Overview of Causal Research

Causal Research is the most sophisticated research market researchers conduct. Its goal is to establish causal relationships—cause and effect—between two or more variables[i]. With causal research, market researchers conduct experiments, or test markets, in a controlled setting. Researchers study how a dependent or response variable—brand sales or brand preference—is effected by changes in a variety of predictor or independent variables: retail price, advertising spending, advertising copy, or other promotional activities. Causality is often expressed as an if/then statement: If X happens, then Y will occur.

To support a causal relationship, the researcher must find more than just a correlation, or an association, among two or more variables. To support a causal inference—a conclusion that if one or more things occur another will follow, three critical things must happen:

Temporal Sequence: There must be an appropriate time order of the events. The "cause" must happen before the "effect." Sometimes there may be a strong correlation between two variables, but we cannot say with any certainty that one is the dependent variable and the other is the independent variable. There is, for example, a strong correlation between students with poor grades and students who use marijuana. But, it is presumptuous to conclude that smoking marijuana causes poor grades without eliminating the other possibility: Poor grades cause student to smoke marijuana.
Concomitant Variation: Concomitant variation means that when the cause changes, we can also observe a change in the effect. For example, if a brand's advertising expenditures have been cut in half and the brands sales fell, we may suspect that the reduced advertising support caused sales to fall.
Elimination of Spurious Correlations: "Spurious Correlation" is a term coined by the great statistician, Karl Pearson. A spurious correlation is a common misinterpretation of cause and effect. It occurs when the presumed cause of an effect is actually caused by an unconsidered variable. Here is a commonly used example of a spurious correlation. Whenever ice cream sales at New York City beaches go up, more people drown at these beaches. Conclusion: Ice cream causes people to drown. This, of course, is a false conclusion. There is a hidden or lurking variable that explains both increased ice cream sales and a higher number of people who drown: Hot, summer weather. Another example of a spurious correlation is based on Dutch statistics. There is a positive correlation between the number of storks nesting on rooftops and the size of Dutch families.[ii] There is, of course, no causal relationship between these phenomena. Contrary to the old wife's tale, storks do not delivery babies. Larger families tend to have bigger homes, which may attract nesting storks To establish causation, the researcher must be certain that there is not a spurious correlation. For a humorous look at spurious correlations, visit Tyler Vigen's website.

Causal Research relies on experiments—test markets—where the researcher can conduct real-world or simulated experiments to ascertain how consumer attitudes, brand market share, and brand sales among other variables respond to changes in marketing mix strategies.

[i] A variable is a characteristic, number, or quantity that changes over time. Marketers are concerned with a wide variety of variables including market share, retail sales, retail price, brand loyalty, consumer characteristics, competitive activity, promotional support, etc. Variables can be defined as independent or dependent. A change in an independent, or predictor variable, predicts a change in the dependent or response variable. In an experiment, there may also be control variables. A control variable is a variable that the researcher holds constant during the experiment.

[ii] http://andrewgelman.com/2012/04/27/how-to-mislead-with-how-to-lie-with-statistics/. Sapsford, Roger; Jupp, Victor, eds. (2006). Data Collection and Analysis. Sage. ISBN 0-7619-4362-5.

toc | return to top | previous page | next page