Correlates of Consumer Safety Behavior

Richard Staelin, Carnegie-Mellon University
Alan G. Weinstein, Carnegie-Mellon University
[ to cite ]:
Richard Staelin and Alan G. Weinstein (1974) ,"Correlates of Consumer Safety Behavior", in NA - Advances in Consumer Research Volume 01, eds. Scott Ward and Peter Wright, Ann Abor, MI : Association for Consumer Research, Pages: 87-100.

Advances in Consumer Research Volume 1, 1974    Pages 87-100

CORRELATES OF CONSUMER SAFETY BEHAVIOR

Richard Staelin, Carnegie-Mellon University

Alan G. Weinstein, Carnegie-Mellon University

[This research is being sponsored by the National Science Foundation under grant #GI-3277X.]

[Members of the group besides the authors include S. Feichtner, D. Pittle, and J. Thompson.]

INTRODUCTION

Almost every report on product safety is quick to point out that approximately 1 out of every 10 persons in the U.S. or 20 million people annually suffer a consumer product related injury requiring medical attention. With the technology explosion which has produced a myriad of new, sophisticated consumer products this accident rate could increase markedly. How can the frequency of consumer product related injuries be reduced? Three major approaches have been suggested. The first stresses the responsibility of industry to reduce accidents by improved product design and dissemination of safety related information. The second places the responsibility on government agencies to insure minimal safety standards by regulation, while the third assumes that consumers must watch out for themselves.

According to Richard Simpson (1972), Chairman of the Consumer Product Safety Commission, product safety "is a shared responsibility--shared by manufacturer, the government and the consumers." Simpson and the Commission view the major role of government and industry as providing safety information, education and persuasion to reduce accidents and injuries involving consumer products. However, the Commission also has forced the removal of products which they believe presented an extreme hazard to the consumer; developed product standards to reduce potential safety hazards; and encouraged manufacturers to develop and maintain its own standards through voluntary action.

One of the major proponents for safety regulation is Mary Gardiner Jones of the Federal Trade Commission who recently stated "..In my opinion informing consumers of possible hazards can never be regarded as sufficient where these hazards can involve loss of life and/or substantial property losses." (1973, p. 81). She reasoned that consumers cannot be expected to make the necessary trade-offs between the extra risks that use of the products may entail and the various advantages of cost, design etc. that these products are said to have. This orientation has led to recent legislation concerning safety of children's sleepwear which now must be made of flame retardant material and the removal of certain unsafe childrens' toys from the market place.

Although regulation may remove potentially hazardous products from the market, it also reduces some freedom of choice of the consumer who is willing to accept a known risk. Thus, it has been argued by Oi (1973) that the appropriate goal of public policy toward product safety should be the maximization of the economic welfare of consumers not the minimization of accident costs. He argues for policies which allow the consumer to select a product grade which minimizes the sum of the expected accident costs and the accident prevention costs. In another analysis Hinich and Staelin (1973) demonstrate that if consumers really want safer products, the market place will respond to meet the needs of the consumer, resulting in "voluntary" compliance by industry without any government intervention.

Each of the above approaches implicitly assumes a different model of consumer behavior. If consumers are capable of evaluating the possible dangers associated with consumer products and use this information in the purchase decision process, then the market place should operate to insure that the manufacturers will provide safe products. However, if some segments of the population cannot evaluate safe from unsafe products or do not use safety information in their decision process, then cheaper, unsafe products will be found in the market place increasing the probability of consumer product related injuries.

The above discussion has been primarily concerned with reducing injuries through the purchase of safe products. Yet injuries also may occur from interactions between consumers and "safe" products, suggesting that safe purchasing behavior alone is insufficient to prevent accidents. Figure 1 illustrates the four situations which can result from consumer-product interactions.

