Assessing the Impact of Increased Product Safety on Consumer Utility

ABSTRACT - One of the costs of product safety regulation may be its effect on consumer utility. The paper presents a pilot application of conjoint analysis as a methodology for measuring the consumer utility impact of proposed safety regulation. Pending government decisions on minimum wet braking standards for bicycles and flame-resistant standards for clothing were selected for the analysis.

Citation:

Stephen C. Burnett (1978) ,"Assessing the Impact of Increased Product Safety on Consumer Utility", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 186-193.

Advances in Consumer Research Volume 5, 1978      Pages 186-193

ASSESSING THE IMPACT OF INCREASED PRODUCT SAFETY ON CONSUMER UTILITY

Stephen C. Burnett (student), Indiana University

ABSTRACT -

One of the costs of product safety regulation may be its effect on consumer utility. The paper presents a pilot application of conjoint analysis as a methodology for measuring the consumer utility impact of proposed safety regulation. Pending government decisions on minimum wet braking standards for bicycles and flame-resistant standards for clothing were selected for the analysis.

INTRODUCTION

Since the creation of the U.S. Consumer Product Safety Commission (CPSC) in 1972, there has been an increasing awareness among marketers that growing public and private concern over product safety will have significant implications for marketing programs. This raises questions about the probable effects of product safety regulation and how marketing should respond. Some of the suggestions offered include incorporating safety testing in new product planning (Trombetta and Wilson, 1975), product safety as a competitive strategy (Edwards, 1975; Loudenback and Goebel, 1974), and more stringent labeling and record-keeping procedures (Jensen, Mazze, and Stern, 1973). In addition to assessing the marketing implications of product safety, there has also been an increasing interest in how consumer research can assist in product safety policy decisions (Staelin and Pittle, 1977; Miller and Parasuraman, 1974) and public policy decision-making in general (Day, 1976; Wilkie and Gardner, 1974).

While consumer research in the product safety area has not been overly abundant and its focus has been somewhat fragmented, as would be expected at this early stage, the research issues emanating from the regulatory role of the CPSC are becoming more apparent. For products within its authority, the Commission has several policy alternatives including: (1) mandatory safety regulation, (2) voluntary industry standards, and (3) consumer information/education campaigns (CPSC, 1976a). Although all of these alternatives require some knowledge of consumers' behavior in order to be evaluated for effectiveness, the specific research issues raised by each are quite different.

It has been reported that only 20% of the total incidence of consumer product-related injury and death is address-able by improved product design (Staelin and Pittle, 1977). The remainder is a consequence of consumers carelessly using relatively safe products. Because of this fact, consumer information and education has become a highly attractive policy alternative and one that has begun to benefit from ongoing consumer research. A continuing project reported by Staelin (1977), Staelin and Weinstein (1974), and Thompson (1974), has concentrated primarily on whether information and education can enhance safer consumer behavior in using products. Although the researchers have found that knowledge of safety principles correlates positively with safer product use behavior, the direction of causality between knowledge and behavior remains unresolved.

A second policy alternative available to the Commission is to encourage and assist industry trade associations in the development of voluntary minimum safety standards. Despite their regulatory cost appeal, benefits of voluntary standards are obviously dependent upon the degree to which they are implemented. Whether improved safety standards will be incorporated in products without government mandate raises the question of safety marketability. The previously cited study by Staelin and Weinstein (1974) attempted to gauge the importance of safety in purchase decisions. Their findings suggest that either because of the demand characteristics or the multidimensional nature of safety, its saliency as a product attribute varies with how its importance is measured. Using open-ended questioning, safety was found not to be an influential decision variable. However, when subjects were asked to rate the importance of safety as one of eight predetermined purchase variables, it took on a more central role in the respondents' purchase decisions. A similar result has been reported by Kuehl and Simon (1973).

OBJECTIVE

The final CPSC policy alternative and the focus of the present study is to set minimum safety standards and require compliance under the threat of fines and even criminal prosecution. While this action eliminates much of the uncertainty in the actual delivery of safer products to consumers, its evaluation requires an estimation of the costs and benefits expected to accrue to consumers from a Mandatory Safety Regulation (MSR).

Consumer benefits from an MSR may take several forms, but primarily these are reduced probabilities of injury and death. There are numerous components of MSR costs. From a macro-economic perspective, some of these costs are fairly tangible such as additional resources to produce the product and expenditures for monitoring and enforcing the standard. Other cost components such as industry disruptions and increased industry concentration are more difficult to quantify.

