Forecasting Consumer Acceptance of New Products For Multiple Market Segments Using Multiple Methods

Arch G. Woodside, Tulane University
Elizabeth J. Wilson, Louisiana State University
Nicholas T. van der Walt, University of Canterbury
Roderick J. Brodie, University of Canterbury
ABSTRACT - Tests of customer acceptance of a new medical-related product are reported. The potential target markets were sampled for the Ultrarest Couch, a consumer durable designed to relieve back pain. The three markets were physician patients, chiropractor patients, and middle-aged office workers. Five prices, two promotional messages, and five endorsement conditions were tested. Results: contrary to expectations, physician patients more readily accepted the new product; a strong segment-by-price interaction effect was found to influence customer acceptance.
[ to cite ]:
Arch G. Woodside, Elizabeth J. Wilson, Nicholas T. van der Walt, and Roderick J. Brodie (1989) ,"Forecasting Consumer Acceptance of New Products For Multiple Market Segments Using Multiple Methods", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 326-331.

Advances in Consumer Research Volume 16, 1989      Pages 326-331

FORECASTING CONSUMER ACCEPTANCE OF NEW PRODUCTS FOR MULTIPLE MARKET SEGMENTS USING MULTIPLE METHODS

Arch G. Woodside, Tulane University

Elizabeth J. Wilson, Louisiana State University

Nicholas T. van der Walt, University of Canterbury

Roderick J. Brodie, University of Canterbury

ABSTRACT -

Tests of customer acceptance of a new medical-related product are reported. The potential target markets were sampled for the Ultrarest Couch, a consumer durable designed to relieve back pain. The three markets were physician patients, chiropractor patients, and middle-aged office workers. Five prices, two promotional messages, and five endorsement conditions were tested. Results: contrary to expectations, physician patients more readily accepted the new product; a strong segment-by-price interaction effect was found to influence customer acceptance.

As a theory, market segmentation is the grouping of potential customers into sets that are homogeneous in response to some elements of the marketing mix. As a strategy, it is the allocation of marketing resources, given a heterogeneous customer population. Thus the identification of segments, homogeneous in response, allows the refinement of marketing strategy (Lilien and Kotler 1983, p. 289).

The present paper is an application of consumer judgment modeling to measure responses to new products among several segments of potential customers. The reported application relates most to the theoretical work and empirical studies on market segmentation by Wind (1978) and his colleagues (Wind & Spitz 1976; Wind, Grashof & Goldhar 1978).

A key finding supported by several studies (cf. Page and Rosenbaum 1987; Wind, Grashof, & Goldhar 1978; Cattin and Wittink 1982; Scott & Keiser 1984; de Kluyver & Whitlark 1986; Woodside, Sanderson, & Brodie 1988) is that market segments (both consumer and industrial) are likely to exhibit different response sensitivities to new product-service designs. Consequently, successful planning and implementation of new product-service launches require fine-tuned matches of product-service bundles to targeted market niches.

Hypotheses and Experimental Design

The following central hypothesis (H1) was tested: consumer acceptance for new product designs differ by market segments; that is, (H1a) each market segment has a unique response function to a set of multi-attribute product alternatives and (H1b) potential customers in one segment versus potential customers in other segments are more interested in buying the product category.

Several specific hypotheses were developed based on the insights related to three potential customer segments and the information needs of the managers directing the marketing strategy for the new product. The managers were concerned most with pricing, positioning strategy (relief versus prevention), and type of medical endorsement to use to promote the new product. The specific hypotheses are stated explicitly following descriptions of the product and potential markets.

The New Product

The product tested, "Ultrarest," is a new medical couch designed-to provide relief from pain and discomfort for sufferers of disc and facet joint problems. The product was designed in 1985-87 by a licensed chiropractor practicing in Ashburton, New Zealand. The chiropractor and his wife also owned and operated a small, medical furniture manufacturing enterprise. The medical couch represented a major product line extension for the manufacturing enterprise. At the time of the study (June and July 1987), prototypes of the medical couch had been tested successfully and the designer had applied for a New Zealand patent; manufacturing of the product was about to start.

Market Segments

Management identified three, naturally occurring, market segments as potential customer targets for the new product: back-pain patients of 0 medical doctors, patients of chiropractors, and middle-aged office workers experiencing chronic back-pain. Each of the three market segments posed unique opportunities and problems in marketing the new product: strategic decisions in pricing, distribution, and promoting Ultrarest depended on which of the three market segments was selected as the primary, potential customer target. Ultrarest's product manager t wanted information on potential customers' levels of interest in receiving information and in buying the new product, as well as information on the potential customers' sensitivities to changes in price, positioning strategy, and medical endorsements on using the product. The product manager believed that chiropractor patients would be the most interested and middle-aged office workers would be the least interested market segment in buying the new product. Research evidence gathered independently of Ultrarest's product management was requested by the managers to support or refute the following speculations.

