Intern. J. of Research in Marketing 22 (2005) 127 – 139 www.elsevier.

com/locate/ijresmar

Country-of-origin effects in consumer processing of advertising claims
Peeter W.J. Verlegha,*, Jan-Benedict E.M. Steenkampb, Matthew T.G. Meulenbergc
a

ERIM and Department of Marketing Management, Erasmus University Rotterdam, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands b Department of Marketing, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlands c Marketing and Consumer Behaviour Group, Department of Social Sciences, Wageningen University, Hollandseweg 1, 6706 KN, Wageningen, The Netherlands

Abstract We propose that country of origin has a dual impact on product evaluations, acting as informational cue, but also as source variable, moderating the impact of ads on product evaluations. In support, we find a direct effect of country of origin on product evaluations, and a three-way interaction between country of origin, claim favorability and ad involvement. Further analyses show that country of origin influences the way in which consumers respond to moderate and extreme claims under conditions of low and high ad involvement. The dual impact of country of origin on consumer behavior emphasizes its relevance to (international) marketing. D 2005 Elsevier B.V. All rights reserved.
Keywords: Country of origin; Source credibility; Attitude change; International marketing

1. Introduction Consumers differentiate between products from different origin countries, a phenomenon that has become known as the country-of-origin effect (for reviews see Agrawal & Kamakura, 1999; Verlegh & Steenkamp, 1999). The country-of-origin effect is rooted in consumers’ images of the quality of specific products marketed by firms associated with a country of origin. These images have been referred to as
* Corresponding author. Tel.: +31 10 4082732; fax: +31 10 4089011. E-mail address: pverlegh@rsm.nl (P.W.J. Verlegh). 0167-8116/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.ijresmar.2004.05.003

product–country images (Heslop & Papadopoulos, 1993), and may be based on actual product experience, but also on information gathered through advertising and other sources of product information, including word-of-mouth and articles in the popular press. Previous research on country-of-origin effects has examined the composition of product–country images (e.g., Han, 1989; Roth & Romeo, 1992; Verlegh, 1999), and how consumers use country of origin as a cue for determining product quality — either by itself, or in conjunction with other product information (e.g., Haubl & Elrod, 1999). These studies have shown that ¨ consumers use product–country images as a cognitive

128

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

shortcut when evaluating products, especially when other information is scarce (Verlegh & Steenkamp, 1999). Country of origin has a greater impact on product evaluations when consumers are less motivated to process available information, for example when involvement is low (Han, 1989; Maheswaran, 1994). Erickson, Johansson and Chao (1984) found that country of origin may bias consumers’ product beliefs. They found for example that perceptions of cars’ economy were biased upward for Japanese cars, while perceptions of the quality of the cars were biased upward for German cars. Johansson, Douglas and Nonaka (1985) build on this study and confirm that country of origin affects consumers’ product attitudes by biasing their perceptions of particular attributes like gas mileage or driving comfort. The authors also find that this bias is stronger when product knowledge is low. The present study extends this research by examining whether country of origin has a dual role when it is presented in conjunction with other product information (i.e., advertising claims). We propose that country of origin does not only act as an informational cue, but also affects consumers’ interpretation of advertising claims. This view extends the research of Johansson and colleagues (Erickson et al., 1984; Johansson et al., 1985), who showed that country of origin biases perceptions of product attributes. We use the term bsource effectQ to refer to the proposed moderating influence (cf., Goldberg & Hartwick, 1990). The persuasion literature (e.g., Eagly & Chaiken, 1993; Petty, Wegener & Fabrigar, 1997) has examined a dual role for several elements of (advertising) messages, but not for country of origin, although Keller (2003) recently suggested that theories on source credibility may be used to achieve a better understanding of country-of-origin effects. To illustrate the dual role of country of origin in product evaluations, consider an ad for a car made in Russia. Based on actual experience or information obtained from other sources, consumers may have a negative a priori image of the quality of cars made in Russia. The product information supplied by claims in the ad may be used by consumers to update their judgments of Russian cars (cf., Gatignon, 1984). The extent to which consumers use the claims in the ad to update their image of Russian cars will increase with the amount of attention given to the ad (i.e., their level

of ad involvement — McKenzie & Lutz, 1989; Petty & Cacioppo, 1986). In addition, however, the a priori image of Russian cars will affect the perceived credibility of the claims in the ad, especially when consumers are involved with the ad and process it more carefully (cf., Goldberg & Hartwick, 1990). In the remainder of this paper, we examine two ways in which country of origin can influence product evaluations: as an informational variable, and as a source variable. We also discuss the role of ad involvement in both types of influences. In the empirical part of our paper, we present an experiment in which we test hypotheses derived from this framework. Our study uses a fairly large and representative sample of consumers. The paper concludes with a discussion of our findings, and their implications for the study and use of country of origin as a marketing tool.

