An Exploration of the Effects of Perceived Social and Performance Risk on Consumer Information Acquisition

Citation:

Richard J. Lutz and Patrick J. Reilly (1974) ,"An Exploration of the Effects of Perceived Social and Performance Risk on Consumer Information Acquisition", in NA - Advances in Consumer Research Volume 01, eds. Scott Ward and Peter Wright, Ann Abor, MI : Association for Consumer Research, Pages: 393-405.

Advances in Consumer Research Volume 1, 1974    Pages 393-405

AN EXPLORATION OF THE EFFECTS OF PERCEIVED SOCIAL AND PERFORMANCE RISK ON CONSUMER INFORMATION ACQUISITION

Richard J. Lutz, University of California, Los Angeles

Patrick J. Reilly, American Can Company

[Richard J. Lutz is Assistant Professor of Marketing at the University of California, Los Angeles. Patrick J. Reilly is a sales representative for the American Can Company, Peoria, Illinois. This research was undertaken while both authors were at Illinois State University, Normal.]

According to the theory of perceived risk (Bauer, 1960; Cox, 1967), consumers, when faced with a purchase situation in which risk is involved, attempt to reduce that risk through one of several alternative strategies. Perhaps the most common risk-reducing strategy employed is to become loyal to a particular brand (Cunningham, 1967; Roselius, 1971).

Since risk is generally conceptualized as a multiplicative combination of two components--uncertainty and importance or danger (Cunningham, 1967)--it seems reasonable that any risk reduction strategy should be operating on one of these two components underlying risk. For instance, brand loyalty can be seen as reducing uncertainty in the consumer's mind as to how the product will perform. However, it is less likely that becoming brand loyal causes any decrease in the importance of the product to the consumer.

Consideration of a range of risk-reducing strategies, most of which appear in an article by Roselius (1971), reveals that virtually all of the strategies employed by consumers to reduce perceived risk appear to operate primarily on the uncertainty component of risk. This is hardly surprising, since any strategy which operated to minimize the importance component would be some sort of defense mechanism (i.e., the consumer convinces himself the decision is not important). For the most part, then, it seems that the "rational" consumer will seek to reduce risk by reducing the uncertainty in the purchase situation.

While brand loyalty and other decision heuristics (e.g., "buy the most expensive," "buy an advertised brand") may be useful in certain instances, perhaps the most desirable strategy for reducing uncertainty in a purchase situation involving choice from among a number of brands is to obtain information about the alternative brands from one or more sources. While cues like price or brand image can be seen as forms of product information, other types of information can be obtained through a more thorough search of the environment. Howard and Sheth (1969) have labeled this process of obtaining product information "overt search." while Hansen (1972) refers to it as "information acquisition." At any rate, what is implied is that the consumer is actively engaged in seeking information regarding the product class. The result of the integration of this information into the consumer's cognitive structure should be the reduction of uncertainty surrounding the decision, which in turn should lead to a reduction in perceived risk.

Although a variety of sources of product information are readily available to the consumer, most past research has focused on word-of-mouth (Arndt, 1967; Cunningham, 1967; Sheth and Venkatesan, 196&). In general, the findings of these studies have shown a rather weak positive relationship between perceived risk and tendency to seek word-of-mouth information regarding the product. There is some indication, therefore, that consumers engage in information search in order to reduce perceived risk. As pointed out by Cox (1967), the consumer in a "risky" situation may have a variety ow information needs to satisfy; i.e. the type of perceived risk being faced by the consumer may dictate his information needs. Jacoby and Kaplan (1972) have recently identified at least five different kinds of perceived risk which may exist in a purchase situation. Presumably, the consumer tends to select those sources which will most likely provide him with uncertainty-reducing information for the type of risk he perceives to be greatest in that situation.

Most past research on risk reduction through information search has employed only a general measure of perceived risk, without attempting to identify differing information needs based on different risk dimensions. Recently, however, two studies have related tendency to engage in word-of-mouth communication with perceived social risk (Perry and Hamm, 1969; Roselius, 1971). Both studies report a relatively strong tendency to seek word-of-mouth information when perceived social risk is high. However, neither study explicitly considered other sources of information which may be expected to become salient when a different type of perceived risk is dominant. Thus there appears to be a need for a more systematic investigation of the types of information sources utilized by the consumer in his efforts to reduce one or more kinds of risk.

