than randomized experiments in many of the field settings where causal conclusions are needed” (p. 492). They defined a field setting as any one “that respondents do not perceive as having been set up for the primary purpose of conducting research” (p. 492).

...