FIGURE 1

FOUR POSSIBLE SITUATIONS RESULTING FROM CONSUMER-PRODUCT INTERACTION

Good purchase behavior and regulation are aimed at eliminating unsafe products and thus address situations 2 and 4. However, consumer information and education programs are designed to influence situation 3 (involving product usage) as well as situations 2 and 4. There are no accurate estimates of the relative magnitudes of injuries which occur in each of these four situations but it has been estimated that at least 50 to 75 percent of all injuries are due to misuse of products by consumers (Simpson, 1972; Tokuhata, 1972).

Numerous studies have attempted to measure the frequency and severity of accidents associated with specific consumer products. One of the most comprehensive projects was conducted by the National Commission on Product Safety (1970) which estimated the number of annual accidents associated with a large number of consumer products. However, these data provided little empirical evidence about why these injuries occurred. More recently the Consumer Product Safety Commission reported estimates based on the National Electronic Surveillance System (NEISS) which measures the frequency and severity of consumer product related accidents treated in hospital emergency rooms. Although this system is designed to give accurate estimates of accident rates, no systematic method ties this reporting system into a method of determining the cause of injury.

In an attempt to relate accident and cause, Tokuhata (1972) studied the circumstances surrounding the accidents of 2,753 children treated in emergency rooms at six central Pennsylvania hospitals over a one year period. Over 75 percent of the accidents occurred in the home and involved common consumer products such as bicycles, electric irons, knives and furniture. No criteria for classification of accidents were given but Tokuhata concluded that 2 percent of the injuries were caused solely by defects in the consumer products, 18 percent were caused by the interaction of product with environment or consumer, and 40 percent were entirely caused by the misuse of the product. In 86 percent of the cases, the human factor contributed significantly to the injury. It is not clear from the report that better product design would have prevented these latter accidents, although based on studies of use behavior for specific products such as lawnmowers and power tools it is likely that a significant proportion of the accidents caused by product misuse were due to the human factor alone (National Commission on Product Safety, 1969). Whether this poor use was due to misunderstanding of the risks involved, ignorance of how the product worked, or because of a lapse of concentration is still an unanswered question.

Psychologists have long studied the problem of identifying predictors which are related to accident indexes. The concept of "accident proneness" (Farmer and Chambers, 1926) has been proposed but not consistently demonstrated to exist. One of the major methodological problems in studies of accident proneness is that reportable accidents occur infrequently and a researcher who wants to develop reliable predictors of accidents must wait several years before reliable data can be accumulated.

To get around this time problem some researchers have looked at near accidents or minor accidents. Guilford (1973) in a laboratory kitchen experiment observed the working patterns of 226 female subjects and classified particular actions into a number of minor accident categories. She correlated the observed poor use behavior (i.e., "accidents") with a number of demographic variables and the subjects' driving record. Significant intercorrelations were found between kitchen and automobile accident histories, thus supporting the hypothesis that accident incidence in one environment has a positive relationship to accident incidence in other settings. Demographic variables were not found to be associated with "accident proneness."

Another study involved an analysis of over 300 case histories of industrial accidents (Hirshfeld and Behan, 1963). The interpretation of these histories suggested that in a vast majority of cases, physical injury resulted from psychological stress preceding the accidents. Thus, the physical injury resulted from behavior attributable to the prior psychological problems of the worker.

On work more closely related to consumer product behavior Kuehl and Simon (1972) reported a survey of over 2,000 households interviewed to deter-mine their attitudes on product safety. In general, consumers said they had a strong safety concern for products such as foods, drugs, cosmetics and toys. Yet, only 30 percent of the respondents when purchasing toys reported considering safety "a lot" while 52 percent considered it "a little". Similarly only one-third said they were highly concerned about the safety of toys in particular. Thirteen percent felt that poor design was the single most important cause of accidents associated with toys, while 87 percent attributed the cause to misuse or failing to follow or read instructions. When asked how often they read directions, only 23 percent indicated they read them all or most of the time and 46 percent did not read directions any of the time.

