The Roles of Country of Origin Information on Buyers' Product Evaluations: Signal Or Attribute?

Wai-kwan Li, University of Illinois
Kwok Leung, Chinese University of Hong Kong
Robert S. Wyer, Jr., University of Illinois
ABSTRACT - Although recent studies have suggested that country of origin information can be used to infer product quality, (a signalling role), or as a product attribute that can provide benefits, (an attribute role), it is unclear under what conditions each of these roles may occur, and whether one may dominate the other. This study found that the occurrence of these two roles are contingent on two factors: Amount of product information, and motivation. It was found that country of origin information signalled the style, but not the functional performance, of a product. Moreover, the signalling role has dominated the attribute role.
[ to cite ]:
Wai-kwan Li, Kwok Leung, and Robert S. Wyer (1993) ,"The Roles of Country of Origin Information on Buyers' Product Evaluations: Signal Or Attribute?", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 684-689.

Advances in Consumer Research Volume 20, 1993      Pages 684-689

THE ROLES OF COUNTRY OF ORIGIN INFORMATION ON BUYERS' PRODUCT EVALUATIONS: SIGNAL OR ATTRIBUTE?

Wai-kwan Li, University of Illinois

Kwok Leung, Chinese University of Hong Kong

Robert S. Wyer, Jr., University of Illinois

[The authors would like to thank Kent B. Monroe and Marielza Martins for their thoughtful comments on the earlier drafts of the article. This article is based on the master's thesis of the first author, submitted to the Department of Psychology, Chinese University of Hong Kong, under the supervision of the second author. This study was supported by the Hong Kong and China Gas Postgraduate Research Scholarship granted to the first author.]

ABSTRACT -

Although recent studies have suggested that country of origin information can be used to infer product quality, (a signalling role), or as a product attribute that can provide benefits, (an attribute role), it is unclear under what conditions each of these roles may occur, and whether one may dominate the other. This study found that the occurrence of these two roles are contingent on two factors: Amount of product information, and motivation. It was found that country of origin information signalled the style, but not the functional performance, of a product. Moreover, the signalling role has dominated the attribute role.

The blooming of international business has brought about important changes in manufacturing locations and marketing of products. Given these changes, marketing managers need to consider whether buyers' evaluations of a product are influenced by knowledge of the country in which it is made. Although previous research has provided evidence of country-of-origin (COO) effects on product evaluations, not much is known about how country of origin affects buyers' product evaluations (Bilkey and Nes 1982; Ozsomer and Cavusgil 1991).

In an in-depth interview, Li and Monroe (1992) reported that, when the products considered were of high technology, informants used COO information to infer product quality, that is, COO information played a signalling role. On the other hand, when the products considered were handicrafts, COO functioned as a product attribute that can provide benefits to consumers, that is, it played an attribute role. However, an issue remaining to be explored is what role(s) COO may play if the manufacturing process of a product, such as watches, involves both technical and handicraft skills. Does one role dominate the other, or does COO play both signalling and attribute roles simultaneously?

Another interesting issue is what conditions may facilitate or inhibit the use of COO in these capacities. In other words, what are the boundary conditions of these two roles?

To investigate these two issues, we manipulated two factors, information amount and motivation, in such a way that the combinations of the factors might (1) facilitate only the signalling role, (2) facilitate only the attribute role, (3) facilitate both roles, or (4) inhibit both roles.

COO AS A SIGNAL OF PRODUCT QUALITY

Johansson (1989, p.55) argued that COO may function as a "summary cue" that produces a cognitive inference effect: "the cue might be used by the customer to guess the attributes of a product". Supporting Johansson's "summary cue" proposition, Han (1989) found that COO may function as a "halo", from which buyers can infer beliefs. Parallel to them, Havlena and DeSarbo (1991) found that COO functioned as an "indicator" of perceived risk.

Although the terminology used by these authors were slightly different, they suggested similar effects: COO may signal perceived quality in product evaluations. [It should be clarified that the signalling process is different from the heuristic process. By signalling, consumers use a piece of information to infer product quality, when information is not available, or difficult to comprehend (Sctovsky 1945). By heuristic, consumers only use a subset of available information to infer the product quality, in order to save their mental effort (Chaiken, Liberman & Eagly 1989).] This tentative conclusion is consistent with the findings in the pricing literature indicating that price (also an extrinsic attribute) may serve as a signal of perceived quality (Rao and Monroe 1989; Steenkamp 1989; Zeithaml 1988).

