Towards Determinants of Acceptable Risk: the Case of Product Risks

Arno J. Rethans, The Pennsylvania State University Gerald S. Albaum, University of Oregon
ABSTRACT - Discussions of product safety should distinguish between risk estimation and risk evaluation. Whereas the former is an essentially scientific and objective activity, the latter involves normative and relativistic judgments on risk acceptability. The acceptability of consumer product risks is explored by examining the relevance of a risk acceptability model emerging from the marketing, product liability and risk assessment literature. The model is found to be deserving of attention by the Consumer Product Safety Commission.
[ to cite ]:
Arno J. Rethans (1981) ,"Towards Determinants of Acceptable Risk: the Case of Product Risks", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 506-510.

Advances in Consumer Research Volume 8, 1981      Pages 506-510

TOWARDS DETERMINANTS OF ACCEPTABLE RISK: THE CASE OF PRODUCT RISKS

Arno J. Rethans, The Pennsylvania State University Gerald S. Albaum, University of Oregon

[The authors would like to acknowledge the intellectual and financial support provided by Dr. Paul Slovic and the Decision Research, Inc., Eugene, OR, in the conduct of this study.]

ABSTRACT -

Discussions of product safety should distinguish between risk estimation and risk evaluation. Whereas the former is an essentially scientific and objective activity, the latter involves normative and relativistic judgments on risk acceptability. The acceptability of consumer product risks is explored by examining the relevance of a risk acceptability model emerging from the marketing, product liability and risk assessment literature. The model is found to be deserving of attention by the Consumer Product Safety Commission.

INTRODUCTION

The assessment of risks associated with consumer products involves both risk estimation and risk evaluation. Risk estimation may be thought of as the identification of (product) hazards and the measurement of their probability and consequences (Slovic, Fischhoff and Lichtenstein 1977). Risk evaluation, on the other hand, is the complex process of anticipating the societal response to risk; this could be termed the "acceptability of risk" (Otway 1975; Rowe 1977).

This distinction between the two major dimensions of risk assessment becomes important in discussions of product safety. Indeed, recognition of the dimensions of measurement and evaluation allows us to clarify the term product safety. A product is deemed safe when its attendant risks are judged to be acceptable (Lowrance 1976). This conceptualization of safety differs sharply from the simplistic dictionary definitions which consider "safe" to mean "free of risk." A product can never be absolutely free of risk. Even if technically approachable, its accompanying price tag would be prohibitive.

More importantly, the distinction between the two facets of risk assessment provides a vehicle to characterize the activities of the Consumer Product Safety Commission (CPSC). The identification of products which pose risks to the American public constitutes risk estimation. This activity is essentially empirical in nature and is in part conducted through the maintenance of the National Electronic Injury Surveillance System (NEISS). Deciding whether the public might be or should be willing to bear the estimated risks constitutes risk evaluation.

The evaluation of risk, then, is more of a normative and political activity. It involves the judging of the acceptability of the product risks. Employment of the term "acceptability'' emphasizes the fact that risk management decisions are relativistic and judgmental. Indeed, the term elicits some interesting and crucial questions: "Acceptable to whom?", "Acceptable in what terms", and "What are the guides to acceptability?"

The study reported here was designed to begin to provide some empirical information on the basis of which some of the above questions may be answered. Its underlying premise is that public attitudes and values must be reflected in the Commission's decisions on whet constitutes acceptable levels of risk associated with consumer products.

CONCEPTUAL FRAMEWORK FOR THE STUDY

The notion of acceptability is quite pervasive, even though it is not always given explicit emphasis. A review of the appropriate literature in the areas of marketing, hazard assessment and product liability revealed several guides to Judge acceptability. In this section we briefly outline these guides and their origin.

Marketing

[In the area of marketing there has been a substantial amount of empirical research on how consumers perceive the risks associated with choice situations (Ross 1975). This research conceptualizes perceived risk as a dual component, multifaceted phenomenon including the dimension of physical risk. The literature does not however address the notion of acceptability of the risk. One possible exception may be Bettman (1973) when he distinguishes between inherent and handled risk.]

