Country-Of-Origin, Perceived Risk and Evaluation Strategy

Dana L. Alden, University of Hawaii
Wayne D. Hoyer, University of Texas
Ayn E. Crowley, Washington State University
ABSTRACT - This study finds that consumers evaluate a new offering from a frequently purchased, non-durable product category in fundamentally different ways depending on whether the product is manufactured in a country perceived as high or low on product quality risk. When consumers associate low risk with the product's country-of-origin (CO) attribute, they evaluate the good based primarily on product category-level generalizations. A high risk CO results in significantly greater use of attribute information associated with the actual product. Based on this research, firms planning to manufacture convenience goods in foreign countries perceived as risky should consider modifying promotional materials to better match consumer evaluation strategies.
[ to cite ]:
Dana L. Alden, Wayne D. Hoyer, and Ayn E. Crowley (1993) ,"Country-Of-Origin, Perceived Risk and Evaluation Strategy", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 678-683.

Advances in Consumer Research Volume 20, 1993      Pages 678-683

COUNTRY-OF-ORIGIN, PERCEIVED RISK AND EVALUATION STRATEGY

Dana L. Alden, University of Hawaii

Wayne D. Hoyer, University of Texas

Ayn E. Crowley, Washington State University

ABSTRACT -

This study finds that consumers evaluate a new offering from a frequently purchased, non-durable product category in fundamentally different ways depending on whether the product is manufactured in a country perceived as high or low on product quality risk. When consumers associate low risk with the product's country-of-origin (CO) attribute, they evaluate the good based primarily on product category-level generalizations. A high risk CO results in significantly greater use of attribute information associated with the actual product. Based on this research, firms planning to manufacture convenience goods in foreign countries perceived as risky should consider modifying promotional materials to better match consumer evaluation strategies.

INTRODUCTION

Developments such as the impending free trade agreement between Mexico and the United States are likely to increase foreign production of lower involvement, consumer convenience goods (Business Week 1991). However, research investigating the effects of country-of-origin (CO) on evaluation strategies and outcomes has focused primarily on higher involvement, durable products such as cars and VCRs (Ozsomer and Cavusgil 1991). Thus while marketing managers have information on the ways that CO can affect evaluation strategies for high involvement goods (cf. Hong and Wyer 1989), far less is known about possible CO effects on evaluation of routinely purchased convenience items such as toothpaste, cereal and film.

Unlike high involvement goods, low involvement goods are rarely evaluated rigorously prior to purchase (Hoyer 1984). For most low involvement purchases, consumers tend to rely on a few salient, surface features (e.g., brand name) that activate generalizations from memory about the product category and brand reputation (Petty and Cacioppo 1986). These evaluation strategies differ from the more deliberate, multiattribute approaches associated with high involvement products (cf. Fishbein and Ajzen 1975; Petty and Cacioppo 1986).

Research in psychology and consumer behavior suggests that certain factors associated with a low involvement product's country-of-origin may cause consumers to shift from relatively passive evaluation strategies to more active strategies associated with higher involvement products. However, this possibility remains untested. Thus, the following study tests for ways that high perceived risk associated with a product's country-of-origin can affect evaluation strategies for a new low involvement brand within a mature product category.

COUNTRY-OF-ORIGIN AND EVALUATION STRATEGY

Research concerning country-of-origin (CO) effects on evaluation strategy has focused on several issues (for a general review, see Ozsomer and Cavusgil 1991). First, researchers have employed multiattribute models to examine interactions between overall evaluation, attribute evaluation and CO (cf., Johansson, Douglas and Nonaka 1985; Han 1989). Second, relationships between product category familiarity, CO and evaluation strategy have been studied (cf., Heimbach, Johansson and MacLachlan 1989). Third, researchers have tested alternative processing models that help explain CO effects. For example, Hong and Wyer (1989) find that CO effects on cognitive processes that mediate evaluation are most consistent with the elaboration likelihood model (Petty and Cacioppo 1986). Having established an elaboration effect, Hong and Wyer (1989, p.185) call for research that will answer "the question of why a product's country of origin stimulates interest in other product information." We now review literature that may help answer this question.

