Preferences of Nationalists and Assimilationists For Ethnic Goods: an Experiment With French-Canadians

Jean M. Lefebvre, University of Hartford
ABSTRACT - This research, building on past studies of ethnic minority consumer behavior, briefly presents a typology of individual minority responses. They are posited to mediate the hypothesized preferences of ethnic minority individuals (MI) for ethnic goods. Two main contrasted types are clearly recovered empirically: nationalists and assimilationists. The later do tend to prefer EC (English-Canadian) to FC (French-Canadian) goods while the former tend, with some ambiguity, to prefer FC to EC products. It is suggested that future research should define and inventory a repertoire of ethnic products, situations and behaviors.
[ to cite ]:
Jean M. Lefebvre (1987) ,"Preferences of Nationalists and Assimilationists For Ethnic Goods: an Experiment With French-Canadians", in NA - Advances in Consumer Research Volume 14, eds. Melanie Wallendorf and Paul Anderson, Provo, UT : Association for Consumer Research, Pages: 497-501.

Advances in Consumer Research Volume 14, 1987      Pages 497-501


Jean M. Lefebvre, University of Hartford


This research, building on past studies of ethnic minority consumer behavior, briefly presents a typology of individual minority responses. They are posited to mediate the hypothesized preferences of ethnic minority individuals (MI) for ethnic goods. Two main contrasted types are clearly recovered empirically: nationalists and assimilationists. The later do tend to prefer EC (English-Canadian) to FC (French-Canadian) goods while the former tend, with some ambiguity, to prefer FC to EC products. It is suggested that future research should define and inventory a repertoire of ethnic products, situations and behaviors.


Research on the consumption behavior of ethnic minorities is important for several reasons. First, both ethnicity (Hirschman 1981, 1983, 1982) and, often overlapping, minority group membership appear to have profound effects on consumer behavior (Sturdivant 1973). It remains a major consumer characteristic even after controlling for socio-economic factors (Robertson, Zielinski and Ward 1984; Schaninger, Bourgeois and Buss 1985).

Second, ethnic minorities live in and are spread across a great many nations around the globe (Runt and Walker 1979). Knowledge about the impact of ethnic minority subculture on consumer behavior could provide guidance to international marketers who operate in countries and world regions divided along ethnic lines.

Third, the similarity of research issues identified in the study of, mostly, Black and Hispanic subcultures in North-America could extend to all ethnic minorities worldwide. They do not constitute a negligible part of the world population.

One of the key issues, yet to be resolved, (Lefebvre 1975; Mallen 1973) concerns the dichotomy in subcultural motivations: "... the preference for cultural integration versus the preference for cultural distinctiveness." (Robertson, Zielinski and Ward 1984, p. 539) as they relate to consumer behavior. The present study focuses on this issue, framed in the Canadian context. FC society is another North-American ethnic minority subculture worthy of attention by its' size and distinctiveness (Engel, Kollat and blackwell 1978). Briefly stated, we attempt to answer the following question: "Do FC nationalists prefer FC to EC products? Do FC assimilationists prefer EC to FC products?"


The framework conceptualizes FCs as members of an heterogeneous ethnic minority subculture. Heterogeneity stems from the various patterns of adjustment of each member to his minority status - minority responses ranging from nationalist to assimilationist. The different types of individual adjustments are posited to mediate their preferences for ethnic goods. Ethnicity, being a common dimension to the person and the choice object should provide a good anchor to preferences for ethnic goods. Ethnicity, being a common dimension to the person and the choice object should provide a good anchor to preferences for ethnic goods.

A Typology of Individual Minority Responses

Minority response refers to the way in which a minority individual (HI) adjusts socio- psychologically to his disadvantageous situation ascribed to him as member of a MG. The MG situation can be viewed as the antecedents, and the minority response as the consequence of the situation. Some of the key elements of the situation, are: 1) the trait or symbol that sets MIs apart from the DG - its' visibility and stability over time (For FCs, the French language nd, less visibly, catholicism; Johnstone 1966), 2) the nature of the relationship between DG and MG - relative power and extent of the differences between the two groups, frequency and kind of contacts between MG and DG persons ant, 3) the severity of discrimination and prejudice from the DG (Simpson and Yinger 1985). In the response per se, the MI adjusts in terms of 1) ego- involvement (Rollins 1973) and direction of his ethnic identifications with his own group (Caplan 1970, Noel 1964, Parker and Kleiner 1964, Plax 1972) and with the DG (Frazier 1957) the two key subcultural motivations of group distinctiveness versus integration - as well as 2) his desired goals for the MG in its' relations with the DG, ranging from separation from to assimilation into the DG (Gordon 1964, Kurokaw 1970, Shemerhorn 1970). Clearly, the adjustment of a MI to his disadvantageous status will contain both his perceptions of the situation together with the key dimensions of his response per se. Conceived as a momentary state of the individual, and not as a dynamic process evolving over time where situation and response would feed back on one another, MI response will be taken to e an both the MI's perceptions of the situation and his response per se.