The CPSC is charged with the responsibility of protecting the public against unreasonable risk of injury. Unfortunately, the Product Safety Act left it to the Commission to decide what "unreasonable risk" is and how to reduce it. The National Commission on Product Safety gave the Commission some guidelines by adopting a statement by Curves D. Edwards:

"Risks of bodily harm to users are not unreasonable when consumers understand that risks exist, can appraise their probability and severity, know how to cope with them and voluntarily accept them to get benefits that could not be obtained in less risky ways. When there is risk of this character, consumers have reasonable opportunity to protect themselves; and public authorities should hesitate to substitute their value judgments about the desirability of the risk for those of the consumers who choose to incur it." (Final Report 1970)

This statement identifies a number of product risk characteristics which are believed to influence acceptability: knowledge of risk, ability to control the risk, and voluntariness of the risk.

Product Liability Litigation

Insight into the acceptability of risks may also be gained from an examination of product liability laws. A recent report by an Interagency Task Force of Product Liability (1978) noted that to determine whether a product is designed defectively (a plaintiff's argument), risks presented by the product must be weighed against the product's utility. Hence, a product which presents a substantial risk is not necessarily defective in the eyes of the law, since it may also have "great utility."

In balancing the risks and "utility" of the product, a list of factors has been proposed by Wade (1973) and has been adopted by courts in Oregon, Arizona, and Pennsylvania (Taskforce Final Report 1978). The factors to be considered are: usefulness and desirability of product, likelihood of injury, seriousness of injury, availability of substitutes, manufacturer's ability to eliminate unsafe character, user's ability to avoid danger and user's anticipated awareness of danger.

Risk Assessment Literature

The urgent need to help society cope with risks has produced a new discipline: Risk Assessment. When performing a risk assessment, the aim is to determine the magnitude and scope of a hazard and whether society should be exposed to it (Slovic and Fischhoff 1979). Researchers in this area have hypothesized and/or identified a number of variables which influence the acceptability of risks.

Chauncey Starr (1969) employing a method of revealed preferences found a number of variables to influence risk acceptability: voluntariness of risks, benefits accruing to society and the number of people involved in the risks. He concluded that the public is willing to accept "voluntary'' risks roughly 1,000 times greater than "involuntary" risks. Furthermore, it appeared that the acceptability of risk is roughly proportional to the third power of the benefits. Finally, his data indicated that the social acceptance of risk was directly influenced by public awareness of the benefits of an activity, as determined by advertising and usefulness, and the number of people participating.

Starr's initial work stimulated further work, notably by researchers at the International Institute for Applied Systems in Vienna by research associates at Decision Research, A Branch of Perceptronics, Inc. and the Clark University Center for Technology, Environment and Development. Much of the work by these researchers centers around the question of how people perceive and respond to risks. The assumption underlying this research is that individual and societal acceptance of risk, in general, is predicated upon how risks are perceived, not necessarily by actual levels or risk (Otway and Pahner 1976; Slovic, Fischhoff and Lichtenstein 1979). Indeed, it has been found that people's perceptions of risks are only moderately related to actual levels of risk and qualitative risk factors are now being examined as possible explanatory variables for perceptions of risk and risk acceptability.

A review of the work by these researchers as well as the work by Lowrance (1976) and Rowe (1977) yields a number of risk characteristics which are thought to be responsible for judgments of risk acceptability: voluntariness of risk, degree of control over risk, knowledge of risk, potential for misuse, degree of risk to special groups, the benefit/risk ratio, ease of hazard reduction, extent of societal exposure, severity of consequences, and necessity of exposure.

The Emerging Model

The literature reviewed above suggests that perceptions of risk and acceptability of risks are determined by both the quantitative and qualitative characteristics of those risks. Specifically, acceptability perceptions are hypothesized to be negatively related to such "objectively" determinable characteristics as injury frequency and injury severity. Furthermore, a set of qualitative risk characteristics are thought to be perception influences. This set of characteristics and their hypothesized relationships with acceptability perceptions include: voluntariness (+), risk knowledge (+), risk control (+), product necessity (+), foreseeability (+), exposure (+), ease of risk reduction (-), user error (-), and risk to children (-).

OBJECTIVES OF THE STUDY

The study reported in this paper represents a first attempt to explore consumer perceptions of the acceptability of the hazards associated with consumer products. The general objective of exploration translates into two research objectives: (a) determine consumers subjective perceptions of product risks and (b) examine the relevance of the emerging model of risk acceptance to the judgments of the acceptability of product risks.