The research cited above suggest that some people are more likely to have accidents than others and that the tendency towards injury may be related to psychological stress. Also, although safety is felt to be important, the consumer does not seem to pay much attention to this dimension while using and purchasing a consumer product. Finally, although regulation and/or better standards may reduce some of the accidents caused by defective products, it is clear that misuse of "safe" consumer products is still a major problem in consumer related accidents.

RESEARCH GOALS

The work reported in this paper is part of an ongoing project aimed at developing an educational consumer product safety program for high school students. The program is designed to increase a) students awareness of safety and b) their ability to select and use consumer products more wisely. This paper reports efforts to empirically determine the relationship between safety behavior (i.e., purchase and use behavior) and the level of knowledge of safety principles. Among the questions investigated are: Do persons with more knowledge about product safety behave safer? Do these consumers seek more safety information before purchase? What types and sources of information do they use?

SAMPLE

Three different samples are discussed in this study. The first, which is the primary sample used in this paper, is a probability sample of Pittsburgh residents. The sample was stratified by 1970 property values which were used as a surrogate for income. In all 328 households were selected and 187 completed personal interviews were obtained. Of the nonrespondents, 58 were due to refusals while 83 could not be contacted after a minimum of 3 call-backs (all during day-time hours). Thus of the 328 households selected, 25 percent could not be contacted, but, of those contacted the response rate was approximately 81 percent. Refusals and non-contacts were slightly more prevalent (but not significant) in the lower income intercity areas.

The second sample was obtained from the records of two Pittsburgh hospitals associated with the NEISS data collection system. Questionnaires were mailed to heads of 600 households, each household having a member of the family treated for an accident related to a consumer product. Of the 600 questionnaires mailed, only 115 were returned and found to be usable.

The last sample was a survey of 211 sophomores in three Pittsburgh high schools selected to obtain a cross-section of economic, ethnic, and racial backgrounds. A questionnaire was administered in 40 minute classroom periods. Non-response was limited to those students absent on the day the questionnaire was administered.

SURVEY INSTRUMENT

The interview-questionnaire was divided into 8 sections, and although there were slight differences for each sample, the instrument followed the same general format. The first two sections focused on general purchase information. For the hospital and the personally interviewed community samples, a set of five product categories were chosen to obtain specific information about the purchasing patterns and decision variables involved in the purchase process. The product classes used were small and large appliances, expensive and inexpensive toys and household tools. For the high school sample, products more relevant to a high school student were chosen. These data were collected using two methods. The first method asked respondents to name a product they purchased, if any, within each product class and list the information about the product which they considered important prior to purchase. The second method asked the respondents to rate eight decision variables using a Likert type scale with respect to the importance of the variable in the purchase decision process for the same set of consumer products. One of the decision variables was safety. The remainder of the survey instrument focused more specifically on safety.

The next two sections assessed the frequency of use and usefulness of information sources with respect to product safety. Section 5 consisted of six situationally-based questions involving risk. In some cases respondents were asked to indicate how much money they would be willing to pay for a safer product. In other cases, they were given the choice of several alternative behaviors to select in response to a risk related situation.

Section 6 consisted of a number of questions which asked how each respondent behaves in a given situation. The intent of this section was to measure the respondents reported safety use behavior. Typical situations included use of a three prong plug, pouring or handling volatile liquids, use of power mowers, selection of plastic toys and use of glass jars. These responses were scored with respect to safe use behavior on a scale ranging from-l-to 7.- Section 6 also assessed the degree to which the respondent felt capable of repairing and explaining to a neighbor the operation of common household products.

The next section focused on the respondents' level of knowledge of safety principles. These principles concerned electricity, insulation, materials, mechanical energy and flammability and were measured by a series of multiple choice questions.

The second to last section asked if anyone in the household had an accident related to a consumer product in the last two years and if so the circumstances surrounding the injury. The last section asked the respondent to provide some standard demographic data. On the average, the interview took 35 minutes. Both the use behavior and safety knowledge subsections were developed with the assistance of experts in product safety. Responses to these items were prescaled according to these experts' opinions of the appropriateness of the responses with respect to safety.