2. Informational and source variable roles of country of origin 2.1. Country of origin as product information Consumers have (often well-developed) product– country images, i.e., sets of associations related to quality and other attributes of products from a specific country of origin. For a given country, this image and its valence may vary across products, so that it is best defined at the level of product categories (Balabanis & Diamantopoulos 2004; Gurhan-Canli & Maheswaran ¨ 2000). A large body of research shows that consumers use product–country images as information when they evaluate products. This results in different evaluations of identical products with different country-of-origin labels, even when additional product information is presented (for reviews, see Agrawal & Kamakura 1999; Verlegh & Steenkamp, 1999). H1a. Products from a country with a relatively favorable product–country image will be evaluated more positively than products from a country that has a relatively unfavorable product–country image. Country of origin is often presented together with advertising claims that contain product information. Dual-process models like the Elaboration Likelihood Model (Petty & Cacioppo, 1986), and the Heuristic

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

129

Systematic Model (Eagly & Chaiken, 1993) hold that the influence of product–country images depends on the level of consumers’ involvement with the ad (hereafter referred to as bad involvementQ). Dualprocess models distinguish between two modes of information processing. The first mode, viz., heuristic or peripheral processing, refers to low-effort processing in which consumers use shortcuts or heuristics to form an evaluation. The second mode, viz., systematic or central processing, is more elaborate, and entails effortful scrutinizing of attitude-relevant information. A low level of ad involvement implies that consumers rely heavily on heuristic cues such as country of origin (Eagly & Chaiken, 1993; Petty & Cacioppo, 1986). As the level of involvement increases, consumers will be less influenced by country of origin, and elaborate more on additional product information provided in the ad. Hence, consumers should rely more strongly on country of origin in low involvement situations than in high involvement situations (cf., Maheswaran, 1994). H1b. The impact of country of origin on evaluations of an advertised product will be larger when consumers have a low (as opposed to a high) level of ad involvement. 2.2. Country of origin as a source variable When ad involvement is higher, consumers are more motivated to process attitude-relevant information in the ad. As a result, they devote more cognitive resources to the processing of product information that is presented in the form of advertising claims (Petty & Cacioppo, 1986). This does not necessarily mean that the claims get a larger impact on product evaluations. Petty, Cacioppo, and Schumann (1983) show that bweakQ ad claims are evaluated less positively when consumers become more involved with an ad. Consumers who process an ad in a more bcentralQ or bsystematicQ way, will more carefully examine the content of ad claims, so that they attach less value to claims that are weak and not convincing. McKenzie and Lutz (1989) propose that an increase in involvement also implies that consumers pay more attention to the credibility of the source of the advertising claims, in order to determine how much weight should be attached to them. Thus, whether or

not an increase in ad involvement leads to an increase in the impact of ad claims on product evaluations depends on the source and the content of the claims. While marketing research on source credibility has been mostly focused on celebrities and other spokespersons, Rossiter and Percy (1997, p.260) argue that the definition of source variables should not be taken too narrowly, and that entities such as companies and brands should be regarded as sources of advertising claims. This view is in line with studies investigating the role of corporate credibility in consumer evaluations of advertising and other marketing tools (Brown & Dacin, 1997; Goldberg & Hartwick, 1990; Keller & Aaker, 1992). In these studies corporate credibility is defined as bthe extent to which consumers believe that a company can deliver products and services that satisfy customer needs and wantsQ (Keller & Aaker, 1992, p. 37; see also Brown & Dacin, 1997). Analogously, we propose that country-of-origin credibility is determined by consumers’ product–country image. As this image and its valence may vary across products, we define credibility at the category level. For a given category, country-of-origin credibility is high when consumers have a favorable image of the country’s products in that category, and low when the product–country image is unfavorable. Goldberg and Hartwick (1990) have shown that corporate credibility influences consumers’ evaluation of advertising claims. They find that corporate credibility moderates the impact of claim favorability on product evaluations. When corporate credibility is high, extremely favorable claims (e.g., bour product ranked #1 in an independent testQ) lead to more favorable product evaluations than moderately favorable claims (e.g., bour product ranked #5 in an independent testQ). When corporate credibility is low, moderately favorable claims lead to better product evaluations than extremely favorable claims, because the latter are perceived to be less credible. The detrimental effects of overclaiming have been widely recognized in the advertising literature (Rossiter & Percy 1997, p.252). Although increasing ad involvement implies that consumers pay more attention to advertising claims, this does not necessarily imply that these claims subsequently have a stronger effect on product evaluations. Building on Goldberg and Hartwick (1990), we propose that the effect of an increase in

130

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

ad involvement will depend on the favorability of the claims and the credibility of the source. If the favorability of the product–country image of a country of origin determines its source credibility, we may expect a three-way interaction between country of origin, claim favorability and ad involvement. This is visualized in Fig. 1, where the left panel shows the effect of increasing involvement when country-oforigin credibility is high, and the right panel shows the effect of increasing involvement when country-oforigin credibility is low. The impact on product evaluations of increasing levels of ad involvement varies with the favorability of consumers’ product–country images, and the favorability of advertising claims. An increase in ad involvement implies that consumers devote more attention to the ad claims. If country-of-origin credibility is high (i.e., the advertised product is from a country with a favorable product–country image) an increase in attention to the claims implies that consumers attach more weight to the ad claims, so that they lead to more extensive updating of consumers’ a priori images of the product (cf., Gatignon, 1984). This effect is visualized in the left-hand panel of Fig. 1. If the country-of-origin credibility is low, (i.e., the advertised product is from a country with an unfavorable product–country image), higher levels of ad involvement will not necessarily lead to a greater impact of advertising claims on product evaluations, as visualized in the right-hand panel of Fig. 1. Instead, a negative interaction between ad involvement and claim favorability may be expected (cf., Goldberg & Hartwick, 1990). That is, as ad
Favorable Product-Country Image: Credibility is HIGH product evaluation