Based on having a continuing, active file of patients seeking prevention of back-pain by receiving periodic treatments, Ultrarest's designer believed that chiropractor patients were more likely to be affected by positioning the product as a method of preventing versus relieving back-pain. The opposite was hypothesized for patients of physicians; patients actively under the care of physicians for disc problems and back-pains were believed to be more sensitive to positioning Ultrarest as providing back-pain relief versus prevention. Middle-aged office workers, experiencing occasional back-pain but not actively seeking medical care, were hypothesized to be more influenced by the positioning Ultrarest as providing relief rather than preventing back-pain; management believed this market segment to be the least interested in buying Ultrarest because their back-pain did not require medical attention.

Ultrarest's management believed that physician patients versus the other two market segments would be less sensitive to price levels of the new product because of the continuing back-pain being experienced most often by physician patients. Also, the new product designer believed physician patients were more likely to have higher, household disposable incomes than chiropractor patients; no independent survey evidence was available in New Zealand to support this observation.

Information on potential customer sensitivities to product endorsements by three independent associations was desired. Having the new product endorsed by the New Zealand Chiropractors Association was believed by Ultrarest's management to have a positive influence on the product's acceptance, especially among chiropractor patients.

A product endorsement by the New Zealand Medical Association, the national association of physicians, was believed to have a positive influence on the product's acceptance, especially among physician patients; this endorsement condition was believed to be more influential than the chiropractor endorsement among office workers because Ultrarest's management believed medical doctors versus chiropractors had more prestige and a more positive image among the general population of adults in New Zealand. -

An endorsement by the New Zealand Industrial Design Council was also tested. The Council's "DesignMark" for excellence in furniture design and quality was hypothesized to increase customer acceptance of the new product, especially among office workers. Ultrarest's management believed that the Council's endorsement would support the additional claim for the product as fitting "the domestic scene as a piece of furniture." Patients for back-pain relief and prevention versus non-patients were perceived to be less sensitive to a non-medical endorsement because of their immediate need for health care attention.

Specific Hypotheses and Models

Based on the discussed insights and information needs of Ultrarest's management, the following specific hypotheses were predicted to be supported by the research study: (H2) positioning Ultrarest as preventing versus relieving back-pain would be more effective in increasing product acceptance among chiropractor patients; (H3) positioning Ultrarest as providing relief versus preventing back-pain would be more effective in increasing product acceptance among medical patient physicians and middle-aged office workers; (H4) physician patients' product acceptance is less sensitive to the price of the new product compared to chiropractor patients and middle-aged office workers; (H5) physician patients are influenced more positively by a medical physician endorsement than a chiropractor or industrial design endorsement for the new product; (H6) chiropractor patients are influenced more positively by a chiropractor endorsement than a physician endorsement for the new product; (H7) middle-aged office workers versus patients of chiropractors and medical doctors are more influenced positively by the industrial design endorsement; and (H8) the combination of a health care endorsement (physician or chiropractor) and industrial design endorsement will influence consumer judgments to a greater extent than using only one endorsement.

The general main effects model tested in the study included the three marketing variables:

Y'i=a + b'1M - b'2P + b'3 E     (1)

where Y'i = is the predicted customer judgments (e.g., willingness-to-buy) for the dependent variables used in the study

M = positioning message

P = specific price level

E = endorsement

b'i = weights of relative influence of the three marketing variables

a = a regression constant.

Market segment was also hypothesized to have a direct influence on customer judgments (H1b). Ultrarest's product manager believed that chiropractor patients would be more likely to buy the new product than other potential customers:

Y'i = a + b'S     (2)

where S = market segment.

If dummy coding (Pedhazur 1982, Chs. 9-10) is used to test the second model (with S = 1 for chiropractor patients and S = O for all other potential customers), then b' > O would support the product manager's hypothesis. Alternatively, the average consumer judgment on willingness-to-buy should be highest for chiropractor patients versus other potential customers.

Several of the specific hypotheses are statements of interactions of marketing variables with market segments. For H4, physician patients are hypothesized to be less price sensitive than the other market segments. To confirm this hypothesis, with S = 1 for physician patients and S = O for all other potential customers, then the following model should be supported empirically:

Y'i = a - b' (S X P)     (3)

where (S X P) = the interaction of market segment and price.

