Chapter 4: Marketing Research Across Cultures

In this chapter we'll consider both the methodological and practical issues in undertaking marketing and market research across cultures. Some further details of our research methods and techniques are also given in the Appendix.

Market research should be distinguished from marketing research. The former is mainly concerned with the task of quantifying the size and nature of markets and buyer behavior whereas the latter is broader and encompasses all aspects of the company's marketing activities.

THE MEASUREMENT OF MEANING

The predominant methodologies that appear in the extant literature on market research are rarely equipped to go beyond any "first order" interpretation, i.e., interpretations of the dominant culture of the researcher. Much of what cross-cultural scholars have suggested in the last 30 years has been consistently ignored. Furthermore, the dominant school of thought derives from psychology. In an attempt to be "scientific," the main conclusions seek to claim universal psychological motivations explaining the behavior of consumers despite the fact that cultures are particular (Callebout et al., 2000; 2003).

All research outcomes are dependent on the methodological stance taken and this is no less so than in international market research. While the physical scientist alone defines a set of variables to be observed , in social sciences in general and in market research in particular, the observational field usually starts with the subject's preselected and preinterpreted cultural meanings. As Cicourel so profoundly says: "The social scientist must attend to the meaning structures employed by the actors of the scene he [ sic ] wishes to observe and describe, while simultaneously translating such meaning structures into constructs consistent with his theoretical interests" (Cicourel, 1964).

Therefore precise measurement of social action, so fundamental in international market research, rests on the study of the problem of meaning and begins with reference to the commonsense world of everyday life: "Measurement presupposes a banded network of shared meanings, i.e., a theory of culture" (ibid.).

In fact the main criticism to be made of market research is that it is often methodologically flawed, even in domestic markets. It is frequently too simplistic, looking for easy answers. Rarely is there any real ontological quest for the truth in spite of the millions of dollars spent as a result of the data gathered. Even at the basic level, there is a fundamental methodological tenet that market researchers can never know the questions that should really be asked in their surveys. Most just make them up; if you knew what questions to ask, you wouldn't need to conduct the survey in the first place. When extended to international market activity, the methodological problems become even worse .

Kant, in The Critique of Pure Reason (1781), examined the bases of human knowledge and created an individual epistemology. Like earlier philosophers , Kant differentiated modes of thinking into analytic and synthetic propositions. An analytic proposition is one in which the predicate is contained in the subject so that the truth is self-evident. But he pointed out that synthetic propositions are those that cannot be arrived at by pure analysis and can only result from experience of the world. Thereby objects of themselves have no existence, and space and time exist only as part of the mind, as "intuitions" by which perceptions are measured and judged. More narrowly, within metaphysics, idealism is the view that all consumer products are mind-dependent and can have no existence apart from a mind that is conscious of them. Metaphysical realism has traditionally led to epistemological skepticism, the doctrine that knowledge of reality is impossible , and has thereby provided an important motivation for theories of idealism, which contend that reality is mind-dependent and that true knowledge of reality is gained by' relying upon a spiritual or conscious source. This may seem very theoretical, but again it reinforces our thesis that culture is a system of shared meaning. Market research must understand the system of meaning in the culture in which it is collecting and interpreting data.

Primary market research usually involves:

• Survey research: typically to judge reaction to a proposed product, awareness of brand name , or to assess attitudes.

• Observation: has the advantage of no interaction with the subject. This could involve observing whether people turn left or right when they enter a store or leave a car park-so as to help with positioning and location planning.

• Experimentation: This can involve tasting panels, or test marketing, or special promotions and discounts for limited periods.

In any survey research or in some experimentation, the observer is part of the field of action. As such, the observer and the instruments used to measure a group of actors' set of shared meaning, act as a "grid" or "filter" for the definition of certain forms of data. The world of the observable is not simply out there to be described and measured, but the observer and the measurement system can influence what is out there.