involvement increases, consumers will pay more attention to the advertising claims. If ad claims are moderately favorable, an increase in attention may lead to more favorable product evaluations, because consumers are less likely to doubt the validity of claims that are moderately favorable. If however, the ad contains claims that are extremely favorable, consumers are likely to doubt the validity of the claims as they pertain to a country of origin that is low in credibility. As a result they perceive these claims to be less credible, and attach less weight to then when they update their a priori product images. Based on the above discussion we hypothesize that: H2. When consumers are presented with an ad for a product from a country with a relatively favorable product–country image, product evaluations will be more favorable when ad involvement is high, than when ad involvement is low, regardless of the level of claim favorability. H3a. When consumers are presented with an ad for a product from a country with a relatively unfavorable product–country image, and this ad contains claims that are moderately favorable, product evaluations will be more favorable when ad involvement is high, than when ad involvement is low. H3b. When consumers are presented with an ad for a product from a country with a relatively unfavorable product–country image, and this ad contains claims that are extremely favorable, product evaluations will not be more favorable when ad involvement is high, than when ad involvement is low.
Unfavorable Product-Country Image: Credibility is LOW product evaluation

Involvement
moderate claims extreme claims

Involvement

Fig. 1. Illustration of Hypotheses 2, 3a and 3b: Country of origin moderates the interaction between claim favorability and ad involvement.

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

131

3. Method 3.1. Study setting We test our hypotheses in a real-world setting, investigating German consumers’ evaluations of tomatoes from The Netherlands. The Netherlands is a major exporter of fruit and vegetables, with an export value of 5.7 billion euros in 2002. About 40% is exported to Germany, rendering it by far the most important export market for Dutch fruits and vegetables (figures obtained from the Dutch Product Board for Horticulture). In the past decade Dutch producers of fresh fruit and vegetables have received a great deal of criticism in Germany. Dutch tomatoes have been targeted in particular, and are referred to as bWasserbombeQ (waterbombs). The criticisms have received widespread attention in the German press. Market research conducted to establish the effects of this negative publicity showed that Dutch tomatoes are well-known (aided brand awareness = 95%), but carry strong negative associations like btastelessQ, bwateryQ and bartificialQ (GfK, 1996, 1998). Thus, Dutch tomatoes suffered from a negative product–country image in their most important export market, which put them at a competitive disadvantage vis-a-vis products from other countries, ` most notably its most important competitor, viz. Spanish tomatoes. Spain is a major exporter of tomatoes to Germany, and German consumers are familiar with Spanish tomatoes (aided brand awareness = 88%). Unlike Dutch tomatoes, Spanish tomatoes are seen as natural and sun-ripened produce (GfK, 1996, 1998). This is confirmed by data obtained from the 1998 edition of the Holland Imago Monitor (HIM), which surveyed 1500 German consumers on their image of agricultural products from different European countries. Table 1 is based on our own analyses of these data. 3.2. Data collection We examined how this difference in product– country images affects the impact of moderately and extremely favorable ad claims on consumers’ evaluations of tomatoes from The Netherlands and Spain. To capture processing effects, we investi-

Table 1 German consumers’ evaluations of Dutch and Spanish tomatoes* Dutch tomatoes Mean (SD) Good taste** Natural Good quality 56.5 (28.3) 51.6 (25.3) 63.3 (26.0) Spanish tomatoes Mean (SD) 81.5 (17.7) 71.7 (19.6) 79.2 (17.6)

* Based on our analyses of data collected by GfK (1998). ** Ratings on a 10-point scale (10–100). bDon’t know Q responses were deleted.

gated this for lower and higher levels of ad involvement. Our questionnaire and related stimulus material were part of a larger survey on consumers’ opinions with regard to food-related issues. We employed a 2 Â 2 Â 2 between-subjects design to manipulate country of origin, claim favorability and ad involvement. Our questionnaire was distributed among a sample of households drawn from a household panel that is maintained by a large market research company. The panel is representative of the general population in terms of age (of the person responsible for grocery shopping) and household size. All our respondents were responsible for grocery shopping within their household, and randomly selected from this panel. Care was taken to match the cells in the design with regard to geographic and demographic characteristics. In total, the questionnaire was sent to 1500 consumers in the western part of Germany. Questionnaires were distributed by mail, and accompanied by a postagepaid reply envelope. We received 813 questionnaires that were filled out completely and properly (54%), but deleted 106 questionnaires of respondents who had indicated that they knew nothing about Dutch or Spanish tomatoes. The remaining 707 respondents were more or less evenly distributed across the eight cells in the design.1 Demographics on this sample are given in Table 2. Data collection was carried out by GfK, a large international market research company. Our questionnaire consisted of two pages containing a short instruction, followed by the measures for our dependent variables. In the instruction consumers were asked to examine the enclosed stimulus material and fill out

1

Cell sizes varied between 81 and 92.