Andreasen (1968) has developed an information source typology which provides a useful starting point for investigating search strategies. Briefly, he outlined five different "channels" of information flow:

1. Impersonal Advocate--e.g., mass media advertising

2. Impersonal Independent--e.g., Consumer Reports

3. Personal Advocate--e.g., sales clerks

4. Personal Independent--e.g., friends' opinions (w-o-m)

5. Direct Observation/Experience

These five types of sources have different characteristics which make them differentially useful to the consumer depending upon his particular information need, which is related to the type of risk the consumer perceives in the situation.

The present research investigated two types of risk--performance risk and social risk. Performance risk is defined as the extent to which the consumer thinks that the various brands of a product perform differently in ways that are important to him. Social risk is defined as the extent to which the consumer thinks that other people judge him on the basis of the brand he uses (Brody and Cunningham, 1968). Newton (1967) has speculated that where the consumer perceives a high degree of performance risk, he will engage in "problem-solving" behavior and will prefer information from a technically competent source; however, under a high degree of social risk, the consumer needs social reassurance and will tend to engage in "approval seeking" behavior, preferring reassurance from a personally acceptable source. Obviously, the consumer can *ace both social and performance risk (as well as three to four other types of risk) in any purchase situation. The general proposition offered by Newton (1967) is that the consumer is more likely to respond to an appeal which reduces risk on the dimension of perceived risk which is dominant.

Newton focused his analysis on the customer-salesman dyad, attempting to specify which type of sales approach would be more successful under various combinations of performance and social risk. Extending his treatment to a more general view of information acquisition, it can be hypothesized which types of information sources w$11 be preferred in different purchase situations. For instance it would be expected that when social risk is high, the consumer would have a relatively high tendency to seek information from either Personal Advocate (PA) or Personal Independent (PI) sources, because his need for social reassurance should be relatively greater. On the other hand, a consumer facing performance risk might be expected to turn to Impersonal Independent (II) or Impersonal Advocate (IA) sources.

The following exploratory hypotheses guided the current research:

1. The tendency to engage in information search is directly related to the degree of social and performance risk in the purchase situation; i.e., regardless of which type of risk is dominant, increased risk leads to increased search.

2. Under conditions of relatively high perceived social risk, the consumer tends to utilize personal sources of information (PA and PI). Of these two types of sources, PI (word-of-mouth) should be dominant.

3. Under conditions of relatively high perceived performance risk, the consumer tends to utilize impersonal sources of information (IA and II). Of these two types of sources, II should be dominant.

4. In general, the consumer will tend to rely on personal experience (CE), whenever possible, to obtain product information.

METHOD

Recently Bettman (1972, 1973) has made a distinction between inherent risk and handled risk in a purchase situation. Inherent risk is defined as the latent risk the product class holds for the consumer, while handled risk is the amount of risk the product class produces when the consumer selects a brand in his usual buying situation. Inherent risk is seen as the initial degree of risk in a purchase situation, and handled risk is the residual risk facing the consumer after engaging in some form of risk reduction. As Bettman pointed out, most past research has paid little attention to the exact nature of the risk being investigated, which is reflected in a wide diversity of measurement procedures. In the present research, an attempt was made to investigate inherent risk only. By limiting the research in this manner, other forms of risk reduction (e.g., brand loyalty) were eliminated as possible confounding factors.

Bettman (1972, 1973) has focused his research on overall perceived risk. Therefore, it was necessary to develop measures which would assess both inherent performance risk and inherent social risk. Following Bettman (1972), an instrument was developed in which the concept of social (performance) risk was first described. Secondly, a hypothetical situation was presented in which all brands of the product class are unidentified and the consumer must make a choice at random. Finally, several stimulus product classes were listed for which a rating on social (performance) risk was requested. It should be noted that Bettman used a paired comparison method for obtaining risk ratings, while the present research relied on the use of a nine-point equal-appearing interval scale, ranging from "No Risk" (1) to "A Great Deal of Risk" (9).

After developing the risk measures, pretest samples of 80 and 75 undergraduate business students at the University of Illinois at Urbana-Champaign responded to the social risk measure and the performance risk measure, respectively. Each questionnaire included a total of 60 products, which were selected from a number of previous studies of perceived risk. The purpose of the pretest was twofold:

1. To provide a test of the measurement instrument.

2. To identity products of varying degrees of social and performance risk, which were needed for the major part of the study.