What does COO Signal?

Although researchers have offered convincing arguments that COO can play a signalling role in evaluating product quality, what it signals has not been specified. For example, while Johansson (1989) postulated that COO may be used by customers to infer product quality, it is unclear whether he referred to the overall product quality, or the quality of some product attributes. Similarly, Han (1989) proposed that COO can be used to infer the beliefs about a product, but the particular beliefs to be inferred was not specified. If COO influences only some beliefs, then the question is: What are these beliefs?

To investigate this issue, the present study asked subjects to evaluate a product with respect to a comprehensive set of product attributes. By factorizing the set of product attributes, different domains for evaluating a product were identified. Then, the significance of COO effects on these domains were assessed, and we could determine whether COO signals beliefs on some domains of product quality, or on the overall product quality.

Effect of Information Amount on the Signalling Role

Under what conditions do consumers use an extrinsic cue to signal product quality? Scitovszky (1945) postulated that when consumers were not able to assess product quality directly, they might use surrogate measures, such as company size, market success and price to infer product quality. Consumers may be unable to assess product quality in at least two situations. First, when consumers are asked to evaluate a product for which they do not have sufficient knowledge to understand the intrinsic attribute information, they may use the product's extrinsic attributes as indicators of product quality (Rao and Monroe 1988). Second, when consumers are asked to evaluate a product for which only limited product information is available, they may also employ the product's extrinsic cues, such as COO, to infer product quality, and lead to COO effects on product evaluations.

On the other hand, when consumers are confronted with sufficient amount of comprehensible product information, they should be able to evaluate the quality of the product based on the information provided, and therefore have little reason to make any inferences from extrinsic cues. Hence, if COO plays the signalling role, then:

H1: The effect of COO is more likely to occur when a limited amount of product information is available, than when large amount of product information is available. (See Table 1)

TABLE 1

PREDICTED EFFECTS OF COO ON PRODUCT EVALUATIONS AS A FUNCTION OF INFORMATION AMOUNT AND MOTIVATION

COO AS A PRODUCT ATTRIBUTE

Three experimental studies have reported that COO, similar to other intrinsic product attributes, was considered as a product attribute in product evaluations (Hong and Wyer 1989, 1990; Li and Wyer 1991). However, COO usually appears as a "made in" label on a product, it is interesting to investigate what benefits COO (as a product attribute) can provide to consumers. Li and Monroe (1992) identified five kinds of benefits: authenticity, exoticness, patriotism, personalization, and enhanced social standing. For example, Americans may prefer Chinese china to American china, because it is authentic and/or exotic; Americans may prefer American automobiles to Japanese automobiles for patriotic reasons.

Effect of Motivation on the Attribute Role

Literature on motivation and information search suggests that in high motivation situations, consumers have been shown to exert extensive effort in information search (Celsi and Olson 1988; Harris 1987; Zaichkowsky 1986), and attend to both salient and non-salient information (Borgida and Howard-Pitney 1983). On the other hand, in low motivation situations, consumers tend to make quick and effortless decisions, suggesting that only limited information processing would take place (Hoyer 1984), and only salient information was considered (Borgida and Howard-Pitney 1983).

Based on this logic, consumers with high motivation will consider more product attribute information than consumers with low motivation. Hence, if COO is considered as one of the many product attributes, highly motivated consumers are more likely to include COO in the product evaluation process, whereas unmotivated consumers will have a lower probability of including it. Therefore, COO is more likely to play the attribute role in product evaluations, and therefore lead to COO effect, when consumers are of high motivation, rather than when they are of low motivation. Hence, if COO plays an attribute role, then:

H2: The effect of COO on product evaluations is likely to be greater when the motivation of consumers is high, than when it is low. (See Table 1)

CAN SIGNALLING AND ATTRIBUTE ROLES OCCUR SIMULTANEOUSLY?

As discussed above, low information amount will enhance COO to play the signalling role on product evaluations, whereas high motivation will enhance COO to function as a product attribute. Integrating these two factors together, an interesting question arises. Which role(s) may COO play if consumers are highly motivated and only a limited amount of information is available? Will both roles co-exist, or one will dominate the other? Since no data have been reported that one role will dominate over the other, we expect that both roles will co-exist. The predicted effects of country of origin based on signalling and attribute hypotheses are summarized in Table 1.