132

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

Table 2 Sample demographics (in percent) Age b 35 35–49 N 49 Household size 1 2 3 4 N4 Children b 15? Yes No Sample 23.6 28.9 47.5

3.4. Independent variables Country of origin was varied across two levels, i.e., The Netherlands and Spain, which differ in favorability of the product–country images for tomatoes (Section 3.1). Claim favorability was varied across two levels, i.e., moderately favorable and extremely favorable. In the bmoderately favorableQ condition, the ad stated that farmers growing Spania (Hollandia) tomatoes paid attention to nature and tradition, which resulted in a quality product, with a good taste and aroma. It also stated that tests had shown that 6 out of 10 consumers rated Spania (Hollandia) tomatoes as tastier than other tomatoes. In the bextremely favorableQ condition, the ad stated that Spania (Hollandia) tomatoes paid attention to nature and tradition more than in any other country, resulting in a product of unmatched quality, with an unsurpassed full taste and a rich aroma. It also stated that tests had shown that 9 out of 10 consumers rated Spania (Hollandia) tomatoes as much tastier than other tomatoes. Ad involvement was varied across two levels, in a manner that is similar to the standard manipulation used in many information processing studies (e.g., Petty et al., 1983). In the low involvement condition, consumers were informed that the advertised tomatoes were not available in their own country, and that the study was a pre-test for a larger study that would take place in a different country. They were instructed to look at the ad, and then fill out the questionnaire. In the high involvement condition, consumers were informed that the tomatoes featured in the ad would soon be introduced in nearby supermarkets, and that the ad was pre-tested on a select group of customers. They were told to carefully look at the ad before filling out the questionnaire. 3.5. Measures Multi-item measures of purchase intentions and product attitudes were used to assess product evaluations. In addition, we included a multi-item measure of ad involvement, and a single-item measure of perceived credibility of the advertising claims. Purchase intention was measured with two items, i.e., bIf I came across X in my store, I would

28.3 34.9 17.5 13.6 5.7

78.6 21.4

the items in the questionnaire. As we discuss below, the introduction was also used to manipulate ad involvement. Throughout the questionnaire items were worded as evaluative statements. Consumers indicated their agreement on a 10-point scale with numbers 10, . . . ,100. The left end of the scale [10] was labeled bnot agree at allQ, and the right end [100] was labeled bfully agreeQ. 3.3. Stimulus materials Stimulus materials were full-page magazine ads (in color) consisting of a body text and pay-off line, supported by a matching visual. The ads were created by one of the authors, and pre-tested in a mallintercept study in which 25 German shoppers were asked to comment on the ads. These comments were used to further improve the ads, which were then sent to professionals at a large advertising agency who made final improvements. We developed four versions of this ad, in which country of origin and claim favorability were varied with two levels each, and the rest of the ads were kept identical. Ad involvement was manipulated in the instruction (cf. Petty et al., 1983). Each ad presented a brand of either Dutch or Spanish tomatoes. We used fictitious brand names derived from the country names to ensure that consumers’ responses reflected their opinion on the advertised tomatoes: we used bHollandiaQ for Dutch tomatoes, and bSpaniaQ for Spanish tomatoes. These brands did not exist in the market at the time of the study.

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

133

definitely buy itQ, and bI would rather buy X than any other brandQ. Reliability of this measure was high, with a correlation between the two items of .79 (Cronbach’s a = .88). Product attitude was measured with three items: bI like XQ, bI appreciate XQ and bI have a positive image of XQ. Reliability of the measure was high (Cronbach’s a = .94). Claim credibility. The ads were evaluated on credibility with one item: bI find the claims in this ad credibleQ. Ad involvement. Consumers’ involvement with the ad was measured by two items, taken from Zaichowsky’s (1985, 1994) involvement inventory: bThis ad is appealingQ, and bThis ad is interestingQ. Construct reliability was good, with Cronbach’s a = .92, and an inter-item correlation of .85.

Table 3 Means of dependent variables in each of the experimental conditions HOLLANDIA Low involved Purchase intention Moderately favorable Extremely favorable Overall Attitude Moderately favorable Extremely favorable Overall Claim credibility Moderately favorable Extremely favorable Overall 45.6 54.6 49.9 High involved 58.9 52.5 55.7 SPANIA Low involved 71.1 65.5 68.4 High involved 61.3 66.3 63.6

