Social Identity and the Meaning of Fashion Brands

Susan Auty, Lancaster University, U.K.
Richard Elliot, Oxford University, U.K.
ABSTRACT - This study considers the importance of social identity in the interpretation of brands of jeans as measured by Snyder’s Revised Self-Monitoring Scale, which discriminates between people who are highly motivated to respond to social cues and those who remain 'true to themselves’. It was found that self-monitoring is a significant mediator of meaning with regard to unbranded, but not branded, jeans. A model of choice by elimination of the unacceptable is suggested by high self-monitoring responses. It has implications for the amount of advertising required to support a fashion brand.
[ to cite ]:
Susan Auty and Richard Elliot (1998) ,"Social Identity and the Meaning of Fashion Brands", in E - European Advances in Consumer Research Volume 3, eds. Basil G. Englis and Anna Olofsson, Provo, UT : Association for Consumer Research, Pages: 1-10.

European Advances in Consumer Research Volume 3, 1998      Pages 1-10

SOCIAL IDENTITY AND THE MEANING OF FASHION BRANDS

Susan Auty, Lancaster University, U.K.

Richard Elliot, Oxford University, U.K.

ABSTRACT -

This study considers the importance of social identity in the interpretation of brands of jeans as measured by Snyder’s Revised Self-Monitoring Scale, which discriminates between people who are highly motivated to respond to social cues and those who remain 'true to themselves’. It was found that self-monitoring is a significant mediator of meaning with regard to unbranded, but not branded, jeans. A model of choice by elimination of the unacceptable is suggested by high self-monitoring responses. It has implications for the amount of advertising required to support a fashion brand.

INTRODUCTION

Diverse theorists have demonstrated the use of clothing as a code, a language which allows a message to be created and (selectively) understood (McCracken and Roth 1989; Holman, 1981; Hollander, 1978). Perceptions of brand-users have been found to differ for nearly identical brands within a product category (Swartz, 1983). The consumer in the street confirms the existence of a clothing code and indicates the importance of branding to the code: 'If I’m wearing a white T-shirt and sneakers, that label [Armani] will fill in the rest of the information about me. I want to give out the right impression’, says a customer in a London shop selecting her tenth pair of Armani jeans (Financial Times, 1995).

Since this consumer’s next door neighbour is a 'Moschino person’, it is clear that not everyone wants to give out the same 'right’ impression nor reads the label in the same way. McCracken and Roth (1989) make this point to distinguish between the code of clothing and the code of language: 'The knowledge of a code may have more uneven distributions for products than it does for language’ (p 29) with the consequence that there is 'much greater variation in code mastery than is the case for language’. Belk, Mayer and Bahn (1981) carried out one of the first studies into the decoding of messages conveyed by a person’s choice of a particular product. The variables of age, gender and social class were hypothesised to account for differences in perception for the highly visible and socially expressive products of automobiles and houses. They found that students were more likely than older adults to rely on 'consumption-based stereotypes’ and thus make more nearly unanimous judgements about the owners of particular cars and houses. They were surprised to find that men had more consistent judgements than women, and explained this with reference to the importance for men of recognising status differentials 'for achievement motivation and career preparation’. Both lower and higher social classes exhibited consumption-based stereotypes in their judgements, but different stereotypes prevailed among the two groups.

Clothing is particularly susceptible to differences in consumption stereotyping, and therefore to differences in ability to decode a range of messages. Noesjirwan and Crawford (1982) make the point that 'clothing is primarily a means of communicating, not personal identity, but social identity’. They conclude that 'clothing is symbolic of that [social] identity and the values espoused by the group. The same values serve as a yardstick for judging the clothing worn by others and the social identity symbolized by it. . . . Advertisements that project an implied social identity through a model’s clothing are likely to have quite different effects on different kinds of consumers.’

Understanding how people interpret clothing, and how different groups of people make different judgements about the same brand of clothing is critical to clothing manufacturers and their advertising agencies. A knowledge of how different groups are likely to differ in their response to a particular set of symbols means that the marketer has more control over the decoding of the brand image when advertising is 'consumed’. Jeans are an ideal case for examining brand meanings: there is a wide range of branded and unbranded jeans available, and they are almost universally worn by people of both sexes from adolescence through at least to middle age.

In their seminal research on the interpretation of clothing 'codes’, McCracken and Roth (1989) found that females were significantly better than men in interpreting the syntax of clothing codes. That is, women more readily recognised a 'look’ and were more sensitive to fashion cues than men were. Recent research has shown that men and women differ in the way they pay attention to cues in advertising (Meyers-Levy and Maheswaran, 1991) and similarly that they read fashion symbols with different criteria (Meyers-Levy and Sternthal, 1991; Elliott, 1994). Females have been found to be more sensitive to the informative details provided in ads than men generally are.