METHODOLOGY OF THE STUDY

To obtain the information necessary to accomplish the stated objectives, participants in the study were asked to complete a questionnaire. Participants were asked to rate a set of consumer products on a series of seven point scales, which included societal hazard and acceptability scales as well as scales representing the risk characteristics identified in the literature review described earlier. The scales are typologically modeled after the perceived risk studies in marketing (Ross 1975) and contentwise after studies by Green and Brown (1978).

A pretest of the questionnaire suggested that the task needed to be shortened in order to secure respondent cooperation as well as to avoid respondents' fatigue and its accompanying errors. Therefore, it was decided to run a split-run technique on the risk characteristic scales. That is, two versions of the research instrument (A and B) were administered. Each version contained the societal hazard and acceptability scales and six of the twelve risk characteristics.

The products which served as stimuli in the study were included on the basis of a number of criteria. First, the set of products was to reflect the frequency range found in the CPSC data tabulation. Second, the products selected must, at the same time, reflect the severity range of the NEISS data. Third, the product definition must be unequivocal. Many of the "definitions" in the NEISS tabulations do not constitute a succinct product, but rather a product category. Finally, products to be included in the questionnaire were to be selected so as to maximize the probability of product familiarity among respondents.

Application of these criteria led to the identification of the products shown in Table 1. The categorization of the severity of injuries into low, moderate, and high follows that of the CPSC and is based on the mean severity values of the products. A mean severity of 17 or less generally indicates low severity; the range 18 to 81 usually implies moderate severity; and a mean severity over 81 implies high severity. The severity values for the products of this study are shown in parentheses in Table 1.

TABLE 1

PRODUCT STIMULI FOR ACCEPTANCE STUDY

Thirty five questionnaires of form A and forty questionnaires of form B were completed by members of three civic groups in a medium sized town on the West Coast. The questionnaires were hand-delivered to the groups and returned by mail in the case of one group and returned by a group representative in the case of the other two groups. In this way, seventy-five completed and usable questionnaires were collected. A donation in the amount of five dollars was made to the organizations' treasury of each completed questionnaire.

The overall sample consisted of 522 males and 48% females, their ages ranged from 20 to 52 years of age with a mean age of 30.5 and a modal age of 32. Sixty-five percent of the respondents were married and 24% of thee were single. With regard to education, about 50% of the respondents completed a university degree. An additional twenty-five percent noted that they had completed some university education, and 12% completed high school and/or vocational school or less. Occupationally, the groups may be characterized as predominantly white collar rather than professional. Only seven percent were blue collar workers.

FINDINGS OF THE STUDY

Risk Acceptability Ratings

Respondents were asked to judge whether or not risks associated with each of the twenty nine products were at a socially acceptable level. The mean scores of the products on the 7-point acceptability scale are shown in the first column of Table 2. This bipolar scale has been coded such that the higher the score, the greater the acceptability of the product risks. On an absolute level, the majority of the products are considered to bear risks which are perceived to be acceptable (i.e., greater than 4.0). The most acceptable risks are those associated with hammers, sewing machines, garage doors, refrigerators, and dehumidifiers. The lesser acceptable products are fireworks, skateboards, sunlamps, power lawn movers and cosmetics.

TABLE 2

PRODUCT RISK ACCEPTABILITY AND HAZARD RATINGS

The third column of Table 2 shows the participants' ranking of the hazardousness of the products included in the study. Comparing the hazardousness rankings with the acceptability rankings suggests the presence of an inverse relationship between risk and acceptability. Indeed, when computing a rank correlation between these two variables, we find a coefficient of -.62. Reviewing the major rank discrepancies suggests that risk acceptability is not only influenced by perceived risk, but by other considerations as well. Swimming pools, knives, and snow skis from cases in point. Each of these products were considered to be quite hazardous from a societal perspective, yet were considered to be quite acceptable. In the case of swimming pools and snow skis, this result my be explained by the high ratings of these products along the voluntariness and foreseeability characteristic. The acceptability of the risks associated with knives is probably due to its high perceived necessity and the high degree of "knowability" of the risk.

Risk Characteristics

Participants were also asked to rate each of the products on twelve, seven point scales, each of which represented a characteristic of risk. The mean ratings on these characteristics for the more hazardous and the less hazardous products are shown in Tables 3 and 4.