THE CATEGORIZATION PERSPECTIVE

People often attempt to match newly encountered persons with existing stereotypes based on one or two salient features (Fiske and Pavelchak 1986). For example, seeing a middle-aged male wearing a gray suit in an office elevator may activate the category stereotype, "typical businessman." If the man in the suit matches the perceiver's stereotype for a typical businessman, "category-based" evaluation is hypothesized (Fiske and Pavelchak 1986). Under category-based evaluation, the evaluator remains relatively uninvolved and bases his or her impression on thoughts and feelings stored in memory about businessmen in general. Actual information about the individual (e.g., type of watch worn) is less likely to be included in the evaluation.

However, certain factors may cause observers to shift from reliance on summary information from memory to closer analysis of individual attributes associated with the person or object. For example, if the observer notices something unexpected about the businessman (e.g., he or she is carrying a red rose) a shift to "attribute-based" evaluation may occur (Fiske and Pavelchak 1986; Sujan 1985). In addition, Neuberg and Fiske (1987) found that subjects shifted to from category-based to attribute-based evaluation of a new person when they had to depend on the person to achieve an important goal. However, when they weren't dependent on the new person to achieve an important goal, subjects continued to used category-based evaluation. This result was obtained regardless of whether information about the new person was consistent or inconsistent with a category label (e.g., "schizophrenic") provided by the researchers describing that person.

The present study applies findings from the categorization literature to the issue of country-of-origin effects on evaluation strategy. It does so by testing whether a third factor, perceived risk associated with a product's country-of-origin, can also cause a shift from category-based to attribute-based evaluation.

APPLICATION OF CATEGORIZATION THEORY TO CO EFFECTS ON EVALUATION STRATEGY

The categorization perspective may be applied to understanding how a product's CO can impact product evaluation strategy for low involvement consumer goods. Based on that stream, category-based evaluation would be predicted when the product's CO does not strongly deviate from general expectations.

The prediction of category-based evaluation for a low involvement good given a CO that does not deviate strongly from expectations appears to run counter to Hong and Wyer (1989) who concluded that the CO attribute's "inherent interestingness" can produce elaboration. However, Hong and Wyer (1989) used two highly involving products (vcr's and pc's). As a result, motivation to process is likely to have been relatively high. Under these conditions, elaboration may have been more easily triggered by a CO cue, particularly if the CO was seen as providing information relevant to the product category (Heimbach, et al. 1989). On the other hand, an expected, low risk CO attribute (e.g., "Made in the USA" for toothpaste) seems unlikely to motivate elaboration when product category involvement is low, experience is high and other attributes match expectations. Thus, for many frequently purchased non-durables, regardless of CO, consumers seem likely to favor category-based over attribute-based evaluation strategies.

However, there may be times when a CO causes a shift to attribute-based evaluation, even when product involvement is relatively low. As noted, certain CO labels (e.g., developing nations) are associated with increased perceptions of risk and negative evaluations (Bilkey and Ness 1982). Higher risk in turn has been found to increase task involvement (cf. Lichtenstein, Bloch and Black 1988). Under conditions of higher task involvement, related research suggests that individuals are likely to become motivated to form more accurate judgements of attribute information (cf., Neuberg and Fiske 1987; Kruglanski and Freund 1983). As a result, they are likely to rely more on attribute-level information and less on category-level generalities cued from memory. Thus, it appears reasonable to predict a greater use of attribute-based evaluation as opposed to category-based evaluation when a high versus a low risk CO is perceived.

While there may be instances in which the CO label simply won't be noticed for low involvement products, it seems likely that as American convenience goods manufacturers shift production to Mexican factories to take advantage of lower labor costs and less stringent environmental regulations (Business Week 1991) the salience of the CO attribute will increase, causing more consumers to more actively search-out or at least notice that attribute. In the longer term, the U.S. Congress could require more prominent display of CO as pressure from labor groups is levied, further increasing the likelihood that the attribute will be noticed.

Finally, based on Bilkey and Ness (1982), it is also likely that consumers will have relatively more negative attitudes towards a high risk versus a low risk CO. If so, given that a shift from category-based to attribute-based evaluation occurs, relatively less favorable attitudes toward the brand with the high versus the low risk CO would be expected (Fiske and Pavelchak 1986; Sujan 1985). These propositions are now tested.