Following on from McCracken and Roth, Elliott (1994) also found a difference in the way different age groups respond to popular brands in an exploratory study of the meanings of brands of sneakers. McCracken and Roth discovered that age can be a highly significant variable in the interpretation of certain fashion codes, particularly 'punk’; Elliott observed a remarkable difference between secondary school and university students in the semantic clusters elicited in relation to sneakers. The younger children made no specific reference to advertising messages, but had very strong opinions on which brands were fashionable and which were definitely not.

FASHION INVOLVEMENT AND SOCIAL IDENTITY

Social class, which might seem to be a critical variable where social identity is involved, is a problematic variable to measure, primarily because it is difficult to obtain accurate measurements for young adults (acknowledged by Belk, Mayer and Bahn). Nevertheless, because of their attention to fashion this age group must figure highly in any study of clothing codes. In any case, the difference in class may not be so important as the difference in the strength of group affiliation and the consequent attention to a particular code. McCracken and Roth note that the degree of fashion involvement may be a relevant variable in the interpretation of clothing codes. Fashion involvement is likely to be associated with differences in sensitivity to social surroundings in that those who are highly motivated to fit into a particular group will need to be aware of the fashion cues not just of that group but of other less desirable groups so that the 'wrong’ cues may be avoided.

Research in psychology, and more recently in consumer behaviour, has made use of Snyder’s Self-monitoring Scale (1974; revised in Snyder and Gangestad, 1986) to distinguish between people who are sensitive to the social cues around them (high self-monitors) and those who are more likely to suit themselves whatever their social surroundings (low self-monitors). The revised scale measures a person’s control over their social presentation using 18 true-false statements as shown in Exhibit 1. People are then usually divided into two groups using the median, where one group comprises the 'chameleons’ (who constantly change their coat to fit into their surroundings) and the other the 'leopards’ (who never change their spots).

EXHIBIT 1

SNYDER'S REVISED SELF-MONITORING SCALE

Several studies have been conducted to show the influence of self-monitoring groups on responses to advertising (Snyder and DeBono, 1985; Lennon, Davis and Fairhurst, 1988; Snyder, 1989; Johar and Sirgy, 1991; Shavitt, Lowrey and Han, 1992). In particular, Shavitt et al. hypothesised that multiple function products (having both social identity and utilitarian functions) would elicit the strongest differences in response to advertising between high and low self-monitors. In their study, jeans were considered to be a multiple function product 'because they are associated with important utilitarian outcomes (e.g., comfort, durability) as well as social image implications (e.g., conveying one’s style, taste).’ In the actual experiments sunglasses and watches were used as multiple function products, and the findings supported the view that 'self monitoring comes into play when a product affords high and low self-monitors the opportunity to focus on different functional goals’. Hence it was found that there are few differences in attitudes to aspirin but quite distinct differences in social identity or mixed function products.

This study takes the case of jeans and looks at the meanings associated with an unbranded product and the most popular brand in the category to see if advertising succeeds in creating meanings which reside in the product. If so, are these consistently decoded differently by different groups of people? By using the same pair of Levi’s jeans with all brand markings removed for the unbranded stimulus (thereby ensuring that all functional attributes are identical) one can measure the strength of the brand’s meanings. It is hypothesised that high self-monitors will associate more positive meanings with branded jeans than unbranded jeans and that they will differ significantly from low self-monitors in the extent of their positive associations. Low self-monitors are expected to respond more favourably to the utilitarian attributes of both pairs of jeans because they should pay more attention to the product’s construction rather than image in their evaluation.

HYPOTHESES

H1. The meaning of a branded fashion product (Levi's) will be different from the meaning of an unbranded fashion product.

H2. The meaning of a fashion product (jeans) will differ by age and gender.

H3. High self-monitors will have more positive attitudes towards a branded fashion product (Levi's) than will low self-monitors.

H4. Low self-monitors will respond more favorably than high self-monitors to the utilitarian attributes of both a branded and unbranded fashion product (jeans).