Another cost of an MSR may be its net effect on consumer utility. As used here, utility refers simply to the total value a consumer derives from the consumption of a good or service. The concept of utility has its origins in economics, but with the growing popularity of multidimensional scaling techniques, it has more recently been applied to marketing decisions (Green and Wind, 1975; Green and Wind, 1973; Green and Rao, 1971). For a particular MSR, there may be several potential sources of utility impact. One of these is generally a price increase although an increase in price may be only one of many utility tradeoffs a consumer must make in order to obtain the utility of enhanced safety. The proposed MSR for flame-resistant treatment of clothing provides a good illustration.

The CPSC estimates that an MSR requiring all consumer apparel to meet proposed minimum flame-resistant standards would require an average price increase of 25%. In addition, the following garment performance characteristics would be affected: (1) a reduction in the wash and wear properties of some garments, (2) a reduction in the variety of fabrics available and wearer comfort, (3) a decrease in the average life of treated fabrics, and (4) some limitations on styling variations (CPSC, 1976c). Assuming that consumers derive utility from any of these attributes, then this particular MSR may have a utility cost if the increase in utility from having flame-resistant clothing fails to offset the disutility from decreases in the performance of other product characteristics.

It is useful to note that the question of trading off utility as posed here is not concerned with whether a safety feature or performance standard is an important purchase decision variable in an absolute sense. For example, the degree to which a bicycle has the capacity to stop under wet conditions may have only minimal influence on most bike purchase decisions. Nevertheless, if utility derived from improved braking at least offsets the disutility of the price increment necessary to provide better braking, then there is no net utility cost to the consumer even though color, style, and retail outlet are the major purchase determinants. The objective of the present study was to explore the problems and potential of conjoint analysis as a technique for assessing such net utility costs. Pending CPSC decisions on minimum wet braking standards for bicycles and flame-resistant standards for clothing were selected for the application.

The idea of measuring the utility cost of government regulation is not new. Walker, Sauter, and Ford (1974) provided a conceptual framework, and Walker and Sauter (1974) used Thurstone's Law of Comparative Judgement to intervally scale consumer preference for alternative retail credit contracts. The researchers' hypothesis was that regulation reducing consumer credit interest rates may cause retailers to compensate for lower interest charges by requiring higher down payments and monthly installments plus increasing the price of the merchandise. When these strategic "secondary effects" are taken into account, total consumer utility may be decreased by the credit regulation.

In terms of the type of public policy evaluation issue addressed, the present study parallels the works cited above. There are, however, two distinctions: one with respect to the requirements of the policy area, the other with respect to methodology. For consumer credit regulation, utility tradeoffs result from strategic responses of retailers. Because of the uncertainty concerning what these changes would be, Walker and Sauter (1974) had to rely on rough estimates. Given this constraint, their conclusions hold only to the extent that these approximations reflect how retailers would actually respond. For product safety decisions, there fortunately is somewhat less imprecision in determining both the product characteristics that would be affected and by what amounts. Product modifications are conducive to the use of engineering tests and prototypes which provide relatively accurate estimates of performance tradeoffs.

The second difference between the two studies concerns the flexibility of Thurstone scaling versus conjoint measurement. Both scaling models may require identical subject tasks when scaling the average utility for a group of consumers. In fact, Curry and Rogers (1977) recommend using the two methods in tandem when aggregating responses across subjects. If there is an interest in utility at the individual subject level, however, the Thurstone model requires each stimulus or product concept to be presented to a subject a large number of times (Torgerson, 1958). With conjoint measurement the scaling of individual respondents is much easier since the stimuli must be judged only once. Because individual subject comparisons were important to the present study, conjoint measurement was the preferred technique.

STUDY DESIGN

Since the purpose of the study was exploratory, a convenience sample of 40 undergraduate students was used. This made it necessary to select products with which students are familiar. Ten-speed bicycles and shirts or blouses were thought to meet this criterion fairly well. Also, a preliminary questionnaire indicated that two-thirds of the subjects owned 10-speed bicycles. As previously noted, the flame-resistant standard now pending before the CPSC would affect a garment's price, durability, and easy-care characteristics (CPSC, 1976c). The only impact of the proposed wet braking standard for bicycles is an estimated price increase of 10% (CPSC, 1976b).