The survey instruments yield three different measures of safety behavior: a) reported purchasing behavior related to safety, b) reported safety use behavior, and c) injury history behavior. This paper will be concerned primarily with the first two behavior measures.

A major reason for concentrating on reported use behavior rather than accident behavior is that consumer product related accidents for a given individual are rare events and are a function of exposure to the hazard; this latter point being extremely difficult to determine. Thus, efforts were centered on a type of behavior which is easier to measure and duplicate. However apriori there is no guarantee that this measure is related to actual injury.

Two points are relevant concerning the distinction between injury history and reported safety use behavior. First even though injury history and reported safety use behavior are different measures, they are positively related (p < .05). This relationship can be seen graphically in Figure 2 which shows the percent of households reporting an accident as a function of the respondents' score on the use behavior test. The results indicate that respondents with better reported use behavior live in a household which tends to have fewer consumer product related injuries. The mean difference between the accident rate for the high scoring behavior group and two lowest groups is significant (p < .05).

In order to partially check on the possibility of response error between the respondents' reported behavior and actual behavior we asked the high school sample two questions for each item in the safety use section; what do YOU do and what should you do. Although there was a moderate degree of-correlation between the two answers (ranging from .3 to .6) the fact that responses were not highly correlated suggests that this set of respondents tended to report their actual behavior instead of the behavior they felt might be appropriate.

The above results indicate that a) the reported paper and pencil safety use behavior measure is a reasonable indicator of what the respondent actually does when confronted with the real life consumer product related situation, and b) the paper and pencil measure is associated with the harder to measure but ultimately more relevant variable, accident history. Finally, since accidents are rare events and can be caused by unsafe products as well as poor use of safe products, the paper and pencil measure may in fact be a reliable measure of the respondent's tendency towards consumer product related accidents (i.e.. accident proneness).

FIGURE 2

ACCIDENT HISTORY VS. REPORTED SAFETY USE SCORE

INDEX CONSTRUCTION

In designing the survey instrument, several items were developed for each construct of interest. In this way it was possible to combine the responses of correlated items so as to increase the reliability of the aggregated measure. In each case, the mean of the relevant aggregated questions was used as the construct measure, although correlation and principal components analyses were performed prior to collapsing the data to insure that the items included were measuring the same construct. The most difficulty encountered in constructing measures occurred for the safety knowledge questions. Originally it was believed that knowledge could be categorized according to standard educational disciplines, i.e., mechanical principles, electrical principles, etc. However, this hypothesis was not empirically supported. Instead the knowledge questions grouped as follows:

1) Questions which were product specific and related directly to safety. For example: when a soft plastic toy is stepped on it: a) breaks into small pieces which have sharp jagged edges, b) bends and crushes, c) etc.

2) Questions which were theoretical but which were concerned with safety. For example: Birds sitting on a high voltage wire are not electrocuted because: .

3) Questions which were theoretical and not specifically related to safety. For example: Efficiency of any machine or appliance is: .

A description of other indices used are found in later sections of this paper.

RESULTS

Information Seeking

A report on consumer education and product safety prepared for the National Commission on Product Safety (1970) states "Consumers in the United States would benefit in terms of lower injuries, reduced operating costs and few product breakdowns if better consumer education materials were provided with their purchases. !t The above assumes that consumers consider safety related information when making a purchase decision. This assumption is not completely substantiated in analyzing the responses of our sample of adult consumers. Table 1 shows the percent of respondents for particular product categories who mentioned safety with unaided recall when asked to list different types of information they considered prior to buying the product. Except for power tools and toys almost none of the respondents mentioned safety. This finding is consistent with the study conducted by Kuehl and Simon (1972).