No two MIs will respond in exactly the same way. However it is possible to classify them among various analytical types. The nationalist feels strongly about his ethnicity. He considers himself similar to FCs and identifies positively with FCs. He feels dissimilar to ECs, identifies negatively with ECs and perceives ECs to differ from FCs. For him the goal of FC society is to separate from EC society. He opposes assimilation. The assimilationist stands in complete symmetry to the nationalist. He wants to abandon his own ethnicity and blend into the dominant group thereby adopting its values, life style and language. Eventually, if the DG concurs with no discrimination and prejudice, the MG will disappear, merged into the wider society. Nationalists and assimilationists should form the majority of subjects from a FC sample (1).


They flow directly from the typology.

H1: Subjects will be classifiable as nationalists or assimilationists, predominantly.

H2: The assimilationist prefers EC products to FC products.

H3: The nationalist prefers FC products to EC products.

The Choice of Ethnic Products

An ethnic product would be perceived, in the present context, by FC consumers along a salient attribute of French to English-Canadianness The products chosen were presumed ethnic by:

1. their ethnic symbolic meanings - eight kinds of woolen hats bearing clearly ethnic symbols, from FC (Quebec flag, "fleur de lys") to EC (Canadian flag, maple leaf) via neutral symbols (stripes or stars) These various kinds of hats were currently worn by students on campus during the winter

2. their brand names - eight brands of cigarettes, from FC ("La Quebecoise") to EC ("Export A", "Players") "La Quebecoise" had been recently introduced and aggressively targeted to the separatists

3. different levels of ethnic cues attached to "Calmine" - a fictitious brand of analgesic - and "Presto" - a fictitious brand of instant coffee - as the different treatment levels in an experiment attempting to simulate the effects of a radio advertising campaign The ethnic cues, and treatment levels, were manipulated via a) the name and location of the manufacturing firms - from EC, "Toronto Laboratories, Inc " manufacturing "Calmine" and "Nutritex of Ontario, Ltd " manufacturing "Presto", to FC "Les Laboratoires de Montreal, Inc " and "Nutrition du Quebec, Ltd" manufacturing "Calmine" and "Presto" respectively, and via b) the accent assumed by the announcer in French language ratio ads for "Calmine" and "Presto" - alternatively EC, international French and popular FC (2)



The sample consisted of 194 students - 2/3 full-time and 1/3 part-time - enrolled at the School of Business at Universite Laval in Quebec City

They form a cohort, appropriate for the testing of a theoretical framework (Hirschman 1981) A recent business school pool hat shown students were just about split on the separatist issue, thus ensuring variability on a key independent variable No external validity of the results may be claimed

Participation was voluntary Only those subjects who participated throughout the experiment, were naive with respect to its purpose and were of FC culture, have been kept in the sample As compared to the Quebec population at large, the sample is upscale in income, younger and more separatist.

(1) Three peripheral analytical types should not appear in great numbers from a sample of FCs For the pluralist, ethnicity is not a salient matter He does not feel strong preferences for one group over the other The two groups should remain distinct, equal and mutually tolerant He favors neither assimilation nor separation The alienated, downrating both his salient ethnic identifications, would live in self-hatred The marginal, forced to stay in a group he would like to leave - MG - and kept outside a group he wants to integrate - DG -, remains at the margin " poised in psychological uncertainty " (Stonequist 1957) between the two societies If FC society forms an ethnic minority subculture, it never hat to endure as severe a disadvantageous position in Canada as that of Blacks in the US, the object of much of the literature upon which the typology is based


