Measuring the Effects of Framing Country-Of-Origin Information: a Process Tracing Approach

Irwin P. Levin, University of Iowa
J. D. Jasper, University of Iowa
Gary J. Gaeth, University of Iowa
ABSTRACT - One appeal to consumer nationalism is to communicate information about the use of American workers in manufacturing a particular product. Phased narrowing, a newly developed process tracing technique, was used to model how the impact of this information, "framed" differently across experimental conditions, changes over successive decision stages in narrowing down the choice of an automobile. In the presence of other cues, the nationalistic cue had its greatest effect among consumers scoring high on a scale of nationalism and when framed as "% American workers employed" rather than "% non-American workers employed." This framing effect increased in magnitude across successive decision stages. These results have implications for how and when country-of-origin information should be provided.
[ to cite ]:
Irwin P. Levin, J. D. Jasper, and Gary J. Gaeth (1996) ,"Measuring the Effects of Framing Country-Of-Origin Information: a Process Tracing Approach", in NA - Advances in Consumer Research Volume 23, eds. Kim P. Corfman and John G. Lynch Jr., Provo, UT : Association for Consumer Research, Pages: 385-389.

Advances in Consumer Research Volume 23, 1996      Pages 385-389

MEASURING THE EFFECTS OF FRAMING COUNTRY-OF-ORIGIN INFORMATION: A PROCESS TRACING APPROACH

Irwin P. Levin, University of Iowa

J. D. Jasper, University of Iowa

Gary J. Gaeth, University of Iowa

ABSTRACT -

One appeal to consumer nationalism is to communicate information about the use of American workers in manufacturing a particular product. Phased narrowing, a newly developed process tracing technique, was used to model how the impact of this information, "framed" differently across experimental conditions, changes over successive decision stages in narrowing down the choice of an automobile. In the presence of other cues, the nationalistic cue had its greatest effect among consumers scoring high on a scale of nationalism and when framed as "% American workers employed" rather than "% non-American workers employed." This framing effect increased in magnitude across successive decision stages. These results have implications for how and when country-of-origin information should be provided.

Country-of-origin information has traditionally been communicated to consumers as simply the nationality of the company manufacturing the product (Bilkey & Nes, 1982; Gaedeke, 1973; Han, 1988; Hong & Wyer, 1989; Obermiller & Spangenberg, 1989). With the advent of multinational or "hybrid" products, however, this simple classification is at best incomplete. Products such as automobiles may have parts made and/or assembled in several different locations. Recent research has thus turned to how people respond to other country-of-origin cues, one of which is the percentage of American workers employed in manufacturing a product. It has been shown, for example, that consumers prefer companies employing a larger rather than a smaller percentage of American workers in manufacturing and assembling automobiles (Levin & Jasper, 1996). It has also been suggested that this particular cue may play its biggest role in the early phases of considering choice options while final choices may be more apt to be governed by price and quality. In other words, it has been suggested that the American consumer as a gesture to supporting the American worker may initially strive to consider companies that employ mostly American workers but will ultimately choose on the basis of value (Levin, Jasper, Mittelstaedt, & Gaeth, 1993).

Several questions, of course, arise in investigating this issue: 1) What is the most effective way of communicating information about employment of American workers? In this study we compare two ways of "framing" this information: giving the "percentage of American workers" employed in manufacturing a product or the "percentage of non-American workers". 2) How is this information weighted at different stages of the decision process? In this study we use a multi-attribute multi-stage decision task. 3) How do individuals differ in using this information? In this study we use an individual difference measure of consumer nationalism derived from Shimp and Sharma's (1987) ethnocentrism scale. The last two questions were addressed with the aid of a new method that we call "phased narrowing" (Levin & Jasper, 1995). This method is an extension of earlier methods designed to trace the process by which choice options are screened and narrowed down before reaching a final decision (Beach & Potter, 1992; Bettman, 1979; Newell & Simon, 1972; Payne, 1976; Payne, Bettman, & Johnson, 1993). What's unique about phased narrowing is that it requires decision makers to use a series of discrete steps and, with the aid of special analytic tools, it is capable of examining changes in attribute importance across stages and of relating these changes to measurable individual differences.