Ideally we sould seek a situation in which measurement is entirely free from error, a situation, however, which cannot be achieved (Campbell, 1928). If we only accepted "true measures," those free from error, market research would be impossible to conduct: "If the chance of error alone were the sole basis for evaluating methods of inference, then we would never reach any decision, but merely keep increasing the sample size indefinitely" (Churchman, 1948).

We can distinguish between three major sources of error that can frequently occur in international market research.

• One is related to the sampling procedure both within and between cultures.

• A second source of error is inherent to the measuring instruments used for the collection of data. It focuses in particular on the validity and reliability of instruments used in international market research. This source of error can be minimized through the design of the questionnaires and via the administration of validating interviews.

• A third source of error accompanies the administration of the measurement instrument. It manifests from the fact that the responses of the subjects under investigation are influenced by the way the study is presented to them in terms of its purpose and possible consequences.

Sampling Errors

One of the problems with cross-cultural market research is sampling. The sampling method and criteria for selection depends largely on the type of research that is being conducted (Campbell and Stanley, 1966; Berry, 1980). Two main strategies are recognized as satisfying the requirements that are intrinsic to the complex procedure of comparing market characteristics in two or more nations or cultures.

The first uses samples that are broad and representative of the culture involved (Smelser, 1976). In this case different social levels of the culture are represented and different characteristics of people are included, such as age, sex, education, occupation , etc. According to this "broad sample strategy" (Roberts, 1977), as many variables as possible are to be taken into account, because otherwise cultural influences cannot be singled out from other determinants . This type of comparison is often used in public opinion polls which, on the basis of representative samples of natural populations, try to randomize out the subcultural differences.

In international market research, one can also choose a second matching strategy. This is based on the use of narrow samples, in which information is drawn from similar subcultures in different countries . Samples are matched which are intended to be functionally equivalent (Hofstede, 1980). In other words they resemble one another sufficiently well for purposes of analysis-with respect to certain definitional criteria. The logic behind this "method of controlled comparison" is that selecting cases that resemble one another in significant respects can control for potential sources of variation (Walton, 1973).

To maximize the functional equivalence, one needs to include only a few specific types of functions in the narrow sample, e.g., professional staff or housewives. This type of matching allows one to observe the effect of the type of function on the differences found in the various other dimensions between the participating cultures (Smelser, 1976). It is, however, very important that in the sampling procedures attention is given to the fact that the specific functions and positions of the group involved is analyzed in the context of the larger society in which they operate . Too many easy assumptions have been made about the comparability of functionally equivalent samples; for example, the role and status of working women is quite different in different cultures. This might have a significant influence on the conclusions drawn from the results of inquiry. Note that in some cultures, the matching sample may not even exist. How could we compare the reasons for buying certain things expressed by senior female managers in the west with the reasons given by senior female managers in the United Arab Emirates? The equivalent role just is not there.

Research Instruments in International Market Research

Even national market research itself has many complexities. It is all aimed at reading the mind of the ( prospective ) client: a difficult job that increases dramatically when you try to go beyond national boundaries. Since the simultaneous launch of products and services is becoming increasingly popular, market research across cultures is becoming more frequent. And cross-cultural differences in this type of research cause similar dilemmas to the ones a researcher is facing in general cross-cultural research. Most dilemmas occur in the area of validity and reliability of the instruments in use - ranging from interviews to questionnaires. In-depth, unstructured, interactionist interviews may give results that are valid (that is, closer to the truth for the individual interviewee) but not reliable; that is to say that the next interviewee will be likely to provide a different range of insights. Survey questionnaires may produce results that are consistent (i.e., reliable) but that are not necessarily true. Thus, as considered below, the researcher is also faced with a fundamental dilemma - to reconcile truth with generalizability.

Cross-Cultural Validity

Validity has generally been defined as the quality that indicates the degree to which an instrument measures the construct under investigation (Bohrnstedt, 1970). To achieve this you first need to develop an insight into the cross-cultural environment before being able to formulate the objectives of research. For example, multi-local research raises some quite different challenges from global market research. In the former you should focus on deciphering the local particularities of the product in the mind of the local user , while in a global environment the focus is much more on finding similarities in users and products. The objective of the research also changes dramatically, whether you want to find out what kind of advertising campaign needs to be launched or what kind of features the product should offer. Obviously the complexity increases significantly once cultural boundaries are crossed in all areas of validity, in particular when socio-cultural and psychological factors influence the attitude of a particular group.

