American media and the smoking-related behaviors of Asian adolescents.
In his own study, the author establishes no causal connection but merely the existence of a correlation between youthful smoking in a particular geographic area, Hong Kong, and the extent of exposure to cigarette advertising, the extent of exposure to American movies, and the extent of ownership of products bearing a cigarette brand name.
The author believes that correlational investigations such as his can supplement and give additional precision to studies showing causal connections. I disagree.
To illustrate the basis for my disagreement, consider the well-known relationship between lawn growth and the application of fertilizers containing nitrogen. Causal connection was established many years ago using experimental designs to winnow out causal connections from correlations whose direction of causality was unknown or which were produced by variables extraneous to the ones being observed as correlated. A Latin Square design, for example, controls for the possibility of the grass being greener toward the north, east, south, or west of the field, and not greener as a result of the fertilizer being applied. in addition, these designs tested the absence as well as the presence of fertilizer, often including levels or amounts of fertilizer within the experiment.
The difference between such experiments and the type of study the author conducted is that the decision of whether to apply fertilizer to a given plot and the decision as to how much to apply is made by the experimenter. What we learn from the author's study is that the children who smoke also own items with the names of cigarettes on them. It's as if we drove around suburban areas, grading lawns on their lushness, and questioned homeowners as to whether fertilizer had been applied and if so, how much.
If we then found a correlation, say, to the same modest extent that the author's study found between "number of American movies seen in past two months" and "ever puffed a cigarette," we would suspect that perhaps some important extraneous variable had been omitted.
In the case of grass, one such variable might be the amount of watering done. In the case of teenage smoking, such an extraneous variable might be the discretionary income level among this teen population, which could have facilitated both smoking and moviegoing.
In any event, it is as a matter of experimental principle that I make this comment, rather than to suggest any specific conjecture about teenagers and smoking. Observational studies do not add to the knowledge already gained through genuine experiments. In my view, the proper sequencing between these study types is that an observational study can suggest a possible cause and effect relationship, which is later tested using a suitable genuine experiment. After a causal connection has been established, subsequent observations do not add to the knowledge of causes and effects. The subsequent observations are just as subject to the limitations of correlation as were the observations made prior to the conduct of genuine experiments.
TABLE 1 Studies Cited by Goldberg (2003) * The Institute of Medicine, 1994 Distefan et al., 1999 USDHHS, 1994 Pechmann and Shih, 1999 USDHHS, 1996 Pierche et al., 1998 Pollay et al., 1996 Wakefield et al., 2001 Siegel et al., 1996 Pechmann and Knight, 2002 Aitken et al., 1991 Biener and Siegel, 2000 Botvin et al., 1993 Andrews and Franke, 1991 Chassin et al., 1984 Lewit et al., 1981 Collins et al., 1987 Saffer, 2000 Hawkins and Hoch, 1992 Laugesen and Meads, 1991 Hasher, Goldstein, and Toppino, 1977 Chaloupka and Laixuthai, 1996 Bandura and Walters, 1959 * Citations may be found in this issue: Goldberg, M. E. "American Media and the Smoking-related Behaviors of Asian Adolescents," Journal of Advertising Research 43, 1 (2003): 2-11.
ROBERT N. REITTER
Guideline Legal, Division of Find/SVP
|Printer friendly Cite/link Email Feedback|
|Author:||Reitter, Robert N.|
|Publication:||Journal of Advertising Research|
|Date:||Mar 1, 2003|
|Previous Article:||American media and the smoking-related behaviors of Asian adolescents.|
|Next Article:||Effects of differential enhancements on mail response rates.|