Home | Business News | Browse by Publication | J | Journal of Consumer Affairs

Cautions and concerns in experimental research on the consumer interest.

Article Excerpt
Most published consumer research presents data from surveys or other data analyses that, at best, report that certain things tend to happen at the same time. However, correlation does not mean causation; cause and effect relationships can only be concluded from controlled experiments. A key problem is that the use of experimental designs calls for various conceptual and pragmatic trade-offs that cannot be ignored.

**********

Researchers in the consumer interest would like to say that various business activities cause specific consumer responses. Public policy makers would like to know that a planned law or regulation would cause a desired outcome. While critics of business practices would like to conclude all sorts of causal relationships well beyond what the data might allow (e.g., as described last issue in Carlson 2008), a stronger and more conclusive statement can only come from an experiment. Unfortunately, all research methods have limitations, and experiments are not exempt. In particular, experimental designs require certain trade-offs, and these trade-offs provide cautionary notes that researchers might like to ignore or otherwise bury.

A pragmatic approach to understanding a body of research and how it contributes to theory comes from a full understanding of what any design reveals or fails to reveal. Just as researchers should not overstate correlation findings as causation, users of experimental designs should not confuse their controls With how consumers interact in the reality of the marketplace. As current editor of the Journal of Advertising, former associate editor of the Journal of Consumer Affairs, and an Editorial Review Board member for several journals, I unfortunately have seen a number of different manuscript submissions that have done just that.

In an experimental design, one or more variables of interest are manipulated to create variation--generally referred to as a "treatment"--and the differences among the dependent variables are assessed to determine which treatment caused which effect. This type of research is ideal for evaluating relationships among variables that were originally identified in another study as critical. For example, in his commentary on content analysis, Carlson (2008) noted that one can conclude that a certain percentage of advertisements might...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!