Based on the results obtained from the two samples, the pretest was judged to be successful. Informal interviews conducted with a number of respondents revealed no problems of understanding what the instrument was designed to measure. While there was a slight tendency for performance risk to be rated higher than social risk (the overall mean ratings were 4.81 and 3.75, respectively) little difficulty was encountered in identifying nine products for use in the subsequent investigation of information search strategies.

In order to provide a test of the exploratory research hypotheses presented earlier, a role-playing experiment was designed. While the limitations of role-playing are recognized, it was felt that such a study would provide useful input to further research conducted under more realistic cognitions. It should be emphasized that the current research is primarily exploratory in nature and is not designed to provide broad generalizations regarding the relationship between perceived risk and information search. Hopefully the results of this research will suggest meaningful hypotheses for subsequent studies of the processes by which consumers acquire information.

A 3 x 3 factorial design was employed in this study. The first factor consisted of three levels--high, moderate, low--of perceived social risk, while the second factor represented three levels of perceived performance risk. Table 1 shows the products selected from the pretest which corresponded to each of the nine cells. These products were selected on the basis of the pretest ratings of social and performance risk. The entries in Table 1 represent the risk ratings obtained from the experimental subjects as a manipulation check. In general, there was little deviation from the pattern of ratings in the pretest. The one exception of note was the performance risk rating of bread, which inexplicably was much higher for the experimental subjects than for the pretest respondents.

TABLE 1

MEANS OF EXPERIMENTAL CONDITIONS FOR SOCIAL AND PERFORMANCE RISK

The experimental treatments consisted of hypothetical purchase situations in which the subject was to place himself--i.e., role-play. The nine different treatments were identical except for the product to be "purchased." The following paragraph defined the hypothetical situation:

"You need to buy (product) for your own personal use, but when you go shopping you discover that all of the brands you are familiar with are unavailable. The only brands available in the entire town are Brands A, B, C, D and E, brands which you know nothing about. Nevertheless, you need the product and therefore must make a choice among the five brands. However, for you to select a brand without any information about the brands would be virtually the same as selecting at random."

Thus the situation description conformed to Bettman's (1972) concept of inherent risk. The fact that the brands were au unknown and selection would be virtually at random should have generated the maximum amount of conflict in the consumer for that particular product class.

Following the situation description, the subject completed a 20-item behavioral differential (Triandis, 1964), which is shown in Appendix A. The 20 behaviors were selected to correspond to the fivefold typology suggested by Andreasen (1968), as well as one behavior ("go ahead and pick a brand") which allowed the subject to extricate himself from the situation without consulting outside sources of information. The codes to the right of the items in Appendix A did not appear on the actual questionnaire, but are used to indicate which items corresponded to the various types of information sources in the Andreasen model.

Subjects were 243 undergraduate business students at the University of Illinois at Urbana-Champaign and Illinois State University, Normal. Each subject responded to three of the nine experimental treatments, which were counterbalanced to control for order effects and level of experimental treatments. Using a Latin square rotation, each subject was exposed to each of the three levels of both types of risk. Following the completion of the three role-playing situations, each subject rated the amount of social and performance risk in each situation, using scales similar to the ones employed in the pretest. These ratings were obtained to provide a check on the experimental manipulations.

RESULTS

The first step in the analysis was to conduct 3 x 3 factorial ANOVA on both the perceived social risk and perceived performance risk ratings obtained from the subjects. Due to disproportional cell sizes, the method of unweighted means analysis was used. Results of these analyses are summarized in Tables 2 and 3.

TABLE 2

MANIPULATION CHECK ON PERFORMANCE RISK

Turning first to Table 2, the expected effect was a main effect for Factor B--performance risk. This would indicate that the products selected were perceived as possessing different degrees of performance risk, as required by the design. The expected effect did obtain (p < .001), explaining approximately 29 percent of the total variance in the performance risk ratings. However. both a main effect for Factor A--social risk--and the A x B interaction term were also significant (p < .001). These unexpected effects, while significant, were relatively weak relationships, failing to account for more than 10 percent of the total variance. In addition, the significance of the interaction term can be traced almost entirely to one product--bread--which inexplicably had a mean performance risk rating of 7.31 (see Table 1), even though its pretest rating mean was only 4.58. The most plausible explanation for the wide divergence of ratings for bread is that the instrument was not providing a reliable indicant of the underlying construct.- Subsequent analyses showed that information search behavior with respect to bread corresponded closely with other products low or moderate in performance risk, so the apparent deviation of the performance risk ratings from the position dictated by the design was not considered to be a serious one. However, care must be exercised in the interpretation of further analyses, due to possible confounding resulting from the interaction term.