In our original study (Levin & Jasper, 1995), for example, subjects were given a multiattribute-multioption phased decision making task in which they began with 18 automobile options and were asked in successive stages to narrow them down to 6, then to 3, and finally to 1. These successive stages roughly parallel what researchers studying the formation and use of consideration sets would label as the transition from "awareness set" to "consideration set" to "choice set" to "final choice" (Nedungadi, 1990; Roberts & Lattin, 1991; Shocker, Ben-Akiva, Boccara, & Nedungadi, 1991). One of the attributes used to describe each of the 18 options was "% American workers employed." The major finding was that the importance of this nationalistic cue decreased for most subjects as they approached a final decision. However, for those scoring highest on a scale of consumer nationalism, this tendency reversed. For these subjects, the attribute "% American workers" actually increased in importance as they approached a final decision.

The present study was designed to extend this research by investigating how the processes uncovered in the preliminary study would be impacted by the manner in which the information about the percentage of American workers was presented. In fact, this will be the first study to evaluate whether information framing effects increase or decrease as one approaches a final decision. Previous research using objectively equivalent product information labeled in different terms (e.g., 50% success rate vs. 50% failure rate of a medical treatment, or 80% lean vs. 20% fat ground beef) has shown that more favorable evaluations are produced with labels evoking positive associations than with labels evoking negative associations (Christensen et al., 1991; Davis & Bobko, 1986; Levin & Gaeth, 1988; Marteau, 1989; Wilson et al., 1990). In the present case, identifying a product as employing "50% American workers" should be more apt to call attention to the virtue of keeping Americans employed than identifying a product as employing "50% non-American workers". Therefore, we predict that employment information presented as "% American workers employed" should lead to more favorable product evaluations than equivalent information presented as "% non-American workers employed". In addition, based on previous research and theory, we predict that the importance or weight of the employment factor relative to other factors should be greater for those scoring high on a scale of consumer nationalism or ethnocentrism than for those scoring low and that changes in the importance of this factor over successive decision stages should be different for high and low nationalism consumers; specifically, only high nationalism consumers should show an increasing emphasis on employing American workers as they approach a final choice. Because the issue has never been investigated before, interactions involving the framing variable and decision stage are an open question, as is the issue of whether the effect of information framing on initial impressions and associations is transient or persistent.

FIGURE 1

STIMULUS DESIGN FOR SET OF 18 OPTIONS

METHOD

Design and Procedure

Subjects (105 undergraduate students at a large midwestern university) were tested in small groups of size 2 to 12. Subjects in the "% American" condition (N=52) were given the following cover story:

You will be presented with a number of different choice situations involving cars. Each car will be described by a retail price, a rating of overall quality, and information concerning the percentage of American workers involved in making that particular brand. The overall quality ratings, in each case, are obtained from a leading independent consumer publication (e.g., Consumer Reports). The lowest (worst) possible rating is 0 and the highest (best) possible rating is 100. Each rating is derived from a weighted average of all important features associated with cars. Percentages of American workers in this study are based on the number of man-hours required to make and/or assemble each car and its components.

For subjects in the "% non-American" condition (N=53) the phrase "percentage of American workers" was replaced by "percentage of non-American workers." Each subject was then given an envelope containing 18 individual cards, on which each "brand" option (identified by the letters A through R) was described by the percentage of American (or non-American) workers involved in making the product, an overall quality rating, and a retail price.

Choice options were constructed under the assumption that subjects who choose on the basis of price will select options of low price, subjects who choose on the basis of quality will select options of high quality, and subjects who choose on the basis of nationalistic cues will select options manufactured by a high percentage of American workers. The 18 options depicted in Figure 1 have the property that no one option dominates any other option on all three attributes. The "X," "Y," and "Z" designations serve to mark those options where a high, medium, and low percentage of American workers (or low, medium, and high percentage of non-American workers), respectively, were employed in the manufacturing process. (The equivalent version of 80% American workers, for instance, is 20% non-American workers, and so on.) These placements were designed to make % American workers independent of price and quality which are negatively related.