METHODOLOGY

Sixteen focus groups were held on a university campus in the north of England with the objective of discovering the meanings people attach to various brands of jeans and to jeans in general. Students aged 18-35 attended and were asked if the meanings they attached to brands had changed since their secondary school days. A list of 32 bipolar adjectives were drawn from the focus group discussions and tested for comprehension on a convenience sample of children aged 14 to 18. Three pairs were removed as a result of the tests. The adjective pairs were randomly distributed with regard to potentially positive and potentially negative connotations along the right hand side. A seven point scale separated each pair. (See Exhibit 1)

A three-part survey instrument was then prepared. The first part was a 7-point semantic differential scale consisting of the 29 bipolar adjectives; the second part contained the 18 self-monitoring true-false statements as devised by Snyder(see Exhibit 2); the third part asked demographic questions. After a pilot survey on campus, two further pairs of adjectives were removed; minor word changes to the Snyder statements were made for adaptation to British usage; and usage questions were added to the demographic section.

803 respondents were interviewed in various town centres in the UK. Males and females, in each of three age bands - 14-17; 18-24; 25-34 - were selected by quota guidelines. These age bands comprise 58% of the market for jeans in the UK (Retail Business, 1994). Half were shown a photograph with Levi's brand markings clearly displayed. Levi's jeans were used because they are the most highly advertised brand by far in the UK; their spend (between .L3m and L5m annually) in the five years prior to this study was four times higher than their closest rival, Wrangler (Retail Business, 1994). ne other half were shown the same pose with the brand markings (pocket stitching, red tab and leather nameplate) removed by computer techniques. The photograph was a back view of a model of indeterminate sex (confirmed by pilot).

98 responses were rejected for analysis by interviewers because of influence by respondents' friends or other interview problems. A further 36 were rejected because of incomplete responses in the self-monitoring section. The final samples were thus 336 who were shown the unbranded stimulus and 333 who were shown the branded.

SPSS was used to analyse each data set. The bipolar scales were reordered so that potentially positive values were all on the right hand side. The mean rating for each attribute was compared across data sets to determine the extent to which branding affects perception. Crosstabulation by sex, self-monitoring score and age was used to identify overall differences between different groups' perceptions of branded and unbranded jeans. The scale of the semantic differential pairs was then collapsed from 7 down to 3, scores of 1-2, 3-5, 6-7 forming groups of strongly positive or negative and neutral for crosstabulation at an individual variable level. The reason for choosing this method of collapsing the categories, rather than the more traditional one whereby only 4 is considered neutral, was to contrast the positive and negative more strongly with a 'buffer' group.

When the seif-monitoring scores were divided into two groups of 'high' (9-18) and 'low' (1-9), as in Snyder's research, virtually 50% was in each group in both data sets. However, in order to pick out self-monitoring effects more strongly than in previous research, it was decided to use three groups, having a 'buffer' group between the highest and lowest self-monitors. Thus the scores were divided up into three groups of low (I to 6), medium (7-12) and high (1318) for the purposes of analysis. This grouping showed a remarkable consistency between the two samples in that the difference in proportions varied by only.1%, giving confidence in the samples (Table 1). By segregating the large group of medium self-monitors, we can be more certain that the results reflect underlying personality differences and not merely statistical fluke.

MEAN DIFFERENCES IN ATTRIBUTE SCORES BETWEEN BRANDED AND UNBRANDED JEANS

The mean scores for the 27 attributes were compared across both data sets. The means of all but six variables were significantly different (p<.000). HI is thus supported. The largest differences are displayed in Table 2.

As one would expect, the Levi's were perceived to be much more well-known than were the unbranded pair. The other large differences testify to the strength of Levi's brand image: they are considered more expensive and high quality than the unbranded pair. The extent to which this mass brand is considered to be ,original' and to 'make a statement' more than its unbranded counterpart is further evidence of Levi's strong image. The nondiscriminating adjectives were: 'for men', 'individual', 'practical', ,associated with cowboys', 'common', 'aggressive'.

AGE AND GENDER DIFFERENCES

For the purposes of crosstabulation of the individual variables, it was decided to collapse the seven point scale down to three, with a strongly positive and negative grouping formed from the two points at each extreme. The attributes were crosstabulated against age, sex and self-monitoring score. (Results of the self-monitoring analysis are discussed in detail below.) The Mantel-Haenszel test for linear association was used to determine significance (p <05) supported by the eta coefficient in the case of sex (a nominal independent variable).

Crosstabulation of the individual variables suggests that 14-17 year olds are more positively disposed to many of Levi's attributes than older age groups are. Indeed, the findings support Elliott's (1994) conclusion that the 'meaning of fashion brands may differ between genders and, less surprisingly, between age groups'. There were significant differences between the youngest and oldest age groups in the perception of Levi's with regard to cut, desirability, fit, sexiness, stylishness and trendiness. It was also found that men and women had different attitudes to branded jeans, specifically to image, modernity, practicality, quality and trendiness, with women having the more favourable opinions. It is interesting that in both cases there is a predominance of image-related attributes, casting doubt on the tendency for women to process functional information more actively than men, or the youngest age group to be exclusively focused on image.

