REFERENCE GROUPS, CONGRUITY THEORY AND

CROSS/CULTURAL PERSUASION*

BY E.S. LORIMOR and S. WATSON DUNN

The findings of this study in the measurement of selected middle-class attitudes in France and Egypt suggest that persuasion messages (such as advertisements) can be transferred across cultures to a greater extent than is generally supposed. It also seems probable that the transferability of messages differs from nation to nation rather than conforming to an exact formula.

** ** *

One of the many puzzling questions facing international communicators and marketers today is the extent to which they can transfer a successful domestic persuasive or promotional campaign to a different culture. Opinions on the subject range from those which support almost complete transferability [2] to those which maintain that every market is so distinct that marketers will face a host of problems if they attempt to use the same approach abroad as they have in their domestic advertisements [7, 12]. Yet there is little evidence reported to support either argument, nor have the factors which affect international communication ever been clearly identified.

The study that follows provided a test in the field of some hypotheses developed from reference group theory and a consistency model of attitude change.

BACKGROUND OF THE STUDY

Among the many variables influencing the effectiveness of most persuasive communications is the source of the message. This point has been well documented by Hovland, Weiss and many others and need not be discussed here. One of the standards by which the audience appraises the source is the reference group to which that source belongs.

Implicit in the concept of reference group is the notion of an individual deriving values and goals as well as frame of reference from the perspective of the group [4, 17, 19]. Consequently, many kinds of units with great variations in size, composition, and structure may become reference groups [19]. A person's reference group may well contain persons whom he has never seen [18], but it is important that the individuals composing a group bear a psychological relationship to each other [3]. We may therefore think of a given nationality as constituting a group. Such phenomena as jingoism, chauvinism, and wartime solidarity suggest that the citizens of a nation have a psychological relationship to each other. We speak of a nation as sharing a common culture, a concept used by Redfield (cited in 19) to refer to a perspective shared by those in a particular group. On this basis it is proposed that one's nationality group serves as a

*Excerpts from "Reference Groups, Congruity Theory and Cross-Cultural Persuasion," The Journal of Communication, XVIII (December 1968), pp. 354–368. Reprinted with the permission of the Journal of Communication, copyright holder and the courtesy of the author.

reference group for both self-evaluation and attitude formation, other things being equal. The qualifying phrase recognizes such exceptions as the individual who has learned to loathe himself and his membership group so that it may become a negative reference group (cf. 11) and such facts as that the individual has many reference groups, one of which may have more effect than another in a given situation [17].

One popular approach to the study of attitude change is Congruity Theory [13] where two objects of judgment-source and concept-and an assertion are linked. The assertion (communication) originates with the source and concerns the concept. Both source and concept may be positive, negative, or neutral, and the assertion may be positive (associative) or negative (dissociative). The basic congruity principle holds that changes in evaluation always occur in the direction of increased congruity with the prevailing frame of reference. For example, if a negative source (an habitual Drunk) says something favorable about a positive concept (Fine Wines), the attitude toward fine wines would become less favorable while the drunk would rise in the individual's estimation.

HYPOTHESES

On the basis of reference group and congruity theory, and the findings of the case history research, certain hypotheses were formulated:

1. A persuasive message is more effective when its source is viewed as a member of the recipient's nationality group than when the source is seen as a non-member, other things being equal (nationality of source).

Since the foreign model in one country was the native model in the other, and vice versa, it was predicted that:

2. A persuasive message from a source in a country to which he is native has a reverse effectiveness from one in a country to which he is foreign (between country).

It is a well-known fact that the language of a group reflects its values and interests. Obviously an outsider will not be so aware of these as a member of the group who is thoroughly at home in the language. Therefore, a third predictive hypothesis was:

3. A message composed originally in the language in which it is read is more effective than a message translated idiomatically from another language (translation).

A weakness of Congruity Theory is the failure to allow for the consequence of the intensity of the bond between source and concept (originator and object of the message, respectively). It seems reasonable that the strength of the bond is partially determined by the distance. between source and concept.

In the study under discussion, it was assumed that a message directly attributed to a source would be seen as more closely linked to that source than one which merely associated the source with the message without making an explicit attribution. For example, if Carl Sandburg is quoted

about a pen, the message is seen more as his than when his face simply appears without comment in a pen advertisement. Closer association of the source with the message should lead to closer association of the source with the object of the message. We would expect that Carl Sandburg would have a stronger association with the pen in the first instance. On the ground that a closer association between source and concept results in a stronger link, it was predicted that:

4. A message which is directly attributed to a positive source is more effective than one which is his merely by implication (attribution).