50.1 54.6 52.2

60.1 55.9 58.2

70.5 67.6 69.1

67.1 69.3 68.1

56.7 57.6 57.1

60.3 49.8 55.0

69.3 66.8 68.1

66.0 68.6 67.2

4. Results The success of the task involvement manipulation was checked by comparing the means for the ad involvement scales between the high and low involvement groups. Ad involvement was significantly higher (t = 2.07; p b .05) in the high involvement condition (64.0) than in the low involvement condition (60.0). 4.1. Hypotheses testing In this paragraph we discuss the results of analyses of variance (ANOVAs) of the data obtained in the experiment. These analyses were used to test our hypotheses. The cell means obtained in the experiment are shown in Table 3 and Fig. 2A and B. In line with H1a, ANOVAs with purchase intention (PI) and attitude (ATT) as dependent variables revealed significant main effects of country of origin; i.e., a positive difference between consumers’ evaluations for Spanish and Dutch tomatoes 2 (PI:
2 In addition, we entered the two measures for product evaluations into a multivariate analysis. The results of this analyses parallel those of the univariate analyses, and revealed a significant main effect of COO ( F(2,698) = 33.32, p b .001), a significant COO Â Ad Involvement interaction ( F(2,698) = 4.54, p = .011) and a significant three-way interaction of COO, Ad Involvement and Claim favorability ( F(2,698) = 8.91, p b .001). All other effect were not significant ( p N .05).

F(1,699) = 56.58; p b .001; ATT: F(1,699) = 65.54; p b .001). We also find significant interactions between country of origin and ad involvement (PI: F(1,699) = 8.30, p = .004, ATT: F(1,699) = 4.12; p = .043), as proposed in H1b. Inspection of the cell means in Table 3 reveals that the country-of-origin effect is strongest when ad involvement is low: The difference in mean scores for purchase intentions (Spanish vs. Dutch tomatoes) is 18.5 when involvement is low, and 7.9 when involvement is high. The difference in mean scores for attitudes is 16.9 when involvement is low, and 9.9 when involvement is high. These results show that the country-of-origin effect is smaller when ad involvement is high (vs. low), which support H1b. Hypotheses 2, 3a and 3b together specify a threeway interaction of country of origin, involvement and claim favorability. This interaction was significant for PI ( F(1,699) = 13.71, p b .001), and ATT ( F(1,699) = 4.63, p = .032). To investigate this interaction in detail, we separately analyzed the results for products from a country with an unfavorable product–country image (i.e., Dutch tomatoes), and products from a country with a favorable product–country image (i.e., Spanish tomatoes).3
In these analyses, the overall MSE (obtained by conducting a full ANOVA on the entire data set) was used as the error term, so that the error-df is always equal to 699.
3

134

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

A

Hollandia - purchase intention
moderate extreme

B

Spania - attitude
moderate extreme

60 50 40

70 60 50

Low ad involvement

High

Low ad involvement

High

Hollandia - attitude
moderate extreme

Spania - purchase intention
moderate extreme

60 50 40

70 60 50

Low ad involvement

High

Low ad involvement

High

Hollandia - credibility
moderate extreme

Spania - credibility
moderate extreme

60 50 40

70 60 50

Low ad involvement

High

Low ad involvement

High

Fig. 2. (A) Purchase intention, Attitude and Claim credibility for Hollandia tomatoes, for moderately vs. extremely favorable claims under low vs. high ad involvement. (B) Purchase intention, Attitude and Claim credibility for Spania tomatoes, for moderately vs. extremely favorable claims under low vs. high ad involvement.

For Spanish tomatoes source credibility is high, which according to H2 should lead to positive effects of ad involvement on product evaluations. Our data, however, indicate that the main effect of ad involvement, as well as that of claim favorability, was not significant for either of our measures, so that H2 is not supported by our results. We do however find an interaction between claim strength and ad involvement that is significant for PI ( F(1,699) = 4.93, p = .037), but not for ATT ( F(1,699) = 1.15, p = .28). Further analyses of the means in Table 3 reveal that

this interaction is due to an absence of an effect of ad involvement for Spanish tomatoes advertised with extremely favorable claims ( F(1,699) b 1), and a negative effect of ad involvement for Spanish tomatoes advertised with moderately favorable claims that is significant for purchase intention ( F(1,699) = 7.98; p = .005), but not for attitude ( F(1,699) = 1.09; p = .30). One possible explanation for this finding is that consumers’ a priori images of Spanish tomatoes were so favorable that they were not affected by extremely favorable claims, while the moderately

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

135

favorable claims led to a downward adjustment of the evaluation. We briefly expand on this in the Discussion section. For products from a country with an unfavorable product–country image (i.e., Dutch tomatoes), the source credibility is low. This should lead to a positive effect of ad involvement when claim favorability is moderate (H3a), and a negative or zero effect when claim favorability is high (H3b). Together, H3a and H3b predict an interaction effect of claim favorability and ad involvement on product evaluations (for products from a country with an unfavorable product–country image). ANOVAs reveal that this interaction is significant (PI: F(1,699) = 10.01, p = .002, ATT: F(1,699) = 3.96, p = .047). The means in Table 3 show that evaluations of Dutch tomatoes advertised with moderately favorable claims are more favorable under high involvement than under low involvement conditions (PI: F(1,699) = 15.15, p b .001; ATT: F(1,699) = 10.43, p = .001). These findings lend strong support for H3a. For Dutch tomatoes advertised with extremely favorable claims we found no effect of ad involvement ( F(1,699) b 1), which is in line with H3b. The interaction is visualized in the upper two panels of Fig. 2A. It is noteworthy that the results for Dutch tomatoes advertised with moderately vs. extremely favorable claims show a reversal in the effects of claim favorability under high and low involvement (see Fig. 2A). For purchase intentions we find a positive effect of claim favorability when involvement is low (PI: F(1,699) = 6.69; p = .010), reversing to a negative effect when involvement is high (PI: F(1,699) = 3.53; p = .061). For attitudes, we find similar, but nonsignificant effects: (Low involvement: F(1,699) = 1.87; p = .172, High involvement: F(1,699) = 2.10; p = .148). 4.2. Effects on claim credibility and tests for mediation For products from a country of origin with an unfavorable product–country image, H3a and H3b propose that moderately favorable claims lead to product evaluations that are more positive when ad involvement is high than when ad involvement is low, whereas extremely favorable claims lead to product evaluations that are equal or less positive