Factor Descriptions and Level Specification

The general procedure in conjoint analysis to to have subjects judge in terms of their preference or some other criterion a series of product descriptions or scenarios. The products systematically differ on several attributes (factors) by specific amounts (levels). With a subject's preference ranks as input, the conjoint measurement algorithm derives intervally scaled utility weights (part-worths) for each factor level. With an additive scaling model, such as Kruskel's MONANOVA used in this study, the sum of the part-worths must preserve the original rank orders (Green and Wind, 1973). [MONANOVA is a popular nonmetric model and has been found to be appropriate in a variety of studies. It should be noted that other methods are also suitable for analyzing conjoint data (e.g., see Cattin and Wittink, 1977).] To determine the degree to which a subject's utility is altered by paying $150 rather than $165 for a particular product, one only needs to examine the absolute change in part-worths for the two prices relative to the changes in part-worths of the other factor levels. The greater the relative utility change, the greater the utility attached to the product attribute.

A critical design aspect of conjoint analysis is the specification of the factor levels. The relative utility of a factor is a direct function of its levels. For example, the part-worth of a lower price could be increased dramatically by selecting polar price levels such as $100 and $300 as opposed to $100 and $120. Thus, when scaling the utility of a proposed safety feature versus the utility of tradeoff attributes, it is crucial that objective criteria exist for selecting the factors and their levels.

For the present study, limited CPSC test reports were available to guide factor level specification. The only quantitative estimates published by the Commission were anticipated price increases for the two safety feature proposals. For this reason, factor levels were set based on data from recent Consumer Reports tests of 10-speed bicycles and men's shirts. Because of this limitation, the relative utilities of the factors reported later in the paper should be interpreted solely as an illustration of the informational value of conjoint analysis. Different factor levels or different ways to operationalize the factors could produce different findings from those obtained here.

Exhibit 1 lists brief descriptions of the bicycle factors and their levels. For both shirts and bicycles, four factors at two levels were used for a 24 full factorial design. Since the wet braking standard causes only a price tradeoff, in theory braking and price factors would have sufficed. However, because prior studies (Kuehl and Simon, 1973; Staelin and Weinstein, 1974) suggest that the topic of safety may be subject to demand characteristics, two additional moot factors were used in conjunction with price and braking. This was done to insure that respondents would be unaware of the study's purpose and to make the task more realistic.

EXHIBIT 1

BICYCLE FACTOR DESCRIPTIONS AND LEVELS

The inclusion of moot factors raises the important methodological question of whether different sets of moot factors will alter the relative utilities of the price and braking factors. To address this question, subjects were randomly assigned to one of two groups (labeled Forms A and B). Form A subjects received overall pedaling ease and frame quality as moot factors while Form B subjects were given handling precision and shifting ease. Although any bicycle attributes would have served equally well, all four of the moot factors were characteristics of bicycles tested by Consumer Reports.

Before discussing the subject's task, one other aspect of the factor descriptions should be mentioned. Past research on safety knowledge and safer product use behavior has not explicitly dealt with the effect of safety knowledge on product choice behavior. To explore how conjoint analysis might be helpful in examining a link between a consumer's knowledge of a product's hazards and preference for a safety feature, subjects were also randomly divided into two groups for the shirt/blouse task (again labeled Forms A and B). Both groups were told in the instructions that a garment was rated as flame-resistant if it would not ignite when exposed to an open flame. The Form B instructions, however, contained the following introductory sentence to the flame-resistant definition:

"Because some 28,000 people are burned annually from clothing fires, Consumer Reports tested each garment for flame-resistant characteristics."

The hypothesis to be tested was that Form B subjects would attach higher utility to a flame-resistant feature than Form A subjects because of greater knowledge of clothing fire hazards.

This hypothesis and its conceptual underpinnings are far too simplistic to yield substantive conclusions. Knowledge of a product's hazard potential could be manipulated on several dimensions such as the probability of injury to a specific subject or to people in general. Moreover, both the probability of an injury and its severity could influence safety feature preference to different degrees. The strength of the statement could also have implications for utility. The study's mild manipulation of knowledge was included in the design primarily to illustrate the technique's potential for experimentation with safety knowledge. Complete descriptions of the shirt/blouse factors and levels are shown in Exhibit 2.