Parts 1, 2, and 3 (the successive stages task) followed the cover story and appeared on separate pages of the experimental booklet. Each part required subjects to select a smaller number of cards from those cards that they had either been given (Stage 1) or had selected previously (Stages 2 and 3). The critical instructions for each part, respectively, were as follows:

Please open your envelopes, and take out all of the cards. After examining each of the 18 brands carefully, choose 6 brands that you would be interested in looking at if window shopping for a car. The 6 brands that you select should be brands that you would want, at some point in time, to examine first-hand at a dealership.

Assume now that you're actually interested in buying a new car. Look over the six cards again that you selected in Part 1. Choose 3 brands from among those six that you would seriously consider buying.

Again, look over the cards. This time examine the three brands that you selected in Part 2. Which 1 of these three brands do you think you would actually buy?

Finally, subjects were given a 10-item attitude survey designed to measure their nationalism/ethnocentrism. Nine of the 10 items in the survey were taken directly from the CETSCALE developed by Shimp and Sharma (1987). The 9 items chosen correspond to numbers 1, 3, 5, 8, 11, 13, 15, 16, and 17 of that scale. A 10th item, extent of agreement or disagreement with "Buy America first" (used by Levin et al., 1993), was also included in the survey. Responses to each statement were made on a 7-point Likert-type scale where strongly agree=7 and strongly disagree=1. The sum of all 10 items defined our nationalism score. In previous research, this score has been found to be significantly higher for subjects owning American cars than for subjects owning foreign cars.

FIGURE 2

MEAN STANDARD SCORES FOR THE NATIONALISTIC CUE AS A FUNCTION OF FRAME, DECISION STAGE, AND NATIONALISM GROUP

The nationalism survey was administered last in order to avoid the possibility of cueing subjects to the fact that we were examining the role of nationalism in their choice behavior. Subjects in each framing condition were classified as High Nationalism if their scores were in the top third of scorers on the nationalism scale. Those scoring in the bottom two-thirds on the scale were classified for present purposes as Low Nationalism, because our previous research has consistently shown that those scoring low or medium tend to respond similarly.

RESULTS

The results of primary interest center around the differential influence that country-of-origin information has across stages depending on whether it is labeled as "% American workers" or "% non-American workers." With phased narrowing, the relative influence or impact of an attribute can be tracked across stages by computing the mean value of that attribute for the options selected at each stage. The logic here is that the more important an attribute is in the selection process compared to other attributes, the more likely it will be that the options selected will be favorable on that attribute (and possibly unfavorable on other attributes).

In order to compare different attributes on the same scale, values are converted into standard scores. For each attribute, values are converted by subtracting the middle level of the attribute in the stimulus design and dividing by the standard deviation of the attribute levels. Positive standard scores represent selection of options that are, on average, above the middle level of an attribute; negative standard scores represent selection of options that are, on average, below the middle level and unfavorable on that attribute. Because options available at one stage depend on selections made at earlier stages, absolute standard score values are not independent across stages. Nevertheless, differences in the mean standard score from one stage to the next represent the marginal incremental influence of the target attribute at that stage. We believe that this way of computing and comparing attribute importance is simpler and more direct than other methods that require inferences from non-choice data.

Figure 2 displays the mean standard scores for the attribute "% American/non-American workers" for options selected at each decision stage and classified on the basis of nationalism group and framing condition. Inspection of Figure 2 reveals the following trends: 1) The mean standard scores for the nationalistic attribute are higher in the "% American" framing condition than in the "% non-American" condition; 2) The mean standard scores are higher for the High Nationalism group than for the Low Nationalism group; 3) The difference between the two nationalism groups and the difference between the two framing conditions (for each nationalism group) increase in magnitude across stages; and 4) In general, scores decrease or remain constant across stages, except for the High Nationalism subjects in the "% American" frame condition, for whom there was actually an increase across stages.