TABLE 1

PERCENT OF RESPONDENTS WHO MENTIONED CONSIDERING SPECIFIC FACTORS WHEN PURCHASING CONSUMER PRODUCTS

The above can be contrasted with the results displayed in Table 2 which show the average rating of six dimensions (including safety) which are commonly considered prior to purchasing a consumer product. These ratings were obtained by giving each respondent a list and having him (her) rate the importance of each dimension. Using the average ratings across respondents as an indication of ranking, it can be seen that safety was felt to be the most important factor by consumers purchasing power tools and expensive toys while safety ranked third for inexpensive toys and fifth for both types of appliances. Why the difference between the structured and open ended responses? Perhaps, the respondents do not typically consider safety as a unique dimension in evaluating products but instead consider it implicitly within the dimensions of product durability, performance, or quality. Or perhaps they feel that most products are safe and thus,as consumers,they do not have to explicitly consider safety. These results clearly suggest that the way in which the question is asked determines the relative ranking of importance for safety. Consumers appear to be concerned about safety, particularly for toys and power tools. However the degree to which consumers use safety in their purchase decision process is still an unresolved issue.

TABLE 2

MEAN SCORE ON IMPORTANCE OF SPECIFIC FACTORS FOR PURCHASERS OF CONSUMER PRODUCTS

TABLE 3

MEAN SCORE ON USE OF SPECIFIC SOURCES OF SAFETY INFORMATION AND THE CORRESPONDING USEFULNESS

The interviewees were asked to rate a number of commonly used-sources of information on how often they used these sources and how useful they found these sources with respect to obtaining safety information. Average ratings for each source are found in Table 3. The most frequently used sources (excluding self or others in household) tend to be those which are either controlled by the manufacturers (i.e., advertising, manufacturer supplied information! or non-authoritative sources such as friends and neighbors. Government and consumer oriented publications were generally not used. These publications were also perceived not to be useful.

Previous work by Newman and Staelin (1973) and Cox (1967) suggests that information sources cluster into four independent categories which are as follows:

1) Technical sources: Government publications and Consumer Reports;

2) Industry supplied information: advertising, material accompanying product and laboratory testing results;

3) Personal sources: Friends and neighbors and sales personnel;

4) Household sources: Experience of buyer or others in household.

In order to test the independence of the above groupings, principal components analysis was performed on the data. Four eigenvalues were found to be greater than one and the factor loads for the eight variables produced the exact same groupings as hypothesized.

One of our goals was to determine the characteristics of the typical consumer using each of the four types of sources. We built indices (based on the average ranking of the items making up the group) to measure the consumer's reported use for each of the four types of sources. As expected these four indices were not highly related, the largest correlation being .28 (between household sources and industrial sources). The average absolute correlation between the four indices was .14.

In predicting the use of each type of source, seven variables were included in the analysis. They are as follows:

1) The consumer's overall score on reported safety use behavior i.e., how safe the consumer says he/she behaves when using specific consumer products.

2) The consumer's aversion to risk as measured by the aggregation of five situationally based questions involving risk.

3) The consumer's self reported ability to explain to a neighbor how 15 different consumer products work. This variable was measured by counting the affirmative responses and is intended to scale the respondents according to their ability to understand how products work.

4-6) The consumer's knowledge of a) practical safety principles, b) theoretical safety principles and c) theoretical principles not related to safety. These variables were measured by the responses to 23 multiple choice questions in the safety knowledge section.

7) The consumer's education level

Table 4 indicates the significant variables for each regression. As reported in previous studies (Katona and Mueller, 1953; Newman and Staelin, 1973) education level was found to be associated with use of the harder to understand technical materials. Interestingly, no other variables are significantly associated with this source. Use of industrial sources seems to be most related to those consumers who a) have better practical safety knowledge b) report safer use behavior and c) say they can explain how consumer products work, i.e., those consumers who seem to understand practical aspects of safety and who behave safely.