TABLE 3

RISK CHARACTERISTICS OF MORE HAZARDOUS PRODUCTS

TABLE 4

RISK CHARACTERISTICS OF LESS HAZARDOUS PRODUCTS

An examination of individual product profiles provides insight into consumers' risk and acceptability perceptions. For example, the ratings for fireworks suggest that consumers feel that risks associated with this product are perceived as rather voluntary, somewhat known to the user, and highly foreseeable. At the same time, there is a high perceived likelihood of user error accompanied by major injuries if accidents do occur. In addition, children are perceived to be rather vulnerable to the product risks. Similar profiles emerge for the other more hazardous products such as skateboards, portable power saws, and power lawn mowers.

In contrast, when examining dehumidifiers, the least hazardous product, a rather different profile is found. Although consumers admit to relatively low levels of knowledge and consider the risk somewhat unforeseeable, the perceived minor injuries associated with the product and the low perceived probability of user error seem to make this product relatively safe and acceptable.

Similar analyses may also be made at a group level. The data in the tables allow the reader to compare "group profiles" of more hazardous versus less hazardous products. The two groups differ substantially on the characteristics of risk control, injury severity, likelihood of user error, risk to children, frequency of injury and foreseeability.

Dependence Structure

The findings discussed in the preceding section suggest that a relationship exists between perceived and acceptable risk on the one hand, and the qualitative risk characteristics on the other, This section explores this relationship by means of a correlational analysis.

In Table 5, the product moment correlations between the mean risk characteristics scores and the mean risk and acceptability ratings are shown. The scores on the risk characteristics have been coded in a positive direction to facilitate interpretation of the data. The hazard and acceptability ratings correlated significantly (p < .01) with the characteristics of risk knowledge, risk control, injury severity, product necessity, user error, risk to children and injury frequency.

TABLE 5

CORRELATION BETWEEN RISK CHARACTERISTICS AND HAZARD/ACCEPTABILITY RATINGS

The majority of the observed correlations conform to the direction of the influence hypothesized in the risk assessment literature described earlier. For example, the higher the likelihood of error, the less acceptable the product is considered. Similarly, the higher the controllability of the product risk the more acceptable the product was.

Yet, not all correlations between perceived and acceptable risk and risk characteristics behaved according to expectations. The characteristic of foreseeability shows a significant negative correlation with acceptability. One would expect, however, that the more foreseeable a hazard, the more acceptable the risk. It appears that foresee-ability may have more than one dimension, and that the specific dimension that is operating is a function of the risk situation. For example, Ms. King, Chairperson of the CPSC, holds that the Commission's focus should be on risks which are not readily foreseeable by the public. The emphasis in this position seems to be on serious and unexpected hazards. Given this position, one would expect a positive correlation between foreseeability and acceptability, Respondents in this study, on the other hand, seem to regard foreseeable as "knowable", i.e., a sort of 'accidents are bound to happen and therefore the product is dangerous' attitude. In this latter perspective, the negative correlation seems reasonable.

A second characteristic that does not conform to expectation is the dimension of voluntariness. The more voluntarily assumed risks are presumed to be more acceptable. Although this relationship seems to be born out in the case of some products, e.g., snow skis, the correlation across all products is near zero. One explanation for this result is that the consumer goods studied here did not differ much along this dimension.

Finally, a comment is in order about the characteristic termed ease of risk reduction. The respondents were asked how easily hazards associated with a particular product could be reduced through product redesign. A priori one would expect a relationship that suggests the more easily reduced the risk, the less acceptable the risk. That hypothesized relationship was not observed.

Predictive Validity of Risk Characteristics

Earlier in the conceptual framework section a number of attempts to predict acceptable levels of risk were reviewed. In this section the approaches to these predictions are further generalized by an attempt to develop a formula which specifies acceptability of product risks as a function of the twelve risk characteristics under consideration in this study.

Table 6 shows the result of a stepwise regression analysis in which acceptability formed the dependent variable and the risk characteristics constituted the independent variables. Specifically, the table shows the entry sequence with the accompanying increases in r-squared as well as the resulting multiple correlation coefficients, Furthermore, in the column labeled "coefficient", are recorded the unstandardized regression coefficients for the variables in the final model.

TABLE 6

SUMMARY TABLE OF STEPWISE REGRESSION RISK ACCEPTABILITY

Although caution should always be exercised when interpreting the results of stepwise regressions, it should be pointed out that the variables entered in the model do have theoretical justifications in the risk assessment literature. That is to say, although the order in which variables enter a stepwise regression procedure is not always reflective of the relative importance of the variables, it is felt that the evidence presented seems to be converging on the variables included. The possible exception here is the voluntariness characteristic. Although no significant correlations were observed between this characteristic and hazard and acceptability ratings, it plays an important role in the regression model.