In order to test the reliability of these trends, a 2 X 2 X 3 ANOVA was conducted with two between-subjects factors (2 levels of information frame and 2 levels of subject nationalism) and one within-subject factor (3 levels of decision stage). The following effects were significant at or beyond the .01 level: frame, F(1, 101)=15.66; nationalism group, F(1, 101)=25.18; decision stage x frame, F(2, 202)=5.43; and decision stage x nationalism group, F(2, 202)=6.73. The interaction between decision stage and frame shows that the magnitude of the framing effect increases reliably across stages. The interaction between decision stage and nationalism group shows that the difference between High and Low Nationalism subjects also increases reliably across stages. As a result of the conjunction of these effects, the nationalistic cue had its largest effect in Stage 3 for High Nationalism subjects receiving the information in terms of "% American workers employed" and its smallest effect in Stage 3 for Low Nationalism subjects receiving the same information in terms of "% non-American workers employed".

Although not the primary interest of the present paper, the use of standard scores also permits a comparison of the relative importance of different attributes at each stage. By comparing standard scores across attributes, we found that quality was generally the most important factor, followed by % American/non-American workers, and then price for the particular array of attribute levels in the present study. An interesting exception, however, was that the importance of the nationalistic cue equaled that of quality in Stage 3 for the High Nationalism group in the % American condition.

DISCUSSION

This research extends previous research (Levin et al., 1993; Levin & Jasper, 1996) in showing that the percentage of American workers employed in manufacturing a particular product is a source of information related to country-of-origin which can potentially affect consumers' choices of "hybrid" products, and, like the earlier studies, we found this to be especially true for those consumers who score high on a scale of nationalism or ethnocentrism. An interesting new finding, however, is that those placing the greatest importance on the employment factor were those who scored highest on a scale of nationalism and who received the information framed in terms of employing American workers rather than in terms of employing non-American workers. Furthermore, this particular group was the only one for whom the influence of the employment factor actually increased across decision stages.

The observed "framing effect" is of particular interest because the literature appears mixed as to whether "accentuating the positive" or "accentuating the negative" has the greatest effect on attitudes and behavior. Prior work by Levin and Gaeth (1988), however, suggests that when a particular attribute of an object to be judged is labeled in such a way as to produce positive rather than negative associations (e.g., lean beef is healthy whereas fat beef is unhealthy), then more favorable evaluations can be expected. To distinguish this from other types of framing effects, this phenomenon has been called "attribute framing" (Levin, Schneider, Gaeth, & Conlon, 1995). In the present case, if employing American workers is seen as a desirable goalCas it certainly appears to beCthen stressing the percentage of American workers employed rather than the percentage of non-American workers employed, makes a product much more attractive.

The specific nature of the interaction between frame and decision stage provides additional new information about attribute framing. In sum, it appears that the focus on positive or negative aspects of choice options produced by positive or negative labels continues to guide decisions and has a cumulative effect that intensifies across stages. An interesting question for future research is whether there is a general tendency for an independent variable like information frame or a subject variable like nationalism that initially focuses attention on a particular factor to have even greater effect in later stages.

Results showing that attribute impact changes over decision stages and across subject characteristics support the use of appropriate process tracing methods, like phased narrowing, to provide further insights into how individual consumers process information. Because of the promise shown by the phased narrowing method, current work is aimed at testing its validity by showing how choices are influenced by the constraints imposed by forced stages and by assessing consumers' reactions to using this technique. In this test, responses will be compared between phased and unphased decision tasks.

In addition to the theoretical and methodological implications of this study, there are practical implications as well. Recently, information has been made public about where the various parts of a multinational or "hybrid" product (such as many current makes of automobile) are made and assembled. The current study shows that in order to provide the maximum appeal to consumer nationalism, this information should stress the percentage of American workers employed. The use of a phased decision making task, in addition, allowed us to show that in some instances the impact of this form of country-of-origin information may actually increase as one approaches a final decision. For some consumers then, information concerning where parts are made should be particularly relevant if supplied at the point of purchase.