In contrast to the branded jeans, there are relatively few differences between age and gender groups in perceptions of unbranded jeans. H2 is thus supported only in relation to a branded product. Men perceive unbranded jeans to be more American and more original than women do (suggesting that they are either rejecting Levi's messages or not processing the cues as accurately as women do), while 25-35 year olds see them as less well-cut than younger groups do (perhaps because they see all jeans as less well cut). A multivariate analysis of variance revealed a significant interaction (p <.04) between age and self-monitoring and the 27 descriptive variables, particularly in relation to the cut, fit and ,authenticity' of unbranded jeans. Otherwise there was no significant interaction between the three independent variables and the descriptive variables in either set.

EXHIBIT 2

SEMANTIC DIFFERENTIAL SCALE

POSITIVE ATTITUDES TO JEANS AND SELF- MONITORING

A composite attitude towards branded and unbranded jeans was determined by respondents' mean rating of 15 pairs of bipolar attributes, each of which could be identified as having a definitely positive and definitely negative pole. That is, twelve pairs such as >American/not American= were removed because they were capable of being interpreted both negatively and positively by the population. The positives were: high quality; hard wearing; comfortable-, well cut; authentic; sexy; practical; 'classy'; easy to wear; desirable; stylish; a good fit; convey an image; modem; trendy.

TABLE 1

PERCENTAGE OF EACH SAMPLE IN THREE SELF-MONITORING GROUPS

TABLE 2

DIFFERENCES IN MEAN SCORES OF UNBRANDED AND BRANDED JEANS

TABLE 3

POSITIVE ATTITUDES TO JEANS

As might be expected, attitudes towards branded jeans were much more positive, with the mean score being 2.8 (where 1 is most positive) as opposed to 3.6 for unbranded jeans. For the crosstabulation of this variable each sample set was divided into three roughly equal groups representing positive, neutral and negative attitudes. Note that the entry standards' differed between the two sample sets, so that a 2.4 mean was required to get into the positive group in the branded sample, whereas 3.3 was sufficient for the unbranded sample. Hence, 33% would be expected in any cell given percentages in terms of the three self-monitoring groups (only two of which are shown in the tables below; also, for comparative purposes, the percentages of the 'unbranded' sample using the cutoff points of the 'branded' sample are shown in brackets.)

Self-monitoring scores were found to correlate negatively with attitudes towards unbranded jeans (Table 3).

High self-monitors were significantly less likely to regard unbranded jeans positively than were low self-monitors (p<.01). However, contrary to H3, there was no significant difference in the 'branded' sample with regard to self-monitoring.

To test the hypothesis (H4) that low self-monitors are likely to respond more favourably than high self-monitors to the utilitarian attributes of both pairs of jeans, a composite 'utilitarian attitude' variable was derived from the individual attitude scores towards fit, cut, practicality, comfort and durability. It was analysed as described above (Table 4).

Low self-monitors appear, as hypothesised, to have more favourable attitudes to functional qualities of both branded and unbranded jeans than high self-monitors do (p<.02), perhaps because they pay more attention to such matters and therefore have stronger opinions on them.

CROSSTABUILATION BY SELF-MONITORING GROUPS

Quite surprisingly, the individual variables produced only one significant difference with regard to branded jeans and self-monitoring groups, with Levi's being perceived as much sexier by high self-monitors than low self-monitors. In contrast, there were many differences among the perceptions of self-monitoring groups with regard to unbranded jeans. The implication is that high self-monitors virtually define jeans as 'branded jeans'. Unlike the youngest age group they do not have an especially positive attitude towards branded versions of the garment so much as negative ones towards unbranded versions. Interestingly, multivariate analysis using the 15 descriptors of the composite variable suggests that it is the youngest high self-monitors above all who reject the unbranded goods as being not well cut, not a good fit and above all, not authentic (p <006). A summary of all the hypothesised findings shows that a new hypothesis is needed to more accurately describe attitude differences among personality groups with regard to branded and unbranded fashion goods (see Table 5)

TABLE 4

POSITIVE ATTITUDES TO UTILITARIAN ATTRIBUTES OF JEANS

TABLE 5

SUMMARY OF HYPOTHESISED FINDINGS

TABLE 6

SELF-MONITORING BY WELL CUT (UNBRANDED AND BRANDED)

The following tables of detailed findings are a composite of the ,strongly positive' columns from the crosstabulations of both data sets, and hence significant differences are as noted in the text and do not apply to the tables displayed. For the same reason the percentages in the tables do not add up to 100% because each column was based on a different data set and extracted from a nine cell table calculated with row percentages in order to indicate differences that the three personality types (only two shown here) have on attitudes.

