Three alternative approaches are possible in creating a research design. They are not mutually exclusive, but in most cases, the design of a research plan is limited to the use of one of the three.
The first approach is the experimental approach. This approach requires that certain procedural rules must be followed. Essentially, the variable of interest—e.g. price, message—must be manipulated and everyone participating in the experiment must have a known and equal chance of being selected.
In a market experiment, information relating to the basic problem is obtained through the use of a small-scale simulated program designed to test a specific research hypothesis. Suppose, for example, that we wish to test the question that families of similar size and economic characteristics living in three different cities purchase different amounts of aparticular formula of a soft drink, such as Dr. Pepper. The first step would be to establish the research question: "For a given time period, the average fluid ounces of a Formula A, B, or C purchased in each city were the same". Next, a sample of the families in each city would be selected and randomly assigned either A, B, or C. Next, a survey would be taken to determine the number of ounces purchased by each family. Once this was done, a statistical test would be used to test the research question. If statistically significant differences in purchases of Formula A, B, or C of Dr. Pepper were noted, it could be concluded that taste does influence the amount of this soft drink purchased by families with the same social and economic characteristics. Of course, other hypotheses about soft drink purchasing could also have been tested using a slightly different method. For example, the effect of television advertising on the purchase of Dr. Pepper might have been studied by inspecting purchases in two or more cities that are in the same general area of the country (such as the southwest) but in which different levels of television advertising had been used.
The second approach is the historical. In this approach, reliance is placed on past experiences in seeking solutions to marketing problems. Historical marketing facts are relevant only to the degree that they can be projected into the future. Fortunately, in many areas of marketing, this can be done with a good deal of confidence. Certain types of changes, such as populations and income distribution, come about rather slowly. The day-to-day effect of these changes on marketing is almost imperceptible. Projections of future population, gross national product, and consumer purchasing power are practically foolproof. Historical analyses of such factors as consumer behavior, competitive selling tactics, and distributors' buying practices tend also to be fairly reliable indicators of future behavior by these same marketing components. Often, it is possible to trace the experience of organizations similar to yours and assess how they dealt with similar problems. There are literally hundreds of case studies on companies such as Microsoft that are useful to many business functions. Learning from the mistakes of others makes good business sense.
The third approach that can be used in designing a marketing research plan is the survey approach. In the survey approach, marketing information is collected either from observation or by questionnaire or interview. In contrast to the experimental and historical methods, in which the data are more or less directly related to the problem, the survey approach necessarily involves far more subjectivity and intuition on the part of the researcher. Watching a customer make a purchase of a new TV reveals something about his motives; simply asking him why he is buying it is much better. Drawing conclusions from either observations of behavior or from the opinions offered by a respondent create important insights. The survey method is flexible. It can be adapted to almost any type of research design. For this reason, and because of the difficulties in creating marketing experiments and in collecting pertinent historical data, the survey approach is the most often used in marketing research.