The regression equation for personal sources was not found to be statistically significant (with all the variables included) although there does exist a negative relationship between better use behavior and the use of personal sources. These findings are in agreement with past theoretical and empirical work 'Cox, 1967; Newman and Staelin, 1973) which conclude that persons with less experience and/or ability tend to seek out and use those sources which have less evaluative content but which are easier for the user to evaluate as to the reliability of the source.

TABLE 4

BETA WEIGHTS AND T STATISTICS FOR MODELS PREDICTING USE OF FOUR TYPES OF SOURCES WITH RESPECT TO SAFETY INFORMATION

Finally, household sources were used more by those respondents who showed a) better knowledge of practical and theoretical safety knowledge, b) felt they could explain how products in general work and c) had poorer theoretical general knowledge.

The above findings indicate that the use of all the sources except technical sources is related to either knowledge of safety principles and/or good safety use behavior. This raises the question of whether reported safety behavior is at all associated with better knowledge of safety principles. If so, we would expect that consumer education and information programs have some reasonable chance of affecting use behavior patterns. However if no relationship exists we must assume either that our measurement and/or analysis techniques are incorrect or that consumer education and information is of little use.

Table 5 indicates the results of a regression analysis with safety use behavior as the dependent variable. Reported ability to explain how a product works, safety knowledge, and risk aversion [It can be argued that the risk measure used in this study and the safety use measure are conceptually similar. However the risk measure was developed to assess as a general psychological construct whereas the use measure was based upon specific situations concerned with consumer behavior.] were positively associated with better use behavior, while general theoretical knowledge and education level were negatively associated with better use behavior. Thus, it seems that safe use behavior is related to specific knowledge about safety, but not with general theory. It is not clear whether consumer safety information can create better consumer safety behavior or whether experience with consumer products increases knowledge related to consumer safety. The authors are presently conducting a controlled experiment on the causal relationship between consumer information and safety behavior.

In an attempt to replicate the above model of safety use behavior the same analysis was conducted for a group of 736 high school sophomores. In general the signs of the beta weights were in agreement, the only reversal occurring for theoretical safety knowledge. However the mean levels on the knowledge variables for this younger sample, after correcting for guessing, were not significantly different from zero indicating that the measures of the students' knowledge of safety seemed to be random. Also, the t-statistics for the knowledge measures were not statistically significant for the high school sample.

TABLE 5

BETA WEIGHTS FOR FACTORS ASSOCIATED WITH REPORTED SAFETY USE BEHAVIOR FOR PROBABILITY

TABLE 6

REPORTED INCIDENCE OF ACCIDENTS WITHIN A TWO YEAR PERIOD INVOLVING A CONSUMER RRODUCT

One of the goals of this research was to study households with recent accident histories. The NEISS system provided easy access to consumers who had been treated for a consumer product related injury. Unfortunately, the response rate for the specially obtained hospital sample was low, thus the results may be biased due to non-response error.

A comparison of the accident histories for the probability and hospital samples is shown in Table 6. Each respondent was asked "Has anyone in your household had an accident related to a consumer product in the last two years?" Approximately 16 percent of the probability sample and 28 percent of the hospital sample answered yes. The hospital sample results are very surprising since according to the NEISS coding, every household should have responded affirmatively. It would seem that the respondents did not make the connection (as did the NEISS coder) between the injury and a consumer product. Perhaps some of the respondents were not aware of the injury or the cause of the injury since it may have occurred to another member of the household. However, the households selected all had injuries within the last three months. Perhaps some of the respondents interpreted the words "related to" to be "caused by", thus reducing the number of yeses. In any case, the 28 percent versus 100 percent response raises a question concerning the validity of either our measure or that used by the NEISS system, since both supposedly measure the frequency of consumer product related injuries.

When asked who or what was responsible for the injury, most respondents felt that the consumer was responsible. These results are in agreement with Tokuhata (1972) and provide strong evidence that consumers feel most accidents are caused by poor usage rather than unsafe products.