EXHIBIT 2

SHIRT/BLOUSE FACTOR DESCRIPTIONS AND LEVELS

Subjects' Task

The task required of the subjects was divided into two sections. They were first asked to rank in terms of their preference all 16 product combinations. To simplify the ranking task, subjects were instructed to sort the products into high and low preference piles, to rank order each pile, and then to finally combine both piles into one strict rank. The subjects' second task was to provide self-explicated factor importance weights. This was accomplished with a constant sum scale. They were asked to divide 100 points among the factors such that the point assignments indicated the relative degree to which each factor and its levels influenced their product preference.

The reason for obtaining self-explicated factor importance weights was to check for convergent validity. Deriving the relative importance of product attributes from rank orders of product scenarios is a distinctly different measurement method from having subjects directly report this importance. The extent to which the two measures correlate should be indicative of convergent validity. The instructions and tasks for bicycles and garments were identical.

ANALYSIS OF RESULTS

The initial step of the data analysis was to identify subjects who either could not perform the ranking task or whose decision rules were inappropriate for MONANOVA's additive composition rule. The criteria for identifying and eliminating these subjects from additional analysis were stress values and factor directionality. Stress is the badness-of-fit measure of the scaling model to the original data. Factor directionality refers to whether a subject's utility increases or decreases in the hypothesized manner with changes in a factor's levels. For both products, an increase in price was hypothesized to lower utility while increased utility was expected for the higher levels of all other factors.

Of the total number of rank orders submitted to MONANOVA (n = 35 for bikes, n = 40 for garments), only three subjects exhibited stress values greater than the .15 cutoff point. Five subjects violated hypothesized directionality. The low percentage of subjects eliminated is evidence that the subjects could perform the task as requested and that the additive composition rule of MONANOVA was adequate. In fact, 82% of the total rank orders had stress values of less than .01.

The means and standard deviations of the MONANOVA and self-explicated factor importance weights are shown in Table 1. The method for obtaining self-explicated weights has been discussed, but the MONANOVA factor importance weights require explanation. Based on a subject's 16 rank orders, the output of the model is eight intervally scaled utility values--one for each factor level. Since these part-worths comprise an interval scale with unique origin and unit for each subject, it is not permissible to compare part-worths across subjects. It is necessary therefore, to convert the part-worths for each subject into four ratio scaled factor importance weights (FIW). These are computed by simply dividing the absolute value of the change in the two part-worths of each factor by the total absolute value of changes in part-worths for all four factors. Factor importance weights may be interpreted as the percentage contribution of a particular factor to a subject's total utility for all factors. [Another measure of a factor's importance is the percentage of total variance it explains.] As an illustration, if subject X's FIW for price is 25 compared to 50 for subject Y, then Y may be said to attach twice as much utility to the lower price than X. Also, Y's utility for the lower price is equal to his combined utility for all other factors.

From Table 1, it is apparent that the standard deviations of the MONANOVA factor weights were high relative to the mean values. This is a strong indication that it may be inappropriate to compute an average group scale since subjects tended to have quite different utility functions. Because the sample was demographically homogeneous, the high variance may suggest the existence of utility profile segments resulting perhaps from non-demographic variables. An extension of the study, which would help better understand safety feature utility, would be to form clusters of subjects with similar factor importance weights. These clusters could then be examined for systematic differences in such variables as attitudes toward safety, product experience, and safety knowledge and behavior.

Another interesting finding is evident when the means of the self-explicated weights are contrasted with the MON-ANOVA weight means. For bicycle Forms A and B separately and the combined shirts/blouses forms, [As discussed later, garment Form A and Form B groups did not significantly differ on any of the factor importance weights and were aggregated for the bulk of the analysis.] the self-explicated price factor weights were greater than the MONANOVA price weights. Similarly, subjects tended to overestimate the importance of the durability factor for garments compared with the importance of durability as reflected by their preference ranking behavior. It is not completely apparent why price and durability were overestimated. One could perhaps speculate that price and durability are common and rational product attributes which may not influence product preference to the extent that self-report measures would imply. Despite these differences, the product moment correlations for the MONANOVA and self-explicated weights shown in Table 2 were statistically significant (p < .05) for all factors. This agreement is evidence of the convergent validity of the two measures.

One last aspect of the individual MONANOVA factor importance weights should be noted. For bicycle Form A and B groups, the utility of improved wet braking more than compensated for the disutility of the 10% price increase when averaged over subjects. Conversely, the utility of the flame-resistant feature was far outweighed by the utility of price, durability, and wash and wear properties. This finding is consistent with the CPSC's opinion that consumers probably would not be willing to forego the benefits of these attributes for a flame-resistant feature (CPSC, 1976c). Again, these findings should be regarded solely as illustrations of the methodology because of the high degree of subjectivity in the factor level specification.

