More repurchase situations can be analyzed, which increases the accuracy of the test results. The above process assumes that the consumer’s behavior throughout the test is realistic because she was forced to pay money for both the initial and repeat purchases. Even so, laboratory test market models are not without their problems. No matter the safeguards used in attempting to control the environment properly, realism can never be totally assured.
It is extremely difficult to simulate the social dynamics involved in the realistic adoption process. In common with most market research, but more than with most methodologies, laboratory tests involve testing effects. That is, the behavior of respondents may be influenced by taking part in an experiment. This may distort the results. There is also the danger of a maturation effect which is especially relevant in tests of repeat purchase simulations. This effect occurs when respondents gain more knowledge about the experimentation process as the test progresses, which may lead to inconsistent results or to a loss of interest over time.
Urban and Katz analyzed the predictive accuracy of ASSESSOR, a pretest market model that has been used by more than 50 organizations to evaluate more than 200 new products. In general, it uses a research methodology similar to that described above. A comparison of pretest market and test market shares ranging over a number of different products showed a mean deviation of about one half of a share point – the initial forecast average share was 7.9 percent versus an actual share of 7.2 percent. This means that the predictions had a slight upward bias.
From experience gained in using ASSESSOR, it was determined that at least three conditions are necessary for its success. First, the product category involved must be well defined in terms of close substitutes; a new product category poses problems. Second, the usage /purchase rate must be the same for the new brand as for the established ones. And last, consumption and learning must occur quickly enough so preferences will stabilize in a short time period.
Industrial goods can be tested in a number of ways, including trade shows, in-use situations, and sales presentations. The first method consists of displaying and demonstrating the product to obtain measures of interest and possible buying intentions. In-use tests place the product with a sample of potential buyers who agree to try it and to provide an evaluation of its performance. Sales demonstrations simply present the product to a sample of prospective customers in an effort to learn how many would purchase it. Pretest market forecast for consumer durables can be obtained from laboratory research in much the same way as discussed above. Urban, Hauser, and Dholakia describe project involving a new 1985 car that was a down sized version of an existing car which promised economy without the loss of luxury. Target group members were brought to a clinic where they saw ads for a new car, drove a prototype auto, and saw video tapes that simulated word of mouth recommendations from consumers. Measures of preference and choice were taken before and after exposure to ads, drive, and videotapes. Two-thirds of the respondents drove the new car, and one-third drove the existing car. By comparing the responses for the new car relative to the existing car and simulating the growth in awareness, dealer visits, and word-of-mouth communication, a four year, life cycle sales forecast was produced.
The results showed that the new car was better but that its sales would be lower than management objectives.