METHOD

Subjects and Design

Subjects in the main study were 256 undergraduates (128 men and 128 women) at the Chinese University of Hong Kong who participated to fulfill a course requirement. Sixteen subjects (8 of each gender) were randomly assigned to each cell of a 4-factor design involving COO (favorable vs. unfavorable), information amount (high vs. low), motivation to make a correct judgment (high vs. low) and information presentation order (COO first vs. COO later in the information sequence). [It should be noted that the purpose of including the last factor in the present design is for counterbalancing any plausible unexpected effect caused by the order of information presented.]

Selection of Stimulus Materials

Product. Watches were chosen as the product to be evaluated for two reasons. First, since virtually every student owned a watch, product information should therefore be comprehensible to all subjects. Second, manufacturing a watch may involve both technical skills, such as to increase the accuracy, and handcrafting skills, such as workmanship in hand-made watches.

COO. Two countries were selected that had substantially different reputations in manufacturing watches but evoked similar amounts of prior knowledge concerning these products. To make these selections, two sets of pilot data were collected. First, 24 undergraduates who did not perform in the main experiment evaluated the quality of watches made in each of 20 countries along a scale from 1 (extremely poor) to 9 (extremely good), with 0 indicating no impression at all. Ten countries were screened out because more than 10% of the subjects reported that they had no impression about watches made in those countries. The two countries with the highest ratings were Switzerland (X=8.71) and Japan (X=8.04). The two countries with lowest ratings were Mainland China (X=3.42) and Taiwan (X=5.83).

A different group of 48 subjects was asked to write down product information about the watches made in each of the four countries noted above. The number of attributes mentioned was greatest for watches made in Japan (X=9.25), followed by Switzerland and China (in each case, X=7.67) and Taiwan (X=4.08). These differences were assumed to reflect the differences in the amount of information (about watch-making) that could be signified by these country names. Based on these differences, Switzerland and China were chosen as the countries of origin to be used in the main experiment. These two countries, therefore, differed extremely in reputation, whereas the number of inferences subjects could make was the same.

Attribute information. A total of 45 different attribute dimensions were mentioned by the pilot subjects described in the previous paragraph. The 32 attribute dimensions that were mentioned most frequently were selected for use as product information. The eight dimensions that were mentioned most frequently by pilot subjects were assumed to be the most commonly considered bases for judging watches. Attributes along these dimensions, which included price, style, durability and accuracy, were used as product information in all conditions. Of these, four conveyed the favorable pole of the dimension in question, and four conveyed the unfavorable pole of the dimension. [English translation of these and other attributes used in the study may be obtained from the first author.] Attributes along the other 24 dimensions, which were mentioned less frequently by pilot subjects, were used as product information in high-information-load conditions only.

Procedure

Subjects participated in the experiment in groups of four. They were introduced to the study with instructions that they would be asked to view some slides containing information about a watch, and that after receiving this information, they would decide whether or not they would like to purchase the product. They were told that if they made a correct decision, they would receive a monetary reward of either HK$10 (high-motivation conditions) or HK$1 (low-motivation conditions). The criterion for "correctness" was ostensibly based on objective criteria they would be told about later.

Subjects were exposed to each stimulus slide for 5 seconds. (Pretesting determined that this duration was sufficient to read and comprehend the information conveyed.) The first, warm-up slide indicated that the experiment was sponsored by the "Hong Kong and China Gas Company Ltd." This procedure was expected to increase the plausibility of the motivation manipulation, suggesting indirectly that subjects would actually receive the monetary reward indicated. The slides that followed each contained a different piece of information.

In low-information-amount conditions, subjects received descriptions of the COO and eight attributes, with the COO information conveyed either first or seventh in the series. In high-information-amount conditions, subjects first received the 24 additional attributes, followed by the eight attributes, with the COO information presented either first or twenty-fifth in the series. (The eight attributes were presented after the 24 additional attributes in order to control for the time interval between subjects' receipt of the important attribute information and the judgments they were asked to make.)

Assessment of Dependent Variables

After viewing the stimulus slides, subjects completed a product-evaluation questionnaire. First, they evaluated the watch with respect to each of the 32 attributes, regardless of whether they had received information about the attribute. These ratings were made along semantic differential scales from -4 (e.g. ugly, inaccurate) to +4 (e.g. beautiful, accurate). Then, they estimated (a) the overall quality of the watch along a scale from -4 (very bad) to +4 (very good), and (b) how well they liked the watch along a scale from -4 (dislike very much) to +4 (like very much).