In order to portray the types of MI responses, from nationalists to assimilationists, we need a set of independent variables to assess the subjects' perceptions of their minority situation and response per se The extent of differences between EC and FC group members as well as prejudice are key elements of the situation, as perceived by FIIs It materializes in the stereotypes " [a] community shared belief reflecting the image one group holds of another" (Gardner, Rodensky and Kirby 1970) On the basis of research conducted on Blacks and Whites in the United-States (Simpson and Yinger 1985) and on FCs and ECs in Canada (Taylor, Bassili and About 1973, Taylor, Simard and Aboud 1972), on the average, differences in stereotypes should exist and be favorable for the DG while derogatory for the MG, as perceived by MIs Stereotypes of ECs and FCs were measured along 10 bipolar adjectives from studies of stereotypes in Canada and 29 life style statements from a comparative proprietary psychographic study (Bell Canada) Five adjectives and five life style statements exhibited the largest degree of difference between the two groups Compared to their EC counterparts, on the average, FCs perceived themselves as more timid, submissive, follower, nervous, naive, afraid of the future, belonging to less clubs, less self-confident, regretting the good old times more often and, disliking to take risks These 10 most contrasted dimensions were kept to compute an index of profile difference/similarity - euclidean distance - between ECs and FCs

To measure the HI response per se, we operationalize ethnic identification in two different ways, assess ego-involvement with FC ethnicity via latitude of non-commitment and, last, attitudes toward the MG goals in its relation with the DG assimilation and separation Two main reasons justify the use of two different methods of measurement of ethnic identification One, it is a key element of the MI response per se Two, evidence from studies of stereotypes in Canada strongly suggest that derogatory images of FCs by FC subjects filter through only indirect measures (Lambert et al 1960) Both are emic measures, the appropriate type, as cogently argued by Hirschman (1981)

First, ethnic identification with (or against) both EC and FC is inferred from a profile similarity index euclidean distance - between self-EC then self-FC Profile similarity would mean positive identification, while profile difference would reflect negative identification Second, ethnic identification was gauged by a direct measure of attitude toward FC and EC borrowing the scale from Dutta, Norman and Kanungo (1969) On the average, the sample was slightly more favorable to FC than EC Latitude of non-commitment provided a basis for estimating ego-involvement with attitude toward FC (Larimer 1968)

Attitudes toward the goals of the minority assimilation and separation - were recovered via a factor analysis of 24 Likert type items designed for that purpose. A third significant factor tapped attitudes toward the survival of FC culture and has been kept to portray Ml responses

Finally, the dependent variables, preferences for ethnic products, were measured by a paired comparison technique, the dollar-metric, (Pessemier et al 1971) This technique produces individual interval preference scales appropriate for the analysis All scales, their sources and their estimated reliabilities appear in Table 1

[Thank you to Robert Tamilia for sharing his radio ads.]



Experimental Design and Data Collection

Hypotheses 2 an 3 were tested correlatively for woolen hats and brands of cigarettes. The ethnicity of the fictitious brands of analgesic "Calmine" and of instant coffee "Presto" were manipulated in an experiment. Presumably, if the brand is presented to subjects as manufactured by a firm located in Ontario, bearing an English name and sponsoring a radio ad in which the announcer assumes an EC accent, it would be perceived as an EC good. At the other extreme if the brand is manufactured by a firm located in Quebec, bearing a French name and sponsoring a radio ad where the announcer assumes a popular FC accent, it would be perceived as an FC good . Initially six different treatment combinations were planned (1) but we ended with only 4: (2 company names, EC and FC) x (2 types of accent in the radio ad, international French and popular FC). The radio ads were very realistic and of professional quality. They differed only by the accent assumed by the same announcer. To control for pre-measurement effects we used a Solomon eight group design. Four groups - classes - were submitted to premeasures of product preferences and independent variables. The next week, they listened to two radio ads, played on a tape recorder. They listened first to the ad for "Presto" and expressed their consequent preferences for brands of instant coffee, then they listened to an d for "Calmine" and expressed their preferences for brands of analgesics. In each class the same level of accent was used for both radio ads. Each class was also split between the students receiving EC or FC company names as the sponsor of the ads. It was more convenient to assign classes randomly to treatment combinations - essentially, level of accent - than individuals. Another four post-only groups - classes were subjected to the same four treatment combinations. A control group filled out all questionnaires without being exposed to ny treatment.