Alternatively, the average decline in willingness-to-buy the new product caused by price increases for physician patients should be less than the average declines caused by price for other market segments.

For the analysis of the data ordinary least squares (OLS) estimations were employed. Research in consumer judgment modeling (Anderson 1981; Jain, Acito, Malhatro, and Mahajan 1979; Wittink and Cattin 1981; Scott and Keiser 1984) has shown that the level of the data (interval or ordinal) and the method of estimation (metric or non-metric) have little effect on predictive accuracy.

Experimental Design Used in the Study

An orthogonal fractional-factorial experimental design (Banks 1965) was used to estimate the impact of marketing variables (price, message, and endorsement) on consumer judgments of representative samples of the three market segments. Five levels of price, two message conditions, and five endorsement conditions were included in the experiment.

The experiment consisted of two experimental blocks (one block for each of the two messages) of five price levels by five endorsement conditions. A between-subjects design was used to measure the message effect; the subjects were assigned randomly to one of the two message blocks. A within-subjects design was used to measure price and endorsement effects; each subject was exposed to 5 of the 25 treatments per block. Each of 80 subjects was assigned randomly to one of five independent combinations (A, B, C, D, and E) pet block. Thus, a subject was exposed once to each level of price and each level of endorsement in combination. The order of presentation of the five treatments was randomized for each subject.

For each subject five sets of measures of three dependent variables were recorded. A total of 400 observations were collected for each dependent variable: 5 observations by 80 subjects.

The subjects in the study were exposed to a limited number (five) treatment conditions to minimize fatigue and to parallel real-life evaluations of a few, not one, product models or brands (cf. Greenwald 1976). The subjects were requested to examine each of the products described by a written summary and shown visually; consumer judgments on three dependent variables were made for each exposure.

Price. Five retail price levels were used (NZ dollars): $500, $600, $720, $870, and $1,000. (NZ $1.00 = US $0.62.) These price levels were selected based on management and research judgment; the product manager had expected the retail price for Ultrarest to be $870, he judged the $500 and $1,000 prices to be immediately outside the range of prices that would provide marginal profits or market acceptance. The differences in prices represent about 20 percent increments; price changes believed likely to influence consumer judgments and permitting coverage of a broad range of price points.

Message Conditions. The two message conditions (relief versus prevention) were varied within the product descriptions and for the two health care endorsement conditions provided to the subjects participating in the study.

The following specific product descriptions are examples of those used in the experiment. Relief: "The Ultrarest Couch [C8] provides the user with welcome relief from chronic lower back pain." Prevention: "The Ultrarest Couch [G7] helps to prevent the user from experiencing chronic lower back pain."

Endorsement Conditions. Five product endorsement conditions were tested in the study: endorsements made by (1) the New Zealand Medical Association of Physicians, (2) the New Zealand Chiropractors Association, (3) the New Zealand Industrial Design Council, (1) and (3), and (2) and (3). The combination of endorsements for the two health care organizations (1 and 2) was unrealistic; consequently, this condition was not tested.

The following endorsements are examples of ones used in the experiment: 'The Ultrarest Couch [G7] has been found to be effective in preventing chronic lower back pain and promoting good spinal health by the New Zealand Chiropractors Association;" 'The Ultrarest Couch [B4] has been found to be effective in providing temporary relief to chronic back pain by the New Zealand Medical Association of Physicians;" "The Ultrarest Couch [C8] has been awarded the DesignMark for excellence in furniture design and quality by the New Zealand Industrial Design Council."

Dependent Measures. Green and Srinivasan (1978) and Scott and Keiser (1984) suggest that multiple methods be used to test hypotheses involving consumer judgments as a check on the robustness of the results. Both rating and discrete choice modeling techniques were used in the present study in response to this suggestion.

In the rating task each subject was requested to "please place a tick- mark somewhere across the line below that would indicate your willingness to purchase The Ultrarest Couch [C8] based on the above description, price, and endorsement." A 35-point dashed line was used in the rating.

Two discrete choice tasks were used. The first was based on the procedure developed by Politz (reported by Banks 1965, p. 40). Each subject was requested to assume s/he had "$2,000 to spend on two to three of the following products (you would keep any remaining funds not spent). Please tick the two or three products you would buy." Five products and services with prices were listed including the Ultrarest Couch. This choice task was used to learn the extent of purchase intention of the new product if the potential customers had unexpected disposable income available for durable purchases.