TABLE 3

MANIPULATION CHECK ON SOCIAL RISK

Table 3 shows the results of the analysis for the social risk manipulation. As expected, a significant main effect was obtained, although the other main effect and the interaction term were once again significant, but trivial. The interaction in this case was not as serious as in the performance risk manipulation, and findings pertinent to social risk can be interpreted with little danger of confounding.

Examination of the means of the experimental conditions for performance and social risk (see Table 1) shows that the significant effects obtained for social and performance risk were in the expected directions; i.e., high risk was greater what moderate risk, which in turn was greater than low risk. It should also be noted that performance risk ratings, in general, were higher than social risk ratings, the grand means being 5.70 and 3.80, respectively. This may indicate a general tendency on the part of the subjects to be more concerned with the performance aspects of products; alternatively it may indicate either a predisposition not to report the effects of others on their purchase decisions, or a lack of insight into these influences. At any rate, for the current investigation, conclusions regarding "high" social risk must be limited to products "relatively high" in social risk, as compared to the ratings of other products on social risk. In only one cell--Beer--did the absolute level of the social risk rating exceed the performance risk rating. This general pattern of findings may be expected to have some impact on the tests of the research hypotheses.

In order to analyze the influence of perceived social and performance risk on information search behavior, it was first necessary to generate scores for each of the five types of sources under consideration. Subjects' responses to the 20-item behavioral differential shown in Appendix A were coded from 1 (would not) to 7 (would). after which those items which referred to a particular source type (shown to the right of the items in Appendix A) were summed to yield individual scores for each source. For comparison purposes, each score was then divided by the number of items comprising the raw total, yielding a set of final scores on each of the five sources which had potential ranges from 1 (low utilization of source) to 7 (high utilization of source). In addition to the information search behaviors, one behavior--"Go ahead and pick a brand"--was included as an alternative to searching for information. Means on each of these variables bs cell are shown in Table 4.

TABLE 4

MEAN VALUES OF BEHAVIORAL DIFFERENTIAL RATINGS OF ALTERNATIVE TYPES OF INFORMATION SOURCES--INDIVIDUAL CELLS

Several interesting relationships appear in Table 4. First of all, the six products which were either low or moderate in performance risk displayed rankings of the information sources which were remarkably similar. In all cases, subjects preferred-to take their--chances and pick-a brand without searching for product information. In contrast, what would appear to be the most reliable sources of information--Impersonal Independent sources--were least preferred. The summary of these six cells shown in Table 5 illustrates this relationship quite clearly. In addition, Table 5 shows that for the low and moderate risk products, Personal Independent sources were the most preferred information sources, while mass media advertising (IA) and Personal Advocate sources were not heavily utilized. [It should be noted that this particular pattern of results, as well as many of the other findings of this research may be a function of the nature of the sample. Students may be expected to have media habits quite different from the population as a whole.]

TABLE 5

MEAN VALUES OF BEHAVIORAL DIFFERENTIAL RATINGS OF ALTERNATIVE TYPES OF INFORMATION SOURCES--POOLED

However, for products high in performance risk, a different pattern emerged. Direct Observation and Experience was the single most preferred source, while the risky behavior of purchasing without information search was generally shunned. Also, Impersonal Independent sources were quite important, as well as Personal Independent (word-of-mouth) sources.

The overall means shown at the bottom of Table 5 indicate that subjects generally tended to rely on their own judgment, either purchasing without information search, or basing their decisions primarily on their own observations. Of the four sources external to the consumer, word-of-mouth (PI) was most preferred. This agrees with Arndt's (1967) conclusion that word-of-mouth is the most important source of risk-reducing information. If the findings of this research prove to be generalizable, there would be strong implications for publications like Consumer Reports. Apparently subjects in this study felt little need to consult an Impersonal Independent source unless performance risk was relatively high. This would suggest that consumers may turn to Consumer Reports for information on only a very limited number of high risk products; identification of these products could prevent unnecessary reports on low risk product categories. [It might be argued that subjects realized that Consumer RePorts typically analyzes only durable products, which would explain their intentions not to use it for nondurables. However, recent research by Lutz (1973) seems to indicate that most consumers have little knowledge of the kinds of products discussed in that publication.]