when ad involvement is high (vs. low). As noted in Section 2.2., this interaction is mediated by claim credibility: when extremely favorable claims are made by a source low in credibility (i.e., a country of origin with an unfavorable product–country image), an increase in consumers’ attention to these claims will lead to a decrease in the perceived credibility. This negative effect of ad involvement will not be present for moderately favorable claims. These predictions are corroborated by ANOVAs with claim credibility as a dependent variable, which revealed a significant interaction between claim favorability and ad involvement ( F(1,699) = 5.51, p = .019). This finding can be attributed to the fact that an increase in ad involvement has no effect on claim credibility when claims are moderately favorable ( F(1,699) = 1.11, p = .29), and a negative effect when claims are extremely favorable ( F(1,699) = 5.08, p = .024). These results are visualized in the bottom graph of Fig. 2A, and the means are represented in Table 3. We obtained no significant effects on claim credibility for ads for Spanish tomatoes (for cell means see Table 3 and the bottom graph of Fig. 2B). This line of reasoning implies that, for products from a country with an unfavorable product–country image (i.e., Dutch tomatoes), the interaction of ad involvement, and claim favorability should be mediated by the perceived credibility of the claims made in the ad. We examined this notion in a moderated mediation analysis, following the procedure recommended by Baron and Kenny (1986). These authors propose that mediation occurs when three conditions are met. Firstly, the independent variable (i.e., the two-way interaction term in our model) should have a significant effect on the dependent variables (i.e., purchase intention and attitude, respectively). Secondly, the independent variable should have a significant effect on the mediating variable (i.e., claim credibility). Thirdly, the effect of the independent variable on the dependent variable should reduce in terms of effect size when the mediating variable is incorporated in the analyses. In the last step, a (substantial) reduction in terms of effect size is referred to as partial mediation. Complete mediation is shown when the effect of the independent variable reduces to nonsignificance. The three steps in this analysis are discussed below.

136

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

Firstly, ANOVAs with purchase intention and attitude as respective dependent variables revealed that the interaction was significant for both measures (PI: ( F(1,699) = 10.01, p = .002; g 2 = .014; ATT ( F(1,699) = 3.96, p = .047; g 2 = .006). Secondly, the interaction is also significant in an ANOVA with claim credibility as a dependent variable ( F(1,699) = 5.51, p = .019, g 2 = .008). Thirdly, ANCOVAs with claim credibility as covariate and purchase intention and attitudes as respective dependent variables, reveals a significant effect of the claim credibility covariate ( F(1,698) = 393.18; p b .001), as well as substantially reduced interaction effects. For purchase intention, the remaining interaction effect is significant ( F(1,698) = 4.45, p = .035), but the effect size is reduced by 55%, when compared to the effect obtained without claim credibility as covariate (g 2 = .006 vs. g 2 = .014). For attitude, the remaining interaction effect is not significant ( F(1,698) = .12, p = .72), and the effect size is reduced by 97%, when compared to the effect obtained without claim credibility as covariate (g 2 = .0002 vs. g 2 = .006). Thus, claim credibility partially mediates the twoway interaction of claim favorability and ad involvement when purchase intention is used as dependent variable, and we find full mediation when attitude is used as dependent variable.

5. Discussion 5.1. Findings and implications This study examined a dual role of country of origin in influencing consumers’ product evaluations. We find support for the notion that country of origin acts both as information variable and as source variable. We find that country of origin strongly influences consumer product evaluations, even in the presence of additional information presented by ad claims. This finding adds to the body of research indicating that consumers use country of origin as an informational variable, and reinforces the notion that country of origin plays an important role in consumer product evaluations (cf., Verlegh & Steenkamp, 1999). Furthermore, our results show that consumers rely more on country of origin when they are less involved with the ad. This supports the notion that