The second discrete choice task involved a behavioroid measure. Each subject was asked if s/he wanted "to receive additional information in the post about using the Ultrarest Couch [AS] as described above? Please tick one space below." A behavioroid measure comes close to requesting some meaningful commitment toward the product from the subject.

Subjects. The subjects in the study included 20 chiropractor patients, 8 medical doctor patients, and 52 middle-aged (ages ranged from 35 to 55) office workers. The chiropractor and physician patients in the study had received back treatments and/or prescriptions for back pain within 30 days prior to data collection; all patients were actively receiving chiropractic or medical care for back problems at the time of the study. Only office workers reporting experiencing periodic, chronic back pain were included in the study.

Procedure. Cooperation for the study was granted by a chiropractor doctor and a medical doctor located in the same neighborhood community in Christchurch, New Zealand. (Christchurch has a population of 500,000 and is located on the South Island, 80 kilometers north of Ashburton.) Patients visiting their doctors and waiting for their appointments were approached and requested to participate in the study. A total of three physician and one chiropractor patients approached refused to participate.

Office workers in four organizations participated in the study. The organizations' administrators were contacted via part-time, evening, graduate students who worked full-time at the organizations. Each organization employed more than 100 office workers at the building where data for the study were collected. Middle-aged office workers agreeing to participate in the study and reporting periodic, chronic back pains were interviewed individually. A total of four office workers meeting the required profile refused to participate.

For each treatment condition a photograph was shown of the Ultrarest Couch being used by a model above a written description of the product. The written description of the product included the positioning message and price combinations. One of the five endorsement conditions appeared below the product's description. The three dependent measures appeared at the bottom of the page.

Limitations. Even though care was taken to include subjects representative of three identified market segments, only very limited numbers of subjects were included in each sample. Unfortunately, the sample of physician patients was particularly small; five physicians were contacted but only one agreed to allow his patients to participate in the study. While the application of multiple methods increases the robustness of consistent significant findings, the validities of how well the models tested actually predict purchase behavior were not tested. Only visual and written descriptions of Ultrarest were provided to subjects.

Results

The first hypothesis was supported. Consumer acceptance of the Ultrarest Couch did differ among the three sample groups. The product manager's prediction of greater acceptance of the new product among chiropractor patients was not supported; greater acceptance was found among the MD patients versus subjects in the other two for all three dependent variables.

Both individual and grouped analysis (cf. Bass, Tigert, and Lonsdale 1968) were performed to test the proposed models. Both H1a and H1b were supported. A major finding in the analysis was that the main effects of price and sample group, and the interaction of these two variables, influenced acceptance of the Ultrarest Couch.

For the grouped data and dummy coding assigned to MD patients and the other subjects (1, 0 respectively), the main and interaction effects model of price and customer segment explained 91 percent of the variation in likelihood of purchase. (The mean purchase likelihoods for the five price levels by MD patients versus other subjects were used to estimate the grouped data model; regression analysis for grouped versus individual data has been recommended by Bass, et al. [1968] to achieve stable and relevant estimates of the systematic effects in market segment analysis.) Results for this analysis are shown in Table 1.

For the individual analysis on the same model the findings were significant but the model explained only 8 percent of the variation in likelihood of buying the Ultrarest Couch.

Promotional Message Results H2 was not supported. The message main effect was not significant on influencing acceptance of the new product among chiropractor patients nor among office workers in the study.

H3 was partially supported. Among the MD patient sample the back pain relief message increased acceptance of the Ultrarest Couch compared to the prevention message. This finding held for all three dependent variables. This pattern was not found for the other two subject samples. The average responses for the two message treatments among the MD patients appear in Table 2. The main and interaction effects of message and price are significant statistically (p < .05) for the discrete choice findings shown in Table 2.

Caution is needed in interpreting these results since the sample sizes for the two message treatments for the MD patient sample are very small (the number of observations across the five prices is 20, n = 4 subjects, for each message treatment).

Price and Customer Segment Results H4 was not supported. MD patients were found to be affected more by the price treatments than the other two subject samples. The analyses of the two discrete dependent variables provided additional evidence supporting a main and interaction effects model of price and physician patients positively influencing acceptance of the Ultrarest Couch.

Endorsement Treatment Results The remaining hypotheses (H5 through H8) were not supported. No consistent and clear pattern of significant results was found for the five endorsement conditions influencing acceptance of the new product. The interaction effects of endorsements by segments were not significant statistically.