METHOD

Subjects

Subjects from a large American university were used in this study because young adults are regular purchasers of consumer non-durables (Peter and Olson 1990). In addition, the study applies an experimental approach recommended in causal research (Cook and Campbell 1979). As such, one's ability to generalize must be weighed against threats to statistical conclusion validity (Cook and Campbell 1979). Thus, while use of a fairly homogeneous subject pool may limit external validity, chances of finding actual effects are increased.

Pretests

Based on an initial pretest with 40 subjects, a regularly purchased product category (toothpaste) rated as low on enduring involvement was selected for the main study. Attributes typically associated with the category were elicited using an open ended approach from 31 new subjects (cf., Sujan 1985). The four attributes listed most frequently (freshens breath, fights cavities, great taste and available in gel or regular) were selected for further pretesting.

Next, two new groups of 15 subjects were randomly exposed to descriptions of a "new toothpaste product being considered for introduction nationally." The description included the four attributes discussed above (plus brief descriptions following Sujan 1985) and one of two CO attributes, "Made in USA" or "Made in Mexico." Mexico was selected based on pretests indicating that it was seen as significantly riskier than the US for toothpaste (p<.001), but similar to other developing countries (e.g., Bangladesh and India) and because of its growing importance as an offshore production site for American consumer goods (Business Week 1991). The CO label was listed as the first attribute below the "Toothpaste" heading, although Hong and Wyer (1989) indicate that CO order effects may be minimal. As expected, the description with the Mexico CO (Mean=3.64) was rated as significantly higher on risk (8="very risky"; 1="not at all risky") than the description with the US CO (Mean=2.20; t[27]=2.76, p<.005).

The same two groups of pretest subjects were also asked to rate the extent to which either description was similar to "toothpastes already on the market." As noted, previous research indicates that a strongly unexpected or strongly atypical CO attribute may produce elaboration effects on processing that are similar to those predicted to result from a high risk CO (cf., Fiske and Pavelchak 1986). Thus, establishing that a product description with a Mexico CO is perceived as highly risky but not strongly unexpected would strengthen confidence that any observed elaboration effects were due primarily to risk and not pleasant surprise or risk-free curiosity. To this end, the same subjects rated the Mexico and US CO descriptions on two 8-point scales designed to tap perceptions of typicality and similarity to existing products in the U.S. (e.g., 8=very typical;1=not at all typical). The Mexico CO description (Mean=6.86) and US CO description (Mean= 6.13) were rated as equally typical of (t[27]= 1.19, p>.25) and equally similar to (Mexico Mean=7.1; US Mean=7.2; t[27]=0.39, p>.70) other toothpastes already on the U.S. market.

Based on these pretests, within a product description context, the Mexico CO seems far less likely to produce amused surprise or curious interest and far more likely to increase perceived risk, probably due to uncertainty concerning developing country production standards (Bilkey and Ness 1982). However, even if the Mexico CO is viewed as somewhat "atypical," this perception could exert an elaboration effect indirectly by increasing perceptions of risk. That is, because a central cause of perceived risk is "uncertainty" (Dowling 1986), an unexpected CO that also heightens uncertainty (rather than simply producing pleasant surprise) should also heighten perceived risk. For these reasons, only the possibility of a direct effect from pleasant, risk-free surprise would appear to constitute a genuine threat to the study's internal validity. This possibility will be examined further in the main study.

A final pretest was conducted to determine whether or not presence of a well-known American brand name would counter higher levels of perceived risk associated with the Mexico CO. Two groups of 20 new subjects were exposed to one of two descriptions that were verbally introduced as a Proctor and Gamble product that would be marketed under the brand name "Crest." In addition, above the description of the new product was the statement, "This product, which will be sold as a CREST brand, has the following features." Below this statement was either the Mexico or the US CO and the same four attributes pretested previously. Despite the leading American manufacturer's brand name, the product description with the Mexico CO (Mean = 2.85) was judged to be significantly higher on a 7-point purchase risk scale than the description with the US CO (Mean = 1.80, t[38]=2.45, p<.02). And, once again, ratings of typicality did not vary by CO. Thus, the presence of a well-known brand name does not appear to negate the perception of higher risk associated with the Mexico CO, nor does it appear to affect the perceived typicality of the two descriptions.