consumers use country of origin as a cognitive shortcut, a strategy that is relied upon when consumers seek to minimize cognitive efforts (Han, 1989; Maheswaran, 1994). Earlier studies (Erickson et al., 1984; Johansson et al., 1985) show that country of origin affects consumers’ perceptions of product attributes. We extend this finding by showing that country of origin may act as a source variable that moderates the effect of ad claims on product evaluations. In line with the literature on (corporate) source credibility, we propose that the source credibility of a country of origin is higher when consumers associate it with a more favorable product–country image. In line with earlier research (e.g., Goldberg & Hartwick, 1990), we argue that country-of-origin credibility determines the extent to which consumers are influenced by moderately vs. extremely favorable ad claims. We extend this research by explicitly incorporating the role of ad involvement in this process. Ad involvement is an important determinant of the extent to which consumers carefully consider both the credibility of the source and the favorability of claims. At lower levels of involvement, consumers are more likely to take ad claims at face value, and incorporate them into their product evaluations. However, as involvement increases, consumers more carefully examine the level of favorability of ad claims vis-a-vis their product–country images. In that ` case, an ad for a product from a country with an unfavorable product–country image will be perceived as less credible when it features extremely favorable claims, than when it features moderately favorable claims. In other words, we propose that increasing claim favorability has a negative effect when source credibility is low, and ad involvement is high. This notion is supported by a significant interaction between ad involvement and claim favorability when product–country image is unfavorable. The impact of this interaction on product evaluations is mediated by the perceived credibility of the advertising claims. When product–country image is favorable, the source credibility of the country of origin is high, and we hypothesized that an increase in ad involvement should have a positive effect on product evaluations, regardless of the favorability of advertising claims. Our results do not support this

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139

137

hypothesis, but instead we find a marginally significant interaction, which can be attributed to the fact that ad involvement has no effect on evaluations of products from a country of origin with a favorable product–country image when ad claims are extremely favorable, and ad involvement has a negative impact on evaluations when ad claims are moderately favorable. This finding can be accommodated in our framework if we assume that consumers’ product–country images for Spanish tomatoes were more favorable than the moderately favorable claims used in our ad. In that case, both averaging and updating processes of impression formation predict that an increase in the weight that is attached to ad claims (caused by an increase in ad involvement) produces a decrease in product evaluations. These results corroborate Rossiter and Percy’s (1997, p. 252) warning that bUnderclaiming is just as much an error as overclaimingQ. Although plausible, this explanation of our findings remains post-hoc, and alternative explanations might be construed. One such alternative explanation4 is that consumers rejected the extreme claims altogether, and instead relied on their a priori images based on the country of origin of the products. These and other explanations remain to be examined in future research on over- and underclaiming and its effects on product evaluation. 5.2. Implications and suggestions for future research Marketers often seek to leverage the effects of country of origin on consumer product evaluations, and attempt to build brand equity by associating their brand to a country of origin with favorable connotations (Leclerc, Schmitt & Dube, 1994; Keller, ´ 2003). A disadvantage of this strategy is its inherent vulnerability, which is due to the fact that consumers’ perceptions of country of origin may be influenced by many factors beyond the control of (individual) marketers, including negative publicity and low quality products by other brands from the same country of origin. These factors may produce changes in consumers’ product–country images that are able to transform country of origin from an asset to a liability.
4

We thank one of the reviewers for suggesting this explanation.

Those who are marketing products or brands from a country of origin with an unfavorable product– country image are often advised to conceal or at least de-emphasize the origin of the product (e.g., Roth & Romeo, 1992). This is not always a feasible practice, as trade regulations often require a clear and legible indication of the country of origin for a product. Moreover, consumers often associate brands with specific countries (cf., Leclerc et al., 1994), so that country of origin is implicitly communicated through the brand name (note that this may be enhanced by past advertising with an emphasis on country of origin). Thus, when faced with negative product– country images, marketers might be forced to address these images directly, for example by means of advertising campaigns. In doing so, it should be taken into account that country of origin is not merely a shortcut that consumers use to form product evaluations. The present study indicates that consumers use country of origin when they seek to determine the credibility of advertising claims, especially if ad involvement is high. In order to effectively improve consumers’ product evaluations, marketers should adapt the (favorability of their) claims to the favorability of consumers’ product– country images. Our study indicates that country of origin affects how advertising claims are processed under conditions of low vs. high ad involvement. An interesting extension of our study might be to examine countryof-origin effects in conjunction with other factors that are known to affect information processing, such as product expertise, time pressure or repeated exposures. All of these factors influence the amount of message elaboration, allowing consumers to more carefully consider the credibility of advertising claims. Another issue that warrants further research is the generalizability of our results to other products and attributes. The advertising claims studied in this paper related to attributes that are not verifiable before purchase. Previous research (e.g., Feick & Gierl, 1996) has shown that consumers are more skeptical toward advertising for experience claims. This suggests that high source credibility is more important when claims cannot be verified before purchase (cf. Jain & Posavac, 2001). In spite of its limitations, we feel that the present study offers an important extension to our current