The question of whether different sets of moot factors would affect the relative factor importance weights of price and wet braking was examined by using analysis of variance. The criterion variable was the ratio of the price and wet braking factor weights. Form A or B, bike ownership, and sex were specified as the independent variables. As shown in Table 3, neither main- nor inter-action-effects were significant at any commonly used significance level, suggesting that moot factors can be added to make the task more realistic without affecting the stability of the relationship between the factors of interest.

Analysis of variance was also used to test the hypothesis that garment Form B subjects would have greater flame-resistant factor weights from Form A subjects because of more knowledge of clothing fire hazards (Table 4). Sex was included as an independent variable because it was thought the sex of the respondent could interact with the manipulated knowledge. The results show, however, that while the main-effects of knowledge and the inter-action-effects of sex and knowledge were not significant, sex as a main-effect was highly significant (p = .001). The average flame-resistant factor weight for females was 23 compared with an average weight of 9 for males. An ANOVA run for the wet braking factor weights revealed that sex had no significant effect for that safety feature. These findings seem to suggest that safety utility should be treated on a product and consumer specific basis.

CONCLUSIONS

Limitations

For both public policy and marketing strategy decision-making, conjoint analysis has several limitations. Foremost among these is the artificiality of the subjects' task. Respondents are not making market place decisions but rather indicating preferences for abstract descriptions of products which differ on a small number of attributes. While this procedure may be preferable to self-report measures of attribute saliency, the question remains whether conjoint analysis conclusions will be manifested in purchase behavior. Such a limitation is not overly troublesome from the standpoint of assessing the utility impact of a proposed safety regulation. It is, however, quite pertinent for the bicycle manufacturer who must estimate the sales response to the 10% price increase which will accompany improved braking performance.

Another limitation of the technique is the possibility that a product attribute may be imputed to possess utility solely as a consequence of its inclusion as a factor. A consumer may have never considered a garment's flame-resistant properties, yet may feel it must somehow be important since the researchers selected it for the task.

The potential for distorted conclusions from inaccurate factor level specification has been repeatedly emphasized. A second source of error which could cause inappropriate conclusions is the reliability of the scaling technique. The time frame of the study prohibited a check of test-retest reliability. However, the finding that garment Form A and B groups displayed no significant differences on any of the factor importance weights is weak evidence of reliability. Before conjoint analysis is used for conclusive research, reliability should receive additional attention.

Extensions

Throughout the report several extensions of this study's very basic application of conjoint measurement have been recommended. In addition, the technique has the potential for assisting policy makers in identifying the utility costs for specific products within a product category. The finding that a student's utility for a shirt or blouse would be substantially reduced by a flame-resistant feature is hardly surprising. But would the same conclusion be reached for pajamas and robes? Or would the utility impact be the same for elderly consumers where the probability of clothing fires are higher? With larger, more heterogeneous samples and multiple products within product categories, conjoint analysis has the capability for answering these questions. Depending on the utility costs, it could be that a clothing flammability standard should not apply to all types of apparel, but instead only to those where consumers are more willing to make the necessary utility tradeoffs.

Product safety regulation has been justified on the basis that consumers are unable to make tradeoffs between the benefits derived from using a product and the hazards this use may entail (Jones, 1973). By combining existing CPSC information with findings from conjoint analysis applications, this question can be empirically tested. From its national information system, the CPSC has product specific estimates of injury probabilities for various socio-economic classes of consumers. The degree to which consumers with different injury probabilities for a given product have concomitant factor importance weights for a safety feature designed to reduce these hazards may be an indication of whether consumers are capable of incorporating safety in their purchase decisions.

TABLE 1

MONANOVA AND SELF-EXPLICATED FACTOR IMPORTANCE WEIGHTS

TABLE 2

CORRELATION COEFFICIENTS: MONANOVA WEIGHTS VS. SELF-EXPLICATED WEIGHTS

TABLE 3

ANALYSIS OF VARIANCE-BIKES (DEPENDENT VARIABLE = RATIO OF PRICE AND BRAKING FACTOR IMPORTANCE WEIGHTS)

TABLE 4

ANALYSIS OF VARIANCE -- SHIRTS/BLOUSES (DEPENDENT VARIABLE = FLAME-RESISTANT FACTOR IMPORTANCE WEIGHTS

REFERENCES

Philippe Cattin and Dick R. Wittink, "Further Beyond Conjoint Measurement: Toward A Comparison of Methods," in William D. Perreault, Jr., (ed.), Advances in Consumer Research, Vol. IV, 1977, 41-45.