Procedures for the Main Study

Forty-four subjects participated in the main study. Two subjects were dropped from the sample, one whose age exceeded the mean sample age by more than three standard deviations and another who had lived in the U.S. for two months. All remaining subjects were raised in the U.S. except for two who had lived in the country for three and ten years respectively. There were no significant differences between treatment groups in terms of age or sex. Subjects were run individually by a trained experimenter blind to the study's hypotheses.

Subjects first read an introductory passage explaining that a company considering introduction of a new toothpaste was sponsoring the study and that the firm's name was confidential. After completing a demographic questionnaire. the experimenter asked them to "form an impression of the new product based on this description" (i.e., the same description used in the pretest). When finished, subjects were asked by the experimenter to verbalize thoughts they had while forming their impression (Wright 1980). Evaluation times and thoughts were recorded using a small camcorder unobtrusively positioned out of the subject's view. Finally, subjects were asked to respond to several scale items which measured attitude toward the brand, purchase intention, perceived risk and product description typicality.

Manipulation Checks

Manipulation check items were collected after the thought listings. Risk perceived for the new toothpaste product was operationalized in terms of purchase risk (7-point scale with 1="not at all risky" and 7="very risky") and in terms of perceived product quality (7-point semantic differential scale with "high" and "poor quality"). On the purchase risk scale, the product description with the Mexico CO (Mean=3.38) was judged as significantly riskier than the description with the US CO (Mean=1.86; F[1,40]= 12.9, p<.001). On the product quality scale, the description with the Mexico CO (Mean=3.71) was judged as significantly lower on quality than the US CO description (Mean=5.29; F[1,40]=12.26, p<.001).

The mean difference in typicality ratings for the two descriptions was significant (Mexico CO Mean=5.24 and US CO Mean=6.05; F[1,40]=4.003; p<.05). However, this difference was less than one scale point. In addition, multiple classification analysis revealed that the perceived risk effect of the CO manipulation was substantially stronger (R2=.24) than the atypicality effect (R2=.09). Finally, analysis of thought listing data indicated that while 60% of the subjects reported thinking about product risk for the description with the Mexico CO, fewer than 10% (2 subjects) expressed surprise without also mentioning risk. Differences in the reported thoughts of the two groups are clearly reflected in Figures A and B. These results, coupled with pretest data, indicated that the Mexico CO enhanced perceived risk relative to the US CO for the overwhelming majority of subjects as intended.

Finally, analysis of thought listings indicated no evidence of hypothesis guessing or other demand effects.

Analysis Overview

Multivariate analysis of variance with CO (U.S. vs. Mexico) as a two-level independent factor, four dependent process measures (described below) and a covariate was used to test the proposition that a high risk CO can cause a shift in evaluation strategy for low involvement, frequently purchased goods.

Dependent Measures

Response Time: In conjunction with other dependent measures, it is assumed that the longer one takes to form an impression, the greater the likelihood of attribute rather than category-based evaluation (Sujan 1985; Neuberg and Fiske 1987). This is because memory-based category evaluation is relatively rapid while evaluation based on actual product attributes is thought to take longer (cf. Sujan 1985). Response time, based on eye contact with the description, was measured post-hoc from videotapes of the session by an experimenter blind to treatment conditions. Time taken to read the introductory passage served as a baseline reading speed measure.

Thought Listings: Verbal responses were collected immediately following exposure to the description (Sujan 1985). Subjects described all thoughts they remembered having while looking at the description. Verbal responses were later coded by two assistants (blind to the study hypothesis) as: category thoughts, attribute evaluation thoughts or other thoughts. Interjudge agreement, based on coder agreement on classification of thoughts across all three categories, was 81% which is acceptable for studies using this approach (Wright 1980). Disagreements were resolved through discussion with a tie-breaking vote by the experimenter needed in only four cases. Three measures of evaluation strategy were used: number of category thoughts, number of attribute evaluation thoughts and number of total thoughts (category, attribute evaluation and all other thoughts). Subjects engaged in category-based evaluation were expected to have more category evaluation thoughts, fewer attribute evaluation thoughts and fewer total thoughts while the opposite was expected for those engaged in attribute-based evaluation (Fiske, Neuberg, Beattie and Milberg 1987).