Various studies of communication effects have found that, in general, picture and verbal messages together are more effective than words alone in changing evaluative judgments (e.g., 8). A possible reason is that the pictured source reinforces or creates favorable feeling toward the source by appearing to be a member of the recipient's reference group. On the basis of an extension of congruity theory, combined with reference group theory, it was predicted that:

5. The effect of a persuasive communication is greater when a favorable source is actually portrayed with the message than when no source is portrayed regardless of whether or not the message is verbally attributed to the source (illustration).

To test the hypotheses, designs were formulated with two countries, France and Egypt; three levels of identified sources, one native and two foreign; two levels of attribution, attributed and not attributed; two levels of translation, idiomatic translation and a message composed originally in the language in which it was read; and five products.

SAMPLE

The subjects selected for study were middle- and upper-class, urban French and Egyptians. The two countries differ markedly in their degree of economic development, standard of living, and literacy level, and hence permit a test of the hypotheses under very different conditions. The sample in France was chosen by selecting one section at random from middle- and upper-class sections of Paris. After the blocks were numbered, 20 were selected randomly and 10 interviews done in each for a sample of 200. The interviewers were given no choice in the selection of respondents; however, the sample was stratified for sex and employment. In Egypt, a sample of 200 was selected randomly from a larger sample used in a refrigerator study. The original sample was obtained by listing all households in Cairo and its environs which had a special meter, indicating that they used electricty for purposes other than lighting. This original sample was judged by Egyptian authorities to contain almost all of the middle- and upper-class inhabitants of metropolitan Cairo. The respondents were approximately half male and half female. In both countries, subjects were limited to individuals 18 years of age or older. The income, education, occupation, and possessions of the sample were congruent with a middle- and upper-class population.

TEST MATERIALS

The persuasive messages used in this experiment were five print advertisements which had been run quite successfully in mass-circulated magazines in the United States. All had received high Starch recognition scores and most had produced evidence of high recall and/or sales response. All were full-page advertisements which would be effective in black-and-white, were of conventional advertising format, and depended for their impact on a combination of illustration, headline, and body text. Headlines and copy were translated by bi-lingual nationals. The second version of each headline and copy was written by professional copywriters from each country who had a copy of the original copy platform. (including advertising objective), certain background information on prospective consumers, copies of the illustrations to be used, and instructions to write appropriate heads and copy of a given length for the advertisement.

The original photographs were replicated, using French and Egyptian nationals. Great attention was devoted to making details correct, and each illustration carried as a caption a typically national name plus the nationality. Copy and headlines were attributed through the addition of quotation marks and the selected name plus a few words describing the pictured source. The finished persuasive messages consisted of 20 advertisements about each product for each country. The experimental ads were produced by an electrostatic process which provided a clear blackand-white reproduction.

Care was taken to select five products which were neither culture-bound nor too drastically different from products normally used in the selected countries. None of the products violated obvious taboos in any of the countries, and an attempt was made to avoid using a product which was associated with a particular country. For example, France is known for its perfume, on which grounds a French national might be expected to be a better perfume salesman than, say, a Turk, even to other Turks. Neutral fictitious brand names were assigned to the products in order to avoid ceiling effects and any pre-communication judgment. The judgments as to whether or not the products met the selection criteria were made by consulting communication and advertising experts in the nations to be studied.

Ten subjects in each of twenty conditions were shown one message promoting each brand for a total of 200 in each country. These conditions permitted a test of each of the hypotheses. Various measures of effectiveness were administered individually by trained native interviewers in each country. Special care was taken to avoid interviewer or order bias.

MEASURING INSTRUMENTS

Four measures of advertising effectiveness were used. As a measure of attitude toward the products, seven-step semantic differential scales [14]

were used. There is evidence for both reliability and validity of the evaluative scales used in this way [13], and variations of the method have received increasing use in testing advertising effectiveness (e.g., 10). Care was taken to select panculturally stable scales. Subjects rated each advertised product against 12 scales and against a would buy-would not buy continuum. In addition, each subject was asked to describe his corresponding ideal product on the same scales. The semantic differential scales were coded from 1 to 7 with 7 being the most favorable.

After factor analysis, four scales which loaded high and purely on the evaluative factor were selected. For both countries, the scales were good-bad, dirty-clean, effective-ineffective, and agreeable-disagreeable. The scores for these four scales were summed for each individual to give his evaluation factor score; this permitted a range of scores form 4 to 28.

Note: Means with the same alphabetical subscript are not signifcantly different from each other at the .05 level by Newman-Keuls test.

As a second measure of attitude toward the products, evaluation factor scores of the advertised brands, were subtracted from the evaluation factor scores of the corresponding ideal products, theoretically permitting a range of individual scores from -24 to +24. These scores are