Three manipulation checks were obtained. First, subjects were asked to recall the product's COO and then to evaluate the general quality of the watches made in that country along a scale from -4 (very poor) to +4 (very good). Then, they rated the reward they were given for correct judgments along a scale from -4 (too little) to +4 (very reasonable), and the amount of product information they had received along a scale from -4 (insufficient) to +4 (too much).

Upon completing the questionnaire, subjects were debriefed about the nature of the experiment. They were asked not to disclose any information about the experiment to any person. Subjects were then paid according to what they were told, and dismissed.

RESULTS

The evaluations of the 32 product attributes were first reduced to two factors. Then, to verify if the data resemble the pattern suggested by the signalling and/or attribute hypotheses, the effect of COO was initially examined via three 4-way ANOVAs, with COO (Switzerland vs. China), information amount (high vs. low), motivation (high vs low) and information presentation order (COO presented first vs. near the end of the stimulus series) as independent variables; the two factors (from the solution of factor analysis) and overall product evaluation as dependent variables. Information presentation order had no significant effects, and so it will not be discussed further. (The results reported below are based on analyses that are pooled over this factor.)

Manipulation Checks

The manipulations were successful. Subjects believed that watches made in Switzerland were generally of better quality than those made in China (X=2.13 vs. X=-1.56), F(1,254)=365.10, p<.01. Subjects felt the incentive was more reasonable if they had been promised HK$10 than if they had been promised HK$1 (X=.98 vs. X=-.20), F(1,254)=22.10, p<.01. Finally, subjects who had seen 32 pieces of information felt they had received more information than those who had seen only eight pieces (X=1.27 vs. X=-2.15), F(1,254)=195.50, p<.01.

Identifying domains of Specific Product Evaluations

Factor analysis, with oblique rotation, was performed to identify the domains of specific product evaluations. The scree test suggested that a two-factor solution was appropriate. Three attributes were deleted due to double loadings on both factors. The remaining 29 attributes were then reanalyzed and condensed to two factors, which explained 41% of variance. The alpha coefficients were .88 and .87 for factor 1 and 2, respectively. All factor loadings were above .43 and .32 for factor 1 and 2, and no double loadings occurred in the final solution.

Examples of attributes that fell into factor 1 were decoration, style, and color matching. This factor can be conceptualized as "style". Examples of attributes that loaded on factor 2 were water resistance, durability, and shocking resistance. This factor can be labelled as "function". The correlation between these two factors was -.04 (n.s.) which suggests these two factors represented two different domains of watch evaluation, namely, style and function.

TABLE 2

EFFECTS OF COO ON PROUDCT EVALUATIONS AS A FUNCTION OF INFORMATION AMOUNT AND MOTIVATION

Specific and Overall Product Evaluations

Specific product evaluations. The evaluations of the two specific factors, style and function, are shown in Table 2 as a function of information amount, motivation, and COO. The effect of COO on style evaluations was significant in the low information amount conditions, but not in high information amount conditions. However, its effect on function evaluations was not significant in any conditions.

This pattern of results is consistent with the signalling hypothesis and therefore implies that COO played a signalling role, but not an attribute role, on the evaluation of the style of a watch. However, in evaluating the function of the watch, COO did not have a significant role. This finding is consistent with our intuition that evaluation of style is more subjective, and therefore more likely to be affected by our inferences.

Overall Product Evaluations. Subjects' estimates of the quality of the product were correlated .76 with their liking estimates. The two ratings were therefore averaged to provide a single index of overall product evaluations. These evaluations are shown in Table 2. The effect of COO was significant only when both information amount and motivation were high.

This pattern of results lends partial support to the attribute hypothesis, and does not support the signalling hypothesis. Hierarchical regression analyses were therefore conducted on overall product evaluations, to verify if COO really played an attribute role in the high-information-amount and high-motivation condition.

If COO was considered as a product attribute which can provide benefits that are unique and different from other attributes, its effect on overall product evaluations should be over and above the effects of the other attributes. Hence, evaluations of style and function were entered first into the regression equation, followed by COO. The adjusted R2 for style and function together was .20 (p<.01), whereas the change of R2 was .08 (p<.01) when COO was entered into the equation.