Attitude Toward the Brand

In addition to the MANCOVA analysis described above, attitude towards the brand was also measured. A four item semantic differential affect scale (like-dislike, good-bad, appealing-unappealing, and desirable-undesirable) was found to exhibit high internal reliability with Cronbach's alpha equal to .95. As a result the scale items were summed to form a single measure of attitude toward the brand.

RESULTS

Analyzing the MANCOVA model first, an initial plot of evaluation response time and the reading speed covariate indicated non-linearity. Optimal fit was achieved with a log transform of the dependent measure (r-squared=.24, p<.001). In the full model, reading speed was significantly related to the log transform of evaluation time (F[1,39]= 14.04, p<.001) but not to the other dependent measures (all Fs<1). In addition, the homogeneity of regression assumption for covariance analysis was satisfied (F<1).

The multivariate model provides evidence for concluding that subjects exposed to the high risk CO relied significantly less on stereotypes from memory and more on actual product description attributes than those seeing the low risk CO (Hotelling's T-square [4,36]=4.21, p<.007). Thus, higher levels of CO risk appear to cause shifts in evaluation strategy away from category-based approaches to attribute-based approaches.

Univariate analyses were consistent with this conclusion (see Table for sample means). As expected, high risk subjects (Adjusted Mean = 41.14) took marginally longer to evaluate the product description than low risk subjects (Adjusted Mean = 36.42; F[1,39]=3.13, p<.08). Furthermore, as predicted, high risk subjects had: 1) significantly more total thoughts than low risk subjects (7.57 versus 4.71; F[1,40]=9.28, p<.004); 2) marginally fewer category thoughts (0.76 versus 1.33; F[1,40]=3.43, p<.06); and 3) significantly more attribute evaluation thoughts (2.86 versus 0.95; F[1,40]= 15.95, p<.001). In addition, the difference on the attribute evaluation measure remained significant even when all thoughts referring to the CO attribute were removed (1.76 versus 0.81; F[1,40]= 4.82, p<.034). This result suggests that the shift in processing strategy due to high risk produced significantly more elaboration regarding all product attributes and not just the CO attribute.

TABLE

CELL MEANS FOR DEPENDENT MEASURES

Finally, as hypothesized, attitude towards the brand with the Mexico CO (Mean=17.1) was significantly lower than for the brand with the US CO (Mean= 21.62; F[1,40]= 8.04, p<.007). Overall, these results provide convergent evidence that subjects exposed to the Mexico CO relied more on attribute-based evaluation and less on category-based evaluation compared to those exposed to the US CO. Furthermore, pretests and manipulation checks suggest that these observed effects were due primarily to perceived risk.

DISCUSSION

This study demonstrates that country-of-origin (CO) can have a significant impact on the evaluation strategy used by consumers for a new product from a routinely purchased product category. Subjects exposed to the high risk CO for a new toothpaste product took longer to form an impression, had more total and attribute evaluation thoughts, had fewer category thoughts and had more negative attitudes toward the brand. These results provide strong convergent evidence of a shift from category-based to attribute-based evaluation due to the presence of the Mexico CO label (Sujan 1985).

In terms of theory development, our findings extend the current consumer behavior literature by demonstrating that perceived risk, possibly in combination with some degree of unexpectedness, can cause a shift in consumer evaluation strategies from category-based to attribute-based evaluation. Prior to this study, only incongruity with expectations (given ability) had been shown to produce such a shift in the consumer behavior field (Sujan 1985). Hence, risk as well as ability and unexpectedness appear to be related to consumers' relative reliance on category-based or attribute-based evaluation.

Second, this study extends the country-of-origin stream by demonstrating a specific type of elaboration response that can occur when consumers encounter a high risk CO. Although Hong and Wyer (1989) note that a CO's "interestingness" may enhance elaboration, they do not test for a shift from category-based to attribute-based evaluation. Neither do they examine low involvement product categories. Thus, these results indicate that Fiske and Pavelchak's (1986) model of alternative evaluation strategies accurately predicts one consumer processing response to perception of high CO risk for a new, low involvement good.