We can distinguish four levels of cross-cultural validity (Douglas and Craig, 1983):

• construct validity

• content validity

• sample validity, and

• instrument validity

Let's examine the first two in some detail.

Construct Validity

This first type of validity, according to the American Psychological Association, "is evaluated by investigating what qualities a test measures, that is, by determining the degree to which certain explanatory concepts or constructs account for performance on the test." It validates the theory underlying the instruments constructed . In fact what is needed are several independent measures of one concept, one construct. In international market research construct validity becomes of utmost importance, in particular when it concerns the validity of concepts: The comparison of concepts is crucial in cross-cultural market research. Even if you try to measure and quantify basic concepts such as friendliness, cleanliness, sex appeal , and (dis)satisfaction, you can observe quite different ways in which they are expressed in behavior and experienced by the individual or group.

It is, for example, well known that complaints about a product or service express a feeling of dissatisfaction in almost all cultures. But in some cultures a complaint is only given in the context of a relationship that needs improvement. In other words, the producer gets another chance. The logic is one where you write down clearly what went wrong so that the next time you'll get better service. In other cultures, on the contrary, a complaint is like a farewell letter, something along the lines of: "Let me tell you why I will never buy your products or services again."

Another area of validity is the one around function. The equivalence of function is very much overestimated in global market research. According to Frijda and Jahoda (1966) you cannot compare any measurement when similar activities or products play different functions in different societies . Quite often similar functions are assumed across cultures in international research by monocultural research teams . The function of an automobile is quite different across cultures. In some it is purely a mode of transportation while in others status plays a dominant role or it is seen as a vehicle to tour around in. A watch is either predominantly an instrument used to read time, a collector's item, or a fashionable attribute.

Another construct validity check is whether measurement equivalence has been attained. The very popular Likert scale, for example, gives quite good discrimination within a culture when you score 23% very good, 17% good, 11% moderate, 42% bad and 7% very bad. In France and Germany people tend to score less extreme on scales (Pras and Angelmar, 1978), which could partly be related to differences in the meaning of translated words. In general, however, there is also evidence that in certain cultures, the extremes (very good or very bad) are used only in highly exceptional cases. Research has also shown that US Americans have a tendency to score significantly more positively than their European counterparts. This viewpoint is confirmed in the analysis of THT's workshop reviews over the past 20 years. The confirmation holds true to both monocultural and international sets of participants attending similar workshops. Although the use of scales is notorious across cultures, it can still be applied in line with the pretested instruments available in the past. The problems frequently arise because there are different interpretations of what "strongly (dis)agree" means, for example, or how respondents seem to be repelled or attracted by the word "undecided."

Different constructs of quality

Most American companies have adopted the Japanese attitude of doing things right the first time. However while Americans pay lip service to the Japanese concept of first-time perfection or zero defects, experience shows that they actually don't want to do it right the first time. Rather there is an attitude based on "no pain, no gain" - you have to learn from your mistakes.

In the auto industry, for example, General Motors found out that selling "perfect" cars with no mistakes decreased the appreciation of their customers significantly. Big jumps in GM's image were made when the small number of mistakes were treated adequately and quickly: "What a great company. When you have a little problem with your car they pick it up, replace it with a new one and the next evening it is brought back. Fixed."The message is clear; the people who had a problem with their cars had a chance to experience how much GM cared..

In a study within AT&T it was revealed that the phrase "total quality control" appealed to Japanese workers, but was deadly for the Americans. None of the managers was interested, so they were asked what they were doing wrong. The program was redesigned, giving managers tasks to do, and they were videotaped failing these tasks . The people who came in talking about "doing it right the first time" and "zero defects" failed again and again, but then learned from their mistakes.