One final interesting finding was that mass media advertising (IA) was generally not utilized heavily, particularly for low and moderate performance risk products. This result tends to correspond with Bucklin's (1965) finding that, while advertising was not generally relied upon as a source of product information, there was a tendency to use it somewhat more heavily when more expensive or unfamiliar items were purchased.

In order to provide a more precise examination of the relationships between the two types of risk and information search behavior, each of the six variables in Table 5 was subjected to a 3 x 3 factorial ANOVA. Results of these analyses are summarized in Table 6. Similar to the ANOVA performed on the performance and social risk variables, the method of unweighted means analysis was used, since the cell sizes were disproportional. Table 6 shows three significant effects for each of the six variables; however, only one of these effects consistently explained a substantial portion of the variance.

TABLE 6

SUMMARY OF ANOVA RESULTS

Turning first to the use of mass media (IA) sources, it can be seen from Tables 4 and 6 that only for products high in performance risk did subjects turn to these types of sources. For the other six products, the tendency was to respond that they "would not" search for information from IA sources. A major portion of the A x B interaction term seems to be traceable to two cells (A2B1 and A2B2), which were represented by Instant Coffee and Bread, respectively. The particular pattern of the interaction would not seem to be generalizable to all products possessing the same degrees of social and performance risk, but may possibly be a result of different levels of advertising for the two product classes. In addition to being a very weak relationship, therefore, it was probably also idiosyncratic to the two products selected for use in this study.

The pattern of results for Personal Advocate sources was virtually identical to the IA findings. The effects of social risk appear to be minimal: although it had been expected that subjects in high social risk conditions would have relied somewhat more heavily on information from PA sources.

The analysis of Impersonal Independent sources shows exactly the same pattern of results as the two preceding analyses. However, in this case, the significant main effect for performance risk accounted for 40 percent of the variance in II scores. It appears that the use of sources like Consumer Reports was very selective--when high performance risk was faced, the II sources were relied upon heavily; when performance risk was low or moderate, II sources were ignored. Once again, social risk had virtually no impact on the usage level of II sources of information.

Taken together, the results based on IA, PA, and II sources would seem to support Cox's (1967) contention that the consumer uses various channels or sources of information in a complementary fashion, rather than treating them as competing sources of information. Similarly, Cunningham (1967) found that consumers in high risk situations tended to utilize more sources of product i-formation, with little regard to the nature or trustworthiness of that information.

As shown by the results of the analysis of Personal Independent sources of information (word-of-mouth), there is general acceptability of information obtained from personAl sources. However, rather unexpectedly, the main effect for performance risk explained more variance than did the effect for social risk. Apparently word-of-mouth information was more valuable in reducing performance risk than it was for reducing social risk. It had been expected that PI sources would enjoy m*Y;mum usage when consumers were facing high degrees of social risk. It should be recalled, however, that none of the social risk ratings in Table 3 actually attained the same high absolute levels as did the high performance risk ratings. The possibility remains that products very high in social risk may encourage consumers to use PI sources to an even greater degree.

In summarizing the results for Direct Observation/Experience, the general pattern of findings closely approximates the findings for IA, II, and PA sources. The main effect on performance risk explained 41 percent of the variance in OE ratings, showing a strong tendency on the part of the subjects to examine high performance risk products personally before making a purchase decision.

The results of the analysis on the single item--"go ahead and pick a brand"--essentially reverse the pattern of the findings on the five information source variables. SubJects tended to select a brand on the basis of very little information when performance was low or moderate. However, when performance risk was high, the ratings on this scale dropped substantially. This may indicate a critical difference in the adoption processes involved in the purchase of high risk versus low risk products. It appears that for products high in performance risk, evaluation may be undertaken before trial through the use of various information sources; on the other hand, the preferred way to evaluate a low performance risk product may be to make a trial purchase.

CONCLUSIONS

With the caution in mind that the present research dealt with a role-playing situation rather than actual behavior, a number of interesting (although highly tentative) conclusions can be drawn:

1. Consumers tend to use more sources of information when faced with increasing degrees of perceived performance risk.

2. Depending upon the level of perceived performance risk, the consumer's relative preference for various information sources shifts dramatically.