David Curry and William Rodgers, "Aggregating Responses in Additive Conjoint Measurement," in William D. Perreault, Jr., (ed.), Advances in Consumer Research, Vol. IV, 1977, 35-40.

Ralph L. Day, "Prescription for the Marketplace--Every-one Listen Better," Business Horizons, 19 (December 1976), 57-64.

Alfred L. Edwards, "Consumer Product Safety: Challenge for Business," University of Michigan Business Review, 27 (March 1975), 18-22.

Paul E. Green and Vithala R. Rao, "Conjoint Measurement for Quantifying Judgmental Data," Journal of Marketing Research, 8 (August 1971), 355-63.

Paul E. Green and Yoram Wind, Multiattribute Decisions in Marketing: A Measurement Approach, (Hinsdale: The Dryden Press, 1973).

Paul E. Green and Yoram Wind, "New Way to Measure Consumers' Judgements," Harvard Business Review, 53 (July-August 1975), 107-117.

Waiter Jensen, Jr., Edward M. Mazze, and Duke N. Stern, "The Consumer Product Safety Act: A Special Case in Consumerism," Journal of Marketing, 37 (October 1973), 68-71.

Mary Gardiner Jones, "Social Responsibility: The Regulator's View," California Management Review, 15 (Summer 1973), 78-84.

P. G. Kuehl and M. E. Simon, "The FDA Listens: A Survey of Consumer Opinions," FDA Consumer, (March i973), 8-13.

Lynn J. Loudenback and John W. Goebel, "Marketing in the Age of Strict Liability," Journal of Marketing, 38 (January 1974), 62-66.

Joseph C. Miller and A. Parasuraman, "Advising Consumers on Safer Product Use: The Information Role of the New Consumer Product Safety Commission," in Ronald C. Curhan, (ed.), American Marketing Association 1974 Combined Proceedings, 1975, 372-376.

Richard Staelin, "The Effects of Consumer Education on Consumer Product Safety Behavior," in William D. Perreault, Jr., (ed.), Advances in Consumer Research, Vol. IV, 1977, 380-387.

Richard Staelin and David Pittle, "Consumer Product Safety: A Discussion Paper," American Marketing Association Consumerism Workshop Monograph, (Chicago: A.M.A. 1977).

Richard Staelin and Alan G. Weinstein, "Correlates of Consumer Safety Behavior," in Scott Ward and Peter Wright, (eds.), Advances in Consumer Research, Vol. I, 1974, 87-100.

Jeffery L. Thompson, "Product Safety: Suggestions for Better Use and Purchase Behavior Through Consumer Education and Information," in Scott Ward and Peter Wright, (eds.), Advances in Consumer Research, Vol. I, 1974, 101-107.

Warren S. Torgerson, Theory and Methods of Scaling, (New York: John Wiley and Sons, 1958).

William L. Trombetta and Timothy L. Wilson, "Foreseeability of Misuse and Abnormal Use of Products by the Consumer," Journal of Marketing, 39 (July 1975), 48-55.

U.S. Consumer Product Safety Commission, Annual Report, Fiscal Year 1976, (Washington: U.S. Government Printing Office, 1976)a.

U.S. Consumer Product Safety Commission, "Product Profiles: Bicycles," October 1976, b.

U.S. Consumer Product Safety Commission, "Product Profiles: Wearing Apparel," October 1976, c.

Orville C. Walker, Jr. and Richard F. Sauter, "Consumer Preferences for Alternative Retail Credit Terms: A Concept Test of the Effects of Consumer Legislation," Journal of Marketing Research, 11 (February 1974), 70-8.

Orville C. Walker, Jr., Richard F. Sauter, and Nell M. Ford, "The Potential Secondary Effects of Consumer Legislation: A Conceptual Framework," Journal of Consumer Affairs, 8 (Summer 1974), 144-155.

William L. Wilkie and David M. Gardner, "The Role of Marketing Research in Public Policy Decision Making," Journal of Marketing, 38 (January 1974), 38-47.

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