TABLE 1

LIKELIHOOD OF PURCHASE MODEL FOR SEGMENT AND PRICE BY GROUPED DATA

TABLE 2

BACK-PAIN RELIEF VERSUS PREVENTION MESSAGE

Conclusions and Implications

The findings of the study support and illustrate a proposition relevant to market segmentation theory: potential customers grouped into segments by readily identified buying behavior variables, e.g., physician versus chiropractor patients, are likely to differ in their response sensitivities to marketing variables. Empirical tests of hypothesized consumer judgment models are likely to be helpful in selecting potential customer targets and in adjusting product designs and promotional messages.

Four decisions were made by the Ultrarest Couch product manager based on the results of the study. (1) Physician patients as well as chiropractor patients were selected as a target market for the new product; before the study, chiropractor patients only were planned as potential customers. (2) Since the average likelihood of purchasing the new product was lower than expected by the product manager, a retail rent with option to purchase was implemented. (3) The use of the DesignMark endorsement only was included in promotional literature on the Ultrarest Couch; the product manager decided not to attempt to gain endorsements from the two medical related associations. (4) The promotional message selected for the Ultrarest Couch was focused on back pain relief: 'The Ultrarest Couch relieves your back-pain."

Research on acceptance of new products that incorporates consumer judgment modeling can be useful in revising marketing strategies even for small manufacturing enterprises. Some empirical model testing of tentative marketing- decisions is recommended even if the study is limited by sample size and budget. The total expenditure for the Ultrarest Couch study was NZ $3,500.

The use of several methods of measuring customer acceptance of new products is recommended. Consistent results provided from testing customer acceptance of new products across several dependent variables represent an important step in replicating the findings and extending the implications.

REFERENCES

Anderson, James C. (1981), "A Potential Problem in the Use of Orthogonal Arrays in Conjoint Analysis," AIDS 1981 Proceedings, I, in G. R. Reeves and R. J. Sweigart (Eds.) American Institute for Decision Sciences, 278-280.

Banks, Seymour (1965), Experimentation in Marketing, New York: McGraw-Hill.

Bass, Frank M., Douglas Tigert, and Ronald Lonsdale (1968), "Market Segmentation: Group versus Individual Behavior," Journal of Marketing Research, 5 (August), 264-270.

Cattin, Philippe, and Dick R. Wittink (1982), "Commercial Use of Conjoint Analysis: A Survey," Journal of Marketing, 46 (Summer), 4453.

de Kluyver, Cornelis A., and David B. Whitlark (1986), "Segmentation for Industrial Products," Industrial Marketing Management, 15, 273-286.

Green, Paul E., and V. Srinivasan (1978), "Conjoint Analysis in Consumer Research: Issues and Outlook," Journal of Consumer Research, 5 (September), 103- 123.

Greenwald, Anthony (1976), "Within-Subjects Designs: To Use or Not to Use?" Psychological Bulletin, 33, 314-320.

Jain, A. K., Franklin Acito, Naresh K. Malhotra, and Vijay Mahajan, (1979) "A Comparison of the Internal Validity of Alternative Parameter Estimation Methods in Decompositional Multiattribute Preference Models, Journal of Marketing Research, 16 (August), 313-322.

Lilien, Gary L., and Philip Kotler (1983), Marketing Decision Making: A Model-Building Approach, New York: Harper & Row.

Page, Albert, L., and Harold F. Rosenbaum (1987), "Redesigning Product Lines with Conjoint Analysis: How Sunbeam Does It," Journal of Product Innovation Management, 4 (June), 120-137.

Pedhazur, Elizhu J. (1982), Multiple Regression in Behavioral Research, New York: Holt, Rinehart and Winston.

Scott, Jerome, and Stephen K. Keiser (1984), "Forecasting Acceptance of New Industrial Products," Journal of Marketing, 48 (Spring), 54-67.

Wind, Yoram (1978), "Issues and Advances in Segmentation Research," Journal of Marketing, 15 (August), 317-337.

Wind, Yoram, and Lawrence K. Spitz (1976), "Analytical Approach to Marketing Decisions in Health-Care Organizations," Operations Research, 24 (September-October), 973-989.

Wind, Yoram, John F. Grashof, and J. D. Goldhar (1978), Journal of Marketing, 42 (Summer), 27-37.

Wittink, Dick R., and Philippe Cattin (1981), "Alternative Estimation Methods for Conjoint Analysis, Journal of Marketing Research, 18 (February), 101-106.

Woodside, Arch G., R. Hedley Sanderson, and Roderick J. Brodie (1988), 'Testing Acceptance of a New Industrial Service," Industrial Marketing Management, 17, 65-71.

----------------------------------------