138

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139 Goldberg, M. E., & Hartwick, J. (1990). The effects of advertiser reputation and extremity of advertising claims on advertising effectiveness. Journal of Consumer Research, 17(2), 172 – 179. Gqrhan-Canli, Z., & Maheswaran, D. (2000). Determinants of country-of-origin evaluations. Journal of Consumer Research, 27(1), 96 – 108. Han, C. M. (1989). Country image: Halo or summary construct? Journal of Marketing Research, 26(2), 222 – 229. Haubl, G., & Elrod, T. (1999). The impact of congruity between ¨ brand name and country of production on consumers’ product quality judgments. International Journal of Research in Marketing, 16(3), 199 – 215. Heslop, L. A., & Papadopoulos, N. (1993). But who knows where and when: Reflections on the images of countries and their products. In N. Papadopoulos, & L. A. Heslop (Eds.), Product– country images: Impact and role in international marketing. New York7 Haworth Press. Jain, S. P., & Posavac, S. S. (2001). Prepurchase attribute verifiability, source credibility, and persuasion. Journal of Consumer Psychology, 11(2), 169 – 180. Johansson, J. K., Douglas, S. P., & Nonaka, I. (1985). Assessing the impact of country of origin on product evaluations: A new methodological perspective. Journal of Marketing Research, 22(3), 388 – 396. Keller, K. L. (2003). Brand synthesis: The multidimensionality of brand knowledge. Journal of Consumer Research, 29(4), 595 – 600. Keller, K. L., & Aaker, D. A. (1992). The effects of sequential introduction of brand extensions. Journal of Marketing Research, 29(1), 35 – 50. ´ Leclerc, F., Schmitt, B. H., & Dube, L. (1994). Foreign branding and its effects on product perceptions and attitudes. Journal of Marketing Research, 31(2), 263 – 270. Maheswaran, D. (1994). Country of origin as a stereotype: Effects of consumer expertise and attribute strength on product evaluations. Journal of Consumer Research, 21(4), 354 – 365. McKenzie, S. B., & Lutz, R. J. (1989). An empirical examination of the structural antecedents of attitude toward the ad in an advertising pretesting context. Journal of Marketing, 53(2), 48 – 65. Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion. New York7 Springer. Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Central and peripheral routes to advertising effectiveness: The moderating role of involvement. Journal of Consumer Research, 10(2), 134 – 148. Petty, R. E., Wegener, D. T., & Fabrigar, L. R. (1997). Attitudes and attitude change. Annual Review of Psychology, 48, 609 – 647. Rossiter, J. R., & Percy, L. (1997). Advertising communications and promotion management (2nd ed.). Oxford (UK)7 Blackwell. Roth, M. S., & Romeo, J. B. (1992). Matching product category and country image perceptions: A framework for managing countryof-origin effects. Journal of International Business Studies, 23(4), 477 – 497.

knowledge of country-of-origin effects. The finding that country of origin acts as a source variable that moderates the processing of advertising claims, stresses the relevance of studying country-of-origin effects, and at the same time broadens our view of this phenomenon. A thorough understanding of countryof-origin effects continues to be of great importance to academics and practitioners in the field of international marketing.

Acknowledgments The authors thank Alfred Dijs, GfK, and the Dutch Horticultural Board for their kind cooperation. This paper has benefited from the insightful comments of the editor and three anonymous reviewers. Ale Smidts, Luk Warlop, Rik Pieters, and seminar participants at Tilburg University provided helpful comments at earlier stages of the research.

References
Agrawal, J., & Kamakura, W. A. (1999). Country of origin: A competitive advantage? International Journal of Research in Marketing, 16(4), 255 – 267. Balabanis, G., & Diamantopoulos, A. (2004). Domestic country bias, country-of-origin effects, and consumer ethnocentrism: A multidimensional unfolding approach. Journal of the Academy of Marketing Science, 32(1), 80 – 95. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173 – 1182. Brown, T. J., & Dacin, P. A. (1997). The company and the product: Corporate associations and consumer product responses. Journal of Marketing, 61(1), 68 – 84. Eagly, A., & Chaiken, S. (1993). The psychology of attitudes. Worth Fort7 HBJ. Erickson, G. M., Johansson, J. K., & Chao, P. (1984). Image variables in multi-attribute product evaluations: Country-oforigin effects. Journal of Consumer Research, 11(4), 694 – 699. Feick, L., & Gierl, H. (1996). Skepticism about advertising: A comparison of East and West German consumers. International Journal of Research in Marketing, 13(3), 227 – 235. Gatignon, H. (1984). Toward a methodology for measuring advertising copy effects. Marketing Science, 3, 308 – 326. GfK (1996). Image der Hollandischen agrarwirtschaft. Research ¨ Report, AGB Fresh Foods, Dongen (NL) (in German). GfK (1998). Holland imago monitor 1998. Research Report, GfK, Dongen (NL) (in Dutch).

P.W.J. Verlegh et al. / Intern. J. of Research in Marketing 22 (2005) 127–139 Verlegh, P. W. J. (1999). Ingroups, outgroups and stereotyping: Consumer behavior and social identity theory. Advances in Consumer Research, 26, 162 – 164. Verlegh, P. W. J., & Steenkamp, J. -B. E. M. (1999). A review and meta-analysis of country-of-origin research. Journal of Economic Psychology, 20(5), 521 – 546.

139

Zaichowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research, 12(4), 341 – 352. Zaichowsky, J. L. (1994). The personal involvement inventory: Reduction, revision, and application to advertising. Journal of Advertising, 23(4), 59 – 70.