(1) In pretests, it clearly appeared that brands of cigarette were not perceive along ethnic dimensions (woolen hats were) and that the EC accent was perceived as anything but EC. Thus, they were dropped from the analysis

Analysis and Results

Test of Hypothesis 1: Grouping the Respondents by Types

Similarity coefficients, Cattell's Rp, were computed between the profiles of each pair of respondent on the nine independent variables listed in Table 1. Cattell's Rp behaves as a correlation coefficient taking on a value of a when both profiles completely overlap and of -1 when they are symmetrically opposed (Cattell, Coulter and Tsujioka 1966). The resulting matrix of coefficients was submitted to a principal component analysis and a varimax rotation. A three factor solution was retained as it uncovered five types of respondents, the number expected a-priori: two central types, assimilationists and nationalists plus three peripheral types. In order to partially test the validity of the types, the sample was split into two random halves and the analysis repeated on each. A multivariate profile comparison test (Allaire, Silk and Tsang 1973) showed that the average profiles of each pair of equivalent types inferred from each sample half were parallel and at the same level. Further, a discriminant analysis (Johnson 1971) shows the positions of the respondents, that of the independent variables as vectors and that of the group centroids within the axes of two first linear discriminant functions (Figure A). The first two and only significant functions may be interpreted as the two key dimensions of ethnic identifications with FC and EC. The majority of respondents were thus classified: 169 out of 194.



As expected, the two central types account for the great majority of the respondents, 140 out of 169. The mean profiles of the central types appear in Figure B. The assimilationist (type 2, N - 64), as compared to the other types, shows the least favorable attitude toward separatism, FC culture ad FC but the most favorable attitude toward EC. He perceives some difference between EC, FC and self but, contrary to expectations, not differentially. The nationalist, (type 3, N = 19) holds the most favorable attitude toward FC and the most unfavorable attitude toward EC and perceives the most difference between EC and FC. To him, FC ethnicity is ego-involving. He appears to be the extreme nationalist most favorable to separation, FC culture and most opposed to assimilation. The mild nationalist (type 1, N - 57) resembles the separatist's profile closely but remains less extreme on all dimensions except for the highest degree of ego-involvement. Thus, hypothesis 1 receives strong support.



Test of Hypotheses 2 and 3

Examining the preferences of the three types for woolen hats, hypotheses 2 and 3 receive support. Figure C illustrates the profiles of preferences of the nationalist, mild nationalist and the assimilationist. A series of univariate F-tests were all significant at the .05 level except for the difference in preferences of the assimilationist and the mild nationalist for "fleur de lys". Thus H2 and H3 are supported by the results, clearly for the nationalist as contrasted to the assimilationist.



Examining the effects of EC/FC company names and the international French/Popular FC accents assumed by the announcer in the radio ads on the preferences for "Calmine" and "Presto", fictitious brands of analgesics and instant coffee, respectively, called for ANOVAs. As there is no reason to believe that the factors affected the cell sizes, a fixed effects-unequal cells model was used (Winer 1971). Two series of six ANOVAs (3 types, assimilationist, nationalist and mild nationalist x 2 products, "Calmine" and "Presto") were run, first on the differences between the pre and post treatment measures of preferences for "Calmine" and "Presto", second on the post only preference measures. Of the 12 ANOVAs, only five showed an effect, main or interaction, significant at or beyond the .10 level.

For the nationalists there is a significant main effect (.01 level) of company names on their post only preference scores for "Calmine" in the hypothesized direction: they prefer "Calmine" more when manufactured by an FC than by an EC company. In contrast, as expected, there is a main effect of accent (.005 level) for the assimilationists who change their preference for "Calmine" positively when advertised with an international French accent and, negatively when advertised with a popular French accent.

We would expect the mild nationalists to resemble the nationalists in their preferences for ethnic goods. They do not. There is a main effect of accent on the pre/post treatment change in preferences (.07 level) and an the post-only preference (.07 level), both $or"Calmine". The direction of the effects, however, is in the direction Opposed to expectation: the international French accent entails increased and positive preference for "Calmine" as compared to the popular FC accent. Further, there is a significant interaction effect of company names and accents on the post-only preferences for "Presto". Looking at the simple effects (both significant at the .01 level); a) with the EC company name, the mild nationalists prefer Presto" advertised with the international as compared to the popular FC accent; b) with the FC company name, they prefer "Presto" advertised with the popular FC as compared to the international French accent. It is as if the mild nationalist expects the FC company to 'speak" popular FC (otherwise it would be perceived as a turncoat?) and the EC company to "speak" international french and not popular FC (otherwise intention to persuade would become obvious?). Thus, hypothesis 2 and 3 receive further, though weaker, support from the experimental part of the study.

That many expected effects did not occur could be due to a weakness in the experimental design. There are only 19 nationalists in the sample. They are spread too thinly among classes and therefore treatment combinations for many significant effects to emerge in the analysis. However, this argument cannot explain away the quasi absence of significant effects among the 64 assimilationists in the sample or the ambivalent but frequent results for the 57 mild nationalists. It could be that nationalists, more involved than assimilationists with their ethnicity, are more apt to perceive ethnic cues and react to them. Further, most of the significant effects are the result of the accents, even though they did not differ much. Could it be that FCs are more sensitive to language styles, the "trait that sets them apart", than to company origin?