In America "zero defects" means perfection. For the Japanese perfection is attainable; for the Americans, only God can accomplish perfection. The three words ''total quality control" represent the most negative combination possible to the American unconscious. So when AT&T stopped copying the Japanese and starting seeking quality the American way - through trial and error - success was imminent, The training program was devised around quality and started with planned failure. Whereas the Japanese would give managers a rule book to study in order to achieve perfection, AT&T developed a process in which initial failure was built in, so the managers could learn through trial and error to create quality that they would then view as a personal accomplishment.

Companies should not purposefully make products with defects so that they can show customers they care, but they do need to consider another attribute in addition to the quality of the products or service: the quality of the relationship between the product and the customer, ^[1]

 

One way to avoid the problems of differences across cultural scale interpretations is to use forced-choice questionnaires. The standardized questions of a fixed-choice questionnaire can furthermore be designed to incorporate the language and cultural meanings inherent in the respondents' perspectives on daily life and in the questioner's own perspective. By using this type of questionnaire you aim at providing solution types of responses in the different cultural settings related to the "inner" attitudinal states of the (group of) people. In fact, such a use of standardized questions with fixed-choice answers, as Cicourel remarks, provides an empirical solution to the problem of meaning by simply avoiding it. The claim that the "inner" states or "meaning structures" somewhat correspond with the actual response patterns can thus be supported by empirical evidence.

From the outset, however, one needs to realize the limitations of any device which is used to measure meaning in market research, including the forced-choice questionnaire. Note that the "forced" character of the responses restricts the possibility that people's perception and interpretation of the items will be problematic and which, by a static conception of role-taking, does not eliminate the problem of situational definition.

An advantage of this type of survey is that it leads itself to translation into numerical representations. Not many more alternatives are open to overcome these limitations than to minimize the number of possible errors of the measurements of meaning. Any type of questionnaire, however, is an instrument with which the observer communicates with the observed. One other important step, therefore, is the careful analysis of the use of language as the symbolic representation of meaning structures.

Language

Many of the instruments developed for researching the market are in English. The dominance of the Anglo-Saxon world in business and marketing makes problems of translation notorious.

The medium of language acts as a "grid" for defining and letting certain forms of information through to the observer. The error that is caused by this "filter," however, can be reduced. In the process of measuring meaning in international market research it is important that both questions and responses need to reflect the respondents' daily world and be couched in the everyday language that they are familiar with. The evoked answers should not be altered by the particular relevance structures under investigation. Therefore, in the construction of a questionnaire you need to take it as fundamental to use commonsense terms as used in everyday life. These might be quite different in different cultures. Obviously in the case of international research, with questions being presented in a variety of cultural settings, the use of this type of everyday language is even more important.

Standard questions administered in different cultural settings might run the risk of being interpreted differently by virtue of linguistic difference (Evin and Bower, 1952). The use of language, however, is not only important in the process of translation. It starts to be important in the design of the original instruments. Is the language carrying the meaning it purports to carry? Is it stated in such a way that it does not allow for various interpretations in different cultures? Does it express situations which might provoke defensive responsiveness?

There are a variety of ways in which major problems of misinterpretation can be avoided. In particular, in high-context cultures the use of short questions or reactions to a word or concept can be very unreliable. The use of stories is a good way of communicating with the respondents, because people tend to think in stories. Bateson (1980) states the case for the primacy of representatives:

There's a story which I have used before and shall use again: a man wanted to know about mind, not in nature, but in his private large computer. He asked it "Do you compute that you will ever think like a human being?" The machine then set to work to analyze its own computational habits. Finally, the machine printed its answer on a piece of paper, as such machines do. The man ran to get the answer and found, neatly typed, the words: THAT REMINDS ME OF A STORY. A story is a little knot or complex of that species of connectedness which we call relevance. Surely the computer was right. This is indeed how people think.