Realizing that the two factors might not be a good representation of all 32 attributes, a hierarchical regression analysis was again performed with the 32 attributes entered first, followed by COO. The adjusted R2 for the 32 attributes was .42 (p<.01), whereas the change of R2 was .05 (p<.05) when COO was entered into the equation. [It should be emphasized that the COO effect on overall product evaluation is over and above the effect of the other 32 attributes, and therefore a change of .05 in R2 should be appreciable. In fact, when country of origin was entered first into the regression equation, it explained 37.2% of variance on overall product evaluation.] The results of regression analyses, therefore, confirmed that COO was considered as a product attribute in the overall product evaluation of a watch when both information amount and motivation were high.

DISCUSSION

We have considered two plausible roles that COO may play on product evaluations. Results indicated that COO can play both the signalling and attribute roles on product evaluations. This finding is consistent with the integrative framework proposed by Johansson (1989), as well as the verbal protocols reported by Li and Monroe (1992). In addition, the present study identified the conditions for the occurrence of the signalling and attribute roles. Furthermore, the data also suggest that the signalling role dominated the attribute role under the conditions in which both roles were facilitated.

The Signalling Role

The results reported above provide support for Hypothesis 1. Specifically, subjects used COO as a signal for the style, but not for the functional performance, of a watch. Therefore, we conclude that COO may only signal a certain domain, but not every domain, of product quality. Moreover, COO did not directly signal the overall product quality. The indirect effect of COO on overall product evaluation suggests that whether there is a COO effect or not depends on the relative importance of the "affected" and "unaffected" domains in making overall product evaluations. As in the two conditions in which the signalling role was supported, the regression coefficients of style on overall product evaluations were .22 and .32, respectively. However, the regression coefficients of function on overall product evaluations were .47 and .54, respectively, heavier than that of style on overall product evaluations. Since COO effect was only significant in the less important domain (style), but not significant in a more important domain (function), therefore, COO effect on overall product evaluations was not significant.

The finding that COO only signalled a certain domain of product quality when limited information was presented, but not when sufficient amount of information was presented, can be explained in the light of the economics of information. Stigler (1961) argued that consumers would employ cost and benefit analysis in order to decide whether to conduct information search or not. Obviously, the net benefits of conducting an internal information search are higher when subjects only have limited product information. Therefore, the signalling process occurred in low information amount conditions, but not in high information amount conditions.

The Attribute Role

The results also suggest that subjects regarded COO as a product attribute that can provide benefits to them. This occurred when both motivation and information amount were high. This finding provides partial support to hypothesis 2. However, in the high-motivation and low-information-amount condition, in which both signalling and attribute roles were expected to occur, only a signalling effect was observed. This finding suggest that the signalling role dominated the attribute role.

The dominance of the signalling role over the attribute role can also be explained by economics of information. If we assume consumers do not use a piece of information twice to prevent any "double counting", then the benefits of using COO to infer several different attributes are obviously higher than using it as a single piece of information. Therefore, using COO as a signal was preferred to using it as a product attribute when limited information was available.

On the other hand, when information amount is sufficient, highly motivated subjects were more likely to process and use all of the available information, whereas subjects with little motivation selectively attended to and used only part of the available information in the overall product evaluation process.

Conclusion and Direction for Future Research

The findings reported above revealed that COO can play two different roles in making product evaluations, a signalling role and an attribute role. While the signalling role is more likely to occur when information amount is low, the attribute role is more likely to occur when motivation is high. However, when the condition favors both roles to occur (low information amount and high motivation), the signalling role will dominate the attribute role, and therefore they will not occur simultaneously. Nevertheless, this is only the first study to report the dual roles did not occur simultaneously. There may be some situations in which both roles can occur at the same time. Further replications are necessary to validate the present findings.

It is important to note that COO signalled only the subjective aspect (style) of the product but not the objective aspect (function) of the product. Therefore, we speculate that COO may have stronger effect on "affective" products, such as perfume or ear-rings, but weak effect on "functional" products, such as paper and pencil. However, we definitely need more studies to verify this speculation.

It should be noted that the product we used may limit the external validity of the present finding. As Li and Monroe (1992) indicated, the type of product considered can affect the role that COO may play. Therefore, future studies should try to incorporate products of different technological level to extend the external validity of the present findings.