Finally, about 70 percent of the respondents reporting an injury said that the injured person was aware of the hazard before the accident. On the other hand 30 percent said they were not aware of the hazard. Consumer safety information might be most valuable if directed to this uninformed segment of users. Perhaps knowledge of the risks involved with the product would reduce injuries, since risk adverse consumers report that they behave safer and have fewer accidents in their households. The fact that most of the injuries occurred to consumers who were aware of the hazard, however, suggest that better consumer education and information might have only marginal value in reducing accidents unless the level of the consumer's understanding of safety principles can be increased.

SUMMARY

Three different approaches to consumer safety have been discussed. In order to evaluate the adequacy of each of these approaches it is necessary to understand how the consumer behaves when purchasing and using consumer products.

Persons with more knowledge about safety principles report using all sources of safety information, except technical sources, more often than those with less knowledge. However, the number of respondents who explicitly mentioned that they considered safety when purchasing a durable consumer good was low, ranging from 1 percent for small appliances to 21 percent for power tools. In contrast, consumers rank safety high with respect to price and quality as factors which are important in the decision making process. Thus, consumers appear to be concerned with safety, and persons with better safety knowledge tend to use more sources to obtain safety information.

Safe use behavior, as measured in this paper, is positively related to specific and theoretical knowledge of safety principles and risk aversion. General theoretical knowledge about principles of mechanics, physics, etc. was negatively related to safety behavior. The vast majority of consumers reported that the injured person was responsible for the accident, although approximately 30 percent of the time the injured person was unaware of the hazards associated with using the product.

The above findings indicate that consumer education programs should reduce injuries if they a) increase the consumer's level of safety knowledge and b) provide more awareness of potential hazards connected with consumer products. Also, the results tend to confirm the fact that most accidents are the result of poor use behavior. Some of the accidents associated with poor use may have been avoided by the product being designed so as to preclude the consumer from misusing the product or by banning the use of the product altogether. However, a substantial number of injuries were caused by a "safe" product being misused by the consumer. Consumer education may have a positive effect in these situations.

REFERENCES

Cox, D. F. The sorting rule model of consumer product evaluation process. Risk taking and information handling in consumer behavior. Edited by D. F. Cox. Cambridge: Harvard University Press. 1967.

Farmer, E. & Chambers, E. G. A psychological study of individual differences in accident rates. (Industrial Health Research Board Rep. No. 38) London: Her Majesty's Statistical Office, 1926.

Guilford, Joan. Prediction of accidents in a standardized home environment. Journal of Applied Psychology, 1973, 57, number 3, 306-313.

Hinich, M. & Staelin, R. Analysis of inspection procedures for food industry: a final report to FDA. Carnegie-Mellon University, Graduate School of Industrial Administration. Working Paper, 1973.

Hirschfeld, A. H. & Behan, R. C. The accident process. The Journal of the American Medical Association, 1963, 186, 193-199.

Jones, Mary Gardiner. Social responsibility: the regulator's view. California Management Review, 1973, 15, number 4, 78-84.

Katona, G. & Mueller, E. The study of purchase decisions. The dynamics of consumer reaction, 1 of Consumer Behavior. Edited by L. Clark. New York: New York University Press, 1954.

Kuehl, P. G. & Simon M. E. The FDA listens: a survey of consumer opinion. DA Consumer, 8-13, 1973.

National Commission on Product Safety. Final report. Washington, D.C.: U.S. Government Printing Office, 1970.

Newman, J. & Staelin, R. Information sources of durable goods. Journal of Advertising Research, 1973, 13 number 2, 19-30.

Oi, Walter Y. The economics of product safety. The Bell Journal of Economics and Management Science, 1973, 4, number 1, 3-28.

Simpson, Richard. U.S. Department of Commerce release of a speech to the Association of Home Appliance Manufacturers, September 15, 1972.

Tokuhata, George. Childhood injuries caused by consumer products. Pennsylvania Department of Health, Division of Research and Biostatistics, May 1972.

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