CONCLUSIONS AND RECOMMENDATIONS

In this paper we have introduced the distinction between risk estimation and risk evaluation in the determination of product safety. This distinction alerted us to the fact that product risk management decisions include objective and scientific components as well as mere normative, relativistic and judgmental components. This latter realization leads us to develop the premise that public attitudes and values should be reflected in the Commission's decisions as to what constitutes acceptable levels of risk associated with consumer products. Consumer research can play a role in the assessment of consumers' response to the product risks.

The results of the empirical study reported in this paper suggests the relevance of the risk acceptability model which emerged from a review of the marketing, product liability and risk assessment literature. The acceptability of product risks was found to be inversely related to perceived hazardousness. An analysis for the discrepant rankings, however, suggests that consumers may incorporate into their acceptability judgments dimensions other than perceived physical risk.

Several researchers in the area of risk assessment have noted the desirability of examining selected risk characteristics as possible explanatory variables for risk acceptability perceptions. The appropriateness of the characteristics for product risks assessment was examined by a stepwise regression analysis which indicated that about 84% of the variance in the acceptability judgments can be explained by the characteristics of product necessity, user error, voluntariness, and to a lesser extent product knowledge and foreseeability. From a decisional point of view it certainly must be concluded that these qualitative characteristics deserve specific consideration by the CPSC in their product hazard management decisions. User error in particular appears to be an important variable in both risk and acceptability perceptions. This finding may suggest that respondents would be receptive to educational campaigns regarding products rating high on this characteristic. Alternatively, the Commission may review the cost effectiveness of redesigning some of the products most prone to user error.

We further suggest that product "risk profiles" illustrated in Tables 3 and 4 be developed for products with high Hazard Index ratings. Studies of each of the risk characteristics for these products can provide considerable insight into individual and societal behavior toward risk acceptance.

REFERENCES

Bettman, James R. (1973), "Perceived Risk and Its Components: A Model and Empirical Test," Journal of Marketing Research, 10 (May), 184-90.

Fischhoff, Baruch, Slovic, Paul and Lichtenstein, Sarah (1979), "Which Risks are Acceptable," Environment, 21 (May), 17-38.

Green, C. H. and Brown, R. A. (1978), "Counting Lives," Journal of Occupation Accidents, 4 (May), 18-24.

Lowrance, William W. (1976), Of Acceptable Risk, Los Altos, CA: William Kaufman.

National Commission on Product Safety (1970), Final Report, Washington, DC: U.S. Government Printing Office.

Otway, Harry J. (1973), "Risk Assessment and Social Choices," in Proceedings of the IIASA Planning Conference on Energy Systems, Laxenburg, Austria: International Institute for Applied Systems.

Otway, Harry J. and Pahner, Philip (1976) "Risk Assessment," Futures, 8 (April), 122-34.

Rosa, Ivan (1974), "Perceived Risk and Consumer Behavior: A Critical Review," in Advances in Consumer Rematch, Vol. 2, M. J. Schlinger, ed., Urbane, IL: Association for Consumer Research.

Slovic, Paul, Fischhoff, Baruch and Lichtenstein, Sarah (1977), "Risk Assessment: Basic Issues," in Managing Technological Hazard: Research Needs and Opportunities, R. W. Kates, ed., Boulder: Institute of Behavioral Science.

Slovic, Paul, Fischhoff, Baruch and Lichtenstein, Sarah (1979), "Rating the Risks," Environment, 21 (April), 14-39.

Slovic, Paul and Fischhoff, Baruch (1979), "Now Safe is Safe Enough? Determinants of Perceived and Acceptable Risk," in The Management of Nuclear Waste, L. Gould and C. A. Walker, ed., New Haven: Yale University Press.

Starr, Chauncey (1969), "Social Benefit versus Technological Risk," Science, 165 (September), 1232-8.

U.S. Department of Commerce (1978), Interagency Task Force on Product Liability: Final Report, Springfield, VA: National Technical Information Service.

U.S. Consumer Product Safety Commission (1979), Annual Report, Part I and II, Washington, DC: U.S. Government Printing Office.

Wade, John W. (1973), "On the Nature of Strict Liability for Products," Mississippi Law Journal, 44, 825-37.

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