On this note, one might argue that the effects of a high risk CO observed in this study were due not to a shift from category-based to attribute-based evaluation but rather resulted from a shift to another category schema (cf. Meyers-Levy and Tybout 1989). For example, subjects may have taken longer to process the product description with the Mexico CO not because they shifted to attribute-based processing but because they shifted to a general schema for "products from Mexico" which could have then guided their evaluation.

However, two of the study's findings provide evidence for strongly questioning this alternative explanation. First, after removing attribute evaluations that referred to CO, the total number of such evaluations was still significantly higher for the Mexico CO description. This result indicates that subjects increased elaboration of actual description attributes rather than cuing a second schema from memory. Second, the fact that subjects exposed to the Mexico CO expressed significantly fewer category thoughts than subjects who saw the US CO suggests that the former group indeed shifted to attribute-based evaluation rather than remaining at the category level. In sum, while there may be times when consumers continue to use category-level evaluation upon encountering a risky CO, this does not appear to have happened here. At the same time, conditions under such alternative evaluation responses may occur deserve additional attention.

From a managerial perspective, the study suggests the importance of a match between CO, consumer evaluation strategy and the depth and breadth of information provided in promotional materials. For example, an established U.S. consumer goods manufacturer may decide to move production of a new line of toothpastes to a Mexican "maquiladora" to take advantage of lower labor costs and the impending free trade agreement. In light of this study, the firm should consider providing more detailed package information than it would for a domestically manufactured toothpaste brand, e.g., additional information on ingredient quality, a money back guarantee and a toll free number. Should consumers shift to attribute-based evaluation, positive attribute affect should be combined with feelings about toothpastes in general and negative affect for the Mexico CO. Thus, inclusion of such additional positive information should enhance evaluation relative to a standardized, domestic packaging strategy which assumes that a few salient cues are sufficient regardless of CO.

FIGURE A

EXAMPLES OF TYPICAL THOUGHTS EXPRESSED AFTER REVIEW OF PRODUCT DESCRIPTION UNITED STATES COUNTRY-OF-ORIGIN

FIGURE B

EXAMPLES OF TYPICAL THOUGHTS EXPRESSED AFTER REVIEW OF PRODUCT DESCRIPTION MEXICO COUNTRY-OF-ORIGIN

Limitations

Limitations to this study point to potential avenues for future research. First, use of the experimental method which is crucial to ruling-out alternative explanations nonetheless introduces elements of artificiality that may not be operative in the real world. For example, given increased salience of America's ongoing trade problems and the growing numbers of consumer goods produced overseas (e.g., "Made for Hartz in Thailand"), recognition of the CO attribute seems likely to grow for many convenience goods. Even so, for many products, consumers may not notice the "Made-In" label. Thus, having found a strong risk effect in a controlled environment, future researchers should investigate ways in which these findings are affected by factors that may be operative in non-experimental settings such as supermarkets.

Second, one product (toothpaste), one country (Mexico) and one consumer segment (students) were used in the study. Incorporation of additional product categories, countries and consumer segments would increase generalizability. For example, CO effects on evaluation strategy are likely to interact with other important marketing factors such as brand name (e.g., Crest), warranty and store reputation (Thorelli, Lim and Ye 1988). Such potential mediating factors need to be investigated.

Finally, future researchers may want to attempt to include high risk, expected countries in their design so that the specific effects of risk and unexpectedness can be better discerned experimentally. While pretest and experimental data demonstrated that the Mexico CO was first and foremost a manipulation of risk, CO atypicality probably exerted some indirect effect (i.e., by heightening uncertainty and therefore risk) and possibly a mild direct effect (e.g., pleasant surprise or curiosity). However, it may not be possible to remove all unexpectedness from a high risk CO manipulation. If so, future researchers should present evidence (as in this study) of a strongly predominant manipulation effect to demonstrate construct validity.