REFERENCES

Beach, L. R., & R. E. Potter (1992). The pre-choice screening of options. Acta Psychologica, 81, 115-126.

Bettman, J. (1979). An information processing theory of consumer choice. Reading, MA: Addison-Wesley.

Bilkey, W. J., & E. Nes (1982). Country-of-origin effects on product evaluation. Journal of International Business Studies, 13, 89-99.

Christensen, C., P. S. Heckerling, M. E. Mackesy, L. M. Bernstein, & A. S. Elstein (1991). Framing bias among expert and novice physicians. Academic Medicine, 66(9, Suppl.), 76-78.

Davis, M. A., & P. Bobko (1986). Contextual effects on escalation processes in public sector decision making. Organizational Behavior and Human Decision Processes, 37, 121-138.

Gaedeke, R. (1973). Consumer attitudes toward products made in developing countries. Journal of Retailing, 49, 13-24.

Han, C. M. (1988). The role of consumer patriotism in the choice of domestic versus foreign products. Journal of Advertising Research, 25-32.

Hong, S., & R. S. Wyer (1989). Effects of country-of-origin and product-attribute information on product evaluation: An information processing perspective. Journal of Consumer Research, 16, 175-187.

Levin, I. P., & G. J. Gaeth (1988). Framing of attribute information before and after consuming the product. Journal of Consumer Research, 15, 374-378.

Levin, I. P., & J. D. Jasper (1995). Phased narrowing: A new process tracing method for decision making. Organizational Behavior and Human Decision Processes, 64, 1-8.

Levin, I. P., & J. D. Jasper (1996). An experimental analysis of nationalistic tendencies in consumer decision processes: The case of the multi-national product. Journal of Experimental Psychology: Applied, 2, 17-30.

Levin, I. P., J. D. Jasper, J. D. Mittelstaedt, & G. J. Gaeth (1993). Attitudes toward "Buy America first" and preferences for American and Japanese cars: A different role for country-of-origin information. Advances in Consumer Research, 20, 625-629.

Levin, I. P., S. L. Schneider, G. J. Gaeth, & A. B. Conlon (1995). All frames are not created equal: A typology of valence framing effects: Working paper. University of Iowa.

Marteau, T. M. (1989). Framing of information: Its influence upon decisions of doctors and patients. British Journal of Social Psychology, 28, 89-94.

Nedungadi, P. (1990). Recall and consumer consideration sets: Influencing choices without altering brand evaluations. Journal of Consumer Research, 17, 245-253.

Newell, A., & H. A. Simon (1972). Human problem solving. Englewood Cliffs, NJ: Prentice-Hall.

Obermiller, C., & E. Spangenberg (1989). Exploring the effects of country of origin labels: An information processing framework. Advances in Consumer Research, 16, 454-459.

Payne, J. W. (1976). Task complexity and contingent processing in human decision making: An information search and protocol analysis. Organizational Behavior and Human Decision Processes, 16, 366-387.

Payne, J. W., J. R. Bettman, & E. J. Johnson (1993). The adaptive decision maker. Cambridge, England: Cambridge University Press.

Roberts, J. H., & J. M. Lattin (1991). Development and testing of a model of consideration set composition. Journal of Marketing Research, 28, 429-440.

Shimp, T. A., & S. Sharma (1987). Consumer ethnocentrism: Construction and validation of the CETSCALE. Journal of Marketing Research, 24, 280-289.

Shocker, A. D., M. Ben-Akiva, B. Boccara, & P. Nedungadi (1991). Consideration set influences on consumer decision-making and choice: Issues, models, and suggestions. Marketing Letters, 2, 181-197.

Wilson, D. K., K. A., Wallston, & J. E. King (1990). Effects of contract framing, motivation to quit, and self-efficacy on smoking reduction. Journal of Applied Social Psychology, 20, 531-547.

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