It is also important to understand whether other extrinsic cues, such as price and brand names, can also play these two roles. Future studies can incorporate these extrinsic cues to examine if their underlying processes are the same as those we identified in the present research. These studies might stimulate the development of a more general theory of extrinsic cues on product evaluations.

REFERENCES

Bilkey, Warren J. and Erik Nes (1982), "Country-Of-Origin Effects on Product Evaluations," Journal of International Business Studies, 13(Spring/Summer), 89-99.

Borgida, E., and Howard-Pitney, B. (1983), "Personal Involvement and the Robustness of Perceptual Salience Effects," Journal of Personality and Social Psychology, 45(March), 560-570.

Celsi, Richard L. and Jerry C. Olson (1988), "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15(September), 210-224.

Chaiken, Shelly, Akiva Liberman and Alice H. Eagly (1989), "Heuristic and Systematic Information Processing within and beyond the Persuasion Context," in Unintended Thought, eds. James S. Uleman and John A. Bargh, NY: Guilford Press, p.212-253

Han, C. Min (1989), "Country Image: Halo or Summary Construct?" Journal of Marketing Research, 26(May), 222-229.

Harris, Greg (1987), "The Implications of Low-Involvement Theory for Advertising Effectiveness," International Journal of Advertising, 6(3), 207-221.

Havlena, William J. and Wayne S. Desarbo (1991), "On the Measurement of Perceived Consumer Risk," Decision Sciences, 22, 927-939.

Hong, Sung-tai and Robert S. Wyer (1989), "Effects of Country-of-Origin and Product-Attribute Information on Product Evaluation: An Information Processing Perspective." Journal of Consumer Research, 16(September), 175-187.

Hong, Sung-tai and Robert S. Wyer (1990), "Determinants of Product Evaluation: Effects of the Time Interval Between Knowledge of a Product's CO and Information about its Country of Origin Specific Attributes," Journal of Consumer Research, 17(December), 277-288.

Hoyer, W. D. (1984), "An Examination of Consumer Decision Making for a Common Repeat Purchase Product," Journal of Consumer Research, 11, 822-829.

Johansson, Johny K. (1989), "Determinants and Effects of the Use of 'Made in' Labels," International Marketing Review, 6(January), 47-58.

Li, Wai-Kwan, and Kent B. Monroe (1992), "The Role of Country of Origin Information on Buyers' Product Evaluations: An In-depth Interview Approach," in AMA Educators' Proceedings: Enhancing Knowledge Development in Marketing, in press.

Li, Wai-Kwan and Robert S. Wyer, Jr. (1991), "Country-Of-Origin Effects on Product Evaluations: The Role of Information Load, Decision Importance, and Product Familiarity," Working paper, University of Illinois at Urbana-Champaign.

Ozsomer, Aysegul and S. Tamer Cavusgil (1991), "Country-Of-Origin Effects on Product Evaluations: A Sequel to Bilkey and Nes Review," in AMA Educators' Proceedings: Enhancing Knowledge Development in Marketing, Volume 2, eds. Gilly et al., Chicago, 269-277.

Rao, Akshay R. and Kent B. Monroe (1988), "The Moderating Effect of Prior Knowledge on Cue Utilization in Product Evaluations," Journal of Consumer Research, 15(September), 253-64.

Rao, Akshay R. and Kent B. Monroe (1989), "The Effect of Price, Brand Name, and Store Name on Buyers' Perceptions of Product Quality: An Integrative Review," Journal of Marketing Research, 26(August), 351-7.

Scitovszky, Tibor (1945), "Some Consequences of the Habit of Judging Quality by Price," Review of Economic Studies, 12(Winter), 100-105.

Steenkamp, Jan-Benedict E.M. (1989), "A Model of the Quality Perception Process,: from Product Quality, Assen/Maastricht, The Netherlands: Van Gorcum, 99-129.

Stigler, G. (1961), "The Economics of Information," Journal of Political Economy, 69(June), 213-225.

Zaichkowsky, Judith L. (1986), "Conceptualizing involvement," Journal of Advertising, 15(2), 4-14.

Zeithaml, Valarie A. (1988), "Consumer Perceptions of Price, Quality, and Value: A Means-End Model and Synthesis of Evidence," Journal of Marketing, 52(July), 2-22.

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