CONCLUSION

In sum, this study indicates that the riskier the CO, the more likely it is that consumers will use attribute-based evaluation rather than simple cues and category-level images from memory. Managers may find that increasing the depth and breadth of attribute information on the package, in point-of-purchase displays and/or in advertising is a beneficial response. Consumers might then form more positive evaluations than they would in response to standard promotional material as enhanced attribute benefits would be available to compensate for the high risk CO.

REFERENCES

Bilkey, Warren J. and E. Ness (1982), "Country-Of-Origin Effects on Product Evaluations," Journal of International Business Studies 35 (Spring), 89-99.

Business Week (1991), "Assembly Lines Start Migrating from Asia to Mexico," (July 1), New York, NY: Mc-Graw Hill, 43.

Cook, Thomas D. and Donald T. Campbell (1979), Quasi-Experimentation: Design and Analysis Issues for Field Settings, Boston, MA: Houghton-Mifflin Co.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Fiske, Susan T., Steven L. Neuberg, Ann E. Beattie, and Sandra J. Milberg (1987), "Category-Based and Attribute-Based Reactions to Others: Some Informational Conditions of Stereotyping and Individuating Processes," Journal of Experimental Psychology 23, 399-427.

Fiske, Susan T. and Mark A. Pavelchak (1986), "Category-Based versus Piecemeal-Based Affective Responses: Developments in Schema-Triggered Affect," in Handbook of Motivation and Cognition, R.M. Sorrentino and T.E. Higgins (eds.), New York: The Guilford, 167-203.

Han, C-M (1989), "Country Image - Halo or Summary Construct?," Journal of Marketing Research 26(2), 222-229.

Heimbach, Arthur E., Johny K. Johansson, Douglas L. MacLachlan (1989), "Product Familiarity, Information Processing and Country-of-Origin Cues," in Advances in Consumer Research 16, Thomas K. Srull (ed.), Provo, Utah: Association for Consumer Research, 460-67.

Hong, Sung-Tai and Robert S. Wyer, Jr. (1989), "Effects of Country-of-Origin and Product Attribute Information on Product Evaluation: An Information Processing Perspective," Journal of Consumer Research, 16 (2), 175-80.

Johansson, Johny K., Susan P. Douglas, and Ikujiro Nonaka (1985), "Assessing the Impact of Country-of-Origin on Product Evaluations: A New Methodological Perspective," Journal of Marketing Research, 22 (November), 388-396.

Kruglanski, D. and Freund, T. (1983), "The Freezing and Unfreezing Lay-Inferences: Effects of Impressional Primacy, Ethnic Stereotyping, and Numerical Anchoring," Journal of Experimental Social Psychology, 19, 448-468.

Lichtenstein, Donald R., Peter H. Bloch and William C. Black (1988), "Correlates of Price Acceptability," Journal of Consumer Research, 15 (2), 43-52.

Meyers-Levy, Joan and Alice Tybout (1989), "Schema Congruity as a Basis for Product Evaluation," Journal of Consumer Research, 16(1), 39-54.

Neuberg, S.L. and S.T. Fiske (1987), "Motivational Influences on Impression Formation: Outcome Dependency, Accuracy-Driven Attention and Individuating Processes," Journal of Personality and Social Psychology, 53 (3),431-44.

Ozsomer, Aysegul and S. Tamer Cavusgil (1991), "Country-of-Origin Effects on Product Evaluations: A Sequel to Bilkey and Ness Review," in 1991 AMA Educators' Proceedings 2, Mary Gilly, et al. (eds.), Chicago, IL: American Marketing Association, 269-277

Peter, J. Paul and Jerry C. Olson (1990), Consumer Behavior and Marketing Strategy, 2nd Edition, Homewood, IL: Irwin.

Petty, Richard E. and John T. Cacioppo (1986), Communication and Persuasion: Central and Peripheral Routes to Attitude Change, New York: Springer-Verlag.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgements," Journal of Consumer Research 12 (June), 31-46.

Thorelli, Hans B., Jeen-Su Lim, and Jongsuk Lee (1988), "Relative Importance of Country of Origin, Warranty and Retail Store Image on Product Evaluations," International Marketing Review, 6 (1), 35-46.

Wright, Peter (1980), "Message-Evoked Thoughts: Persuasion Research Using Thought Verbalizations," Journal of Consumer Research, 7 (September), 151-175.

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