Summary of Findings

In summary, the two central types - nationalists/milt nationalists and the opposite assimilationists - are recovered empirically, as hypothesized. They account for 140 out of the 169 subjects classified. Hypothesis 1 receives strong support. For the woolen hats, nationalists/milt nationalists do prefer FC to EC symbols, as expected. In their reactions to the company names and accents of ratio ads for "Presto" and "Calmine", the nationalists prefer FC to EC company names and the assimilationist prefers the international French to the popular FC accent. Some effects expected of the assimilationists to not appear empirically. Contrary to expectation, the mild nationalists downrate the popular FC as compared to the international French accent. Thus, hypotheses 2 and 3 receive support but not as strong as for hypothesis l.


First, the findings suggest that limited conditions, the FC market may be fruitfully segmented between nationalists and assimilationists when ethnic symbols are either present in, or attached to, products via advertising. Several studies provide evidence for the equivalent opposed two types among other minorities: Blacks (Bauer, Cunningham and Wortzel 1965, Bauer and Cunningham 1970), Hispanics (O'Guinn and Faber 1985), Chinese (Ellis et. al. 1985) and Jews (Hirschman 1981). A two type typology, consisting of the two central types, may be sufficient to depict most Ml responses. This finding may be valid across minorities.

Second, it follows that in comparative studies which seek to attribute differences in consumption patterns between two ethnic groups to their differing ethnicity, the researcher will need to adopt a heterogeneous view of both the MG and the DG. Besides other confounding factors (e. g., family size, income, social class), the investigator should control for the ethnic identifications of his respondents. For instance, a sample of FCs made up exclusively of assimilationists would appear, falsely, to resemble the sample of ECs to which it is compared. Symmetrically, some ECs are not immune to adopting some FC ways (McDonald 1971).

Last, if the typology has received some measure of cross-subcultural validity, it remains of limited usefulness as a construct mediating the preferences of MIs for ethnic goods and consequently as a basis for segmentation of ethnic MG. If some, and only some, of the present findings are statistically significant, they do not reach the threshold of being managerially actionable. This may be due to shortcomings of the study. We focus on those which bear suggestions for future research.

Suggestions for Future Research

The model tested here requires modifications and further testing. First, it presumes the stability of each type mediating stable - isomorphic preferences for ethnic goods. Are MI responses, or ethnic identifications, stable individual predispositions expressed regularly across many roles? Or, on the contrary, are they driven by circumstances? For instance, French-canadianness would be freely exhibited at home while English-canadienness would be channelled to the realm of work? Alternatively, as dual ethnic identifications with DG and MG may be a source of conflict for many MIs, this conflict may result in unstable preferences or avoidance of choice objects involving ethnic symbols. Negative identification may provide an unclear prescription for overt behaviors. The multiplicity of identifications of a HI may include other non-ethnic subcultures. How would identification with the youth subculture interact with ethnicity? These are questions for future research.

More radically, some argue that ethnicity has but a negligible influence on consumer behavior when compared to environmental circumstances (Thorelli 1985) while others contend that it has a pervasive influence (Hirschman 1981). Rather than take position for one of the two extremes, future research needs to identify the circumstances, obviously restricted, under which ethnic identifications, or Ml responses, will have a major influence on which aspects of consumer behavior. There is evidence that ethnicity may influence other mediating processes (Hirschman 1982, Muller 1985). The present study shows it influences preferences for ethnic goods. The evidence suggests some of the circumstances under which MIs responses could prove to be a major factor in the choice of ethnic goods. Ethnic goods would be socially consumed, exhibit shared symbolic ethnic meanings, all competing brands would perform equally well on all other important attributes (woolen hats; the cigarette brand "La Quebecoise" failed because its taste displeased smokers), non utilitarian products, (Hirschman 1986), for which tangible aspects do not dominate (the results were better for woolen hats than the analgesic), would have been the object of successful ethnic advertising (it was not the case for analgesics or instant coffees) nd, last, perceived even more so ethnic by individuals ego-involved with their ethnicity. Beyond the results of this study, some "goods" may be ethnic even in their core benefits (voting choices) and tangible attributes (package written in French). Before looking further into the determinants of expressing identification with the MG an/or the DG, future research needs to circumscribe and inventory the aspects of consumer behavior, dependent variables, that are saliently ethnic. Toward that goal, the present results outline an initial definition of ethnic goods.


Reference available upon request from the author.