3. There is little evidence of meaningful influences on information search behavior by the level of social risk encountered in a purchase situation.

4. Over a wide range of products low or moderate in performance risk, the most frequently used method of information acquisition is to simply buy the product, presumably on a trial basis. When performance risk is high, trial purchase is the least likely form of information acquisition.

5. Direct observation and/or experience with a product is generally preferred to any secondary source of product information.

6. Of the various secondary sources of product information, word-of-mouth appears to be dominant over a broad range of products. Consumer Reports and like materials are useful only under conditions of high performance risk.

Obviously, these conclusions are broad generalizations which are not justified on the basis of the current research alone. However, several of the relationships tend to coincide with findings of previous research, which lends some credence to their potential generalizability. A more appropriate status for these "conclusions" is that of hypotheses for future research in the area of consumer information acquisition. Considerable attention has recently been focused on the problem of providing the consumer with adequate product information. Understanding the processes by which the consumer acquires and handles product information would appear to be a major research priority. Input from such research would be valuable in identifying optimal information dissemination strategies. The present research attempted to apply an objective information search typology to the problem of information acquisition. Future research should be directed at discovering the consumer's own information search typology--i.e., the structure of the information acquisition process.

APPENDIX A

THE BEHAVIORAL DIFFERENTIAL INSTRUMENT

REFERENCES

Andreasen, A. R. Attitudes and customer behavior: A decision model. In H. E. Kassarjian and T. S. Robertson (Eds.), Perspectives in Consumer Behavior. Glenview, Ill.: Scott, Foresman and Company, 1968, 498-510.

Arndt, J. Word of mouth advertising and informal communication. In D. F. Cox (Ed.), Risk Taking and Information Handling in Consumer Behavior. Boston: Harvard University Press, 1967, 188-239.

Bauer, R. A. Consumer behavior as risk taking. In R. S. Hancock (Ed.), Dynamic marketing for a Changing World. Chicago: American Marketing Association, 1960, 389-398.

Bettman, J. R. Perceived Risk: A measurement methodology and preliminary findings. In M. Venkatesan (Ed.), Proceedings of the Third Annual Conference. Association for Consumer ReseArch, 1972, 394-403.

Bettman, J. R. Perceived risk and its components: A model and empirical test. Journal of Marketing Research, 1973, 10, 184-190.

Brody R. P., S. M. Personality variables and the consumer decision process. Journal of Marketing Research, 1968, 5, 50-57.

Bucklin, L. P. The informative role of advertising. Journal of Advertising Research, 1965, 5, No. 3, 11-15.

Cox, D. F. Clues for advertising Strategists. In D. F. Cox (Ed.), Risk Taking and Information Handling in Consumer Behavior. Boston: Harvard University Press, 1967, 112-1-51.

Cunningham, S. M. The major determinants of perceived risk. In D. F. Cox (Ed.), Risk Taking and Information Handling in Consumer Behavior. Boston: Harvard University Press, 1967, 82-108.

Hansen, F. Consumer Choice Behavior: A Cognitive Theory. New York: The Free Press, 1972.

Howard, J. A. and Sheth, J. N. The Theory of Buyer Behavior. New York: John Wiley and Sons, Inc., 1969.

Jacoby, J. and Kaplan, L. B. The components of perceived risk. In M. Venkatesan (Ed.), Proceedings of the Third Annual Conference. Association for Consumer Research, 1972, 382-393.

Lutz, R. J. Cognitive change and attitude change: A validation study. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign, 1973.

Newton, D. A. A marketing communications model for sales management. In D. F. Cox (Ed.), Risk Taking and Information Handling in Consumer Behavior. Boston: Harvard University Press, 1967, 579-602.

Perry, M. and Hamm, B. C. Canonical analysis of relations between socioeconomic risk and personal influence in purchase decisions. Journal of Marketing Research, 1969, 6, 351-354.

Roselius, T. Consumer rankings of risk reduction methods. Journal of Marketing, 1971, 35, No. 1, 56-61.

Sheth, J. N. and Venkatesan, M. Risk reduction processes in repetitive consumer behavior. Journal of Marketing Research, 1968, 5, 307-310.

Triandis, H. C. Exploratory factor analyses of the behavioral component of social attitudes. Journal of Abnormal and Social Psychology, 1964, 68, 420-430.

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