In the formulation of stories abstract words need to be consistently moved to concrete areas. In doing so the difficulty of achieving equivalent meanings diminishes. In addition, in the design of alternative life situations, it is advised to use those situations that are similarly basic and significant to the potential samples you want to investigate. In doing so the likelihood that a forthcoming translation of it from one language to another will result in a serious alternation of the intended meaning is lessened (Hofstede, 1980). You also need to find samples where the subjects under investigation perform similar activities - thereby sharing common problem areas - which decreases the complexity of the process of achieving this format.

When stories are gathered through questionnaires or through interviews many techniques are available to analyze the meaning of things said. There is software available (words in context) that can sort the essential meanings of what people were trying to say. This software looks at the combination of words and counts the number of words that arise. Obviously lots of extra effort still needs to be put into getting the meaning you want to examine. A multicultural set of analysers can do wonders, in particular by discussing the differences in findings.

Translation

Once these aspects of language are taken into account in your formulations, careful attention needs to be paid to the translation of the questionnaire. Translation problems are notorious. The object of translation - being an important source of error - is to achieve equivalence of meaning, resulting in equivalent interpretations across cultures. A careful translation, followed by a "back-translation" or synchronized by a "parallel translation" does not necessarily dispose of all difficulties (Duijker and Rokkan, 1954). Even between European languages, literal equivalence may result in different connotations , sometimes to the extent of rendering scale items meaningless (Frijda and Jahoda, 1969; Mayer, 1978). Therefore the purpose of translation is to achieve equivalence of concepts and not words. In some cases, however, the utilization of different items of similar intent cannot be avoided, which in turn entails complications when it comes to interpretation. But in types of cross-cultural market research an intuitive judgement is often worth more than an objective standardization (Satori, 1970).

Emic and Etic

In the comparative study of societies two approaches are commonly used: the "unique approach" or nomothetic type of study, and the " comparative approach" or ideographic type of study (Frijda and Jahoda, 1969). The choice between the nomothetic or ideographic study approach needs to be made carefully because its specific mode of explanation, its mode of organizing variables, and the research techniques employed needs to best fit the subject matter.

Since Sapir looked at the different types of cross-cultural research in 1932, the discussion around the uniqueness of cultures has not disappeared. On the one hand, there is the "emic" approach, which believes that any culture has an attitude and behavior that is unique to the group. On the other hand is the "etic" approach, focusing on the search for universal attitudes and behaviors across cultures. Special attention, therefore, needs to be paid to finding concepts that are not so peculiar to a single culture or group of cultures that no instance of the concept can be found in other cultures. The comparability of concepts is improved by simultaneously making them more abstract and inclusive.

In this process of climbing and descending the ladder of abstraction, we face a continuous struggle between the "culture-boundedness" of system-specific categories and the "contentlessness" of system-inclusive categories.

So, in addition, in order to achieve the maximum possible equivalence of concepts translators must be as familiar with the context of the material to be translated as they are with the languages concerned. Despite these precautions many unexpected problems may still arise and translations often need to be anthropological rather than literal.

Before presenting any translation to the participating subjects of the research, versions need to be checked by in-culture people or, even better, by professional local researchers. The various versions of all instruments need to be checked on whether all items are conceptually equivalent and well- adapted to the cultures under survey. Back translations may be used, as also may stories and "thick descriptions" rather than short statements, which are more sensitive to a translation of meaning.

Content Validity

This type of validity refers to the degree to which the instrument being used represents the concept about which generalizations are to be made. The first step that needs to be made in market research is to optimize the content validity of the instruments by searching the international literature carefully in order to determine how various authors have used the concept (Bohrnstedt, 1970). In this procedure special attention needs to be paid to capturing the various shades of meaning of the concepts under investigation. Series of items need to be constructed, which include each of the substrata of the domain of content, a procedure that the literature refers to as "sampling from a domain of content" (Blalok, 1964). Instruments that measure multidimensional concepts need to be constructed in such a way that those items tap into the subtleties of meaning within each of its dimensions. This exercise is an intellectual version of cluster analysis, as used in statistics.

A final check on validity could be attained by conducting validating interviews using a pair of interviewers from different cultural backgrounds. This needs to be done in order to capture as many shades of meaning as possible.

^[1] Source: Adapted from Rapaille (2001).