Six variables gave rise to a significant difference in response towards unbranded, but not branded jeans. One variable showed a difference between high and low self-monitors regardless of branding. Both utilitarian and image-related attributes of unbranded jeans were perceived differently. High self-monitors were much less likely to consider them to be well cut, in comparison both to low self-monitors and to high self-monitors' responses to branded jeans (Table 6).

Another functional attribute with significant differences between high and low self-monitors for the unbranded stimulus is 'hard wearing' (Table 7).

Perceptions of quality are similarly affected by personality differences. High self-monitors were much less willing to ascribe high quality to unbranded jeans than low self-monitors were, although the branded jeans showed no significant difference between the two personality groups (Table 8).

Surprisingly, many of the image-related variables such as ,makes a statement' 'classy' or 'conveys an image' showed no significant difference in perceptions by personality types. The whole sample was in general agreement that unbranded jeans do not score highly in these categories in contrast to branded jeans. Only with regard to 'trendy' did high self-monitors show a significantly different reaction (Table 9).

Here it is the willingness of one-third of low self-monitors to consider unbranded jeans to be trendy which is notable. Given the much lower differential between their attitudes to branded and

unbranded pairs than those of other personality groups, their consideration of jeans here seems to be generic rather than brand specific. High self-monitors appear to limit trendiness much more to branded jeans. Thus, while low self-monitors do recognise that branded jeans convey a different image and are more classy, they still tend to regard wearing jeans in general as the fashion trend, rather than the more code-conscious high self-monitors who do not.

TABLE 7

SELF-MONITORING BY HARD WEARING (UNBRANDED AND BRANDED)

TABLE 8

SELF-MONITORING BY HIGH QUALITY (UNBRANDED AND BRANDED)

TABLE 9

SELF-MONITORING BY TRENDY (UNBRANDED AND BRANDED)

TABLE 10

SELF-MONITORING BY ORIGINAL (UNBRANDED AND BRANDED)

There is a similar finding with regard to the originality of jeans, although in this case the high self-monitors consistently rate jeans, both branded and unbranded, as less original than low self-monitors do (Table 10).

It may be that originality is not highly prized by high self-monitors, and therefore they are less likely to rate branded jeans highly on this attribute than low self-monitors are.

High self-monitors' perceptions differ most from low self-monitors in their assessment of the individuality of unbranded jeans (Table 11).

High self-monitors clearly rate them as 'not individual'. Less than 10% of all but low self-monitors (13%) positively ascribe individuality to any pair of jeans, the majority being neutral. What stands out, however, is the clear view of high self-monitors that unbranded jeans are definitely not individual.

Similarly, while the majority of respondents perceive all jeans to be comfortable, high self-monitors think unbranded jeans are less comfortable than low self-monitors do (Table 12).

However, they are also significantly less likely than low self-monitors to think that branded jeans are comfortable! Their criteria for judging the positive values of jeans may mean that comfort is not an important distinction to make, in keeping with H4.

In contrast, sexiness is an attribute of great importance to high self-monitors (Table 13).

This attribute was the only one for which there was a significant difference in perception among personality types in the branded sample, but not the unbranded. Surprisingly the majority of respondents in both samples are neutral with regard to sexiness. The table shows that it is not so much that high self-monitors are distinguished by their perceptions of branded jeans' sexiness so much as low self-monitors refuse to ascribe this attribute to jeans, especially branded ones.

DISCUSSION

There were distinct differences in the perceptions of branded and unbranded jeans by different self-monitoring groups. As expected, the branded jeans were perceived as much more well-known than the unbranded. This finding merely confirms that the brand markings on the unbranded photograph were successfully removed. Otherwise, the largest differences in perception between the branded and the unbranded jeans were 'Expensive' and 'High Quality', followed by several image-related descriptors: Original, Stylish, Classy and Makes a Statement. In general, therefore, the findings suggest that Levi's have an established brand image close to the image purveyed in their advertising.

TABLE 11

SELF-MONITORING BY (NOT) INDIVIDUAL (UNBRANDED AND BRANDED)

TABLE 12

SELF-MONITORING BY COMFORTABLE (UNBRANDED AND BRANDED)

TABLE 13

SELF-MONITORING BY SEXY (UNBRANDED AND BRANDED)

When the mean scores of respondents on all unambiguously positive attributes are compared, the overall mean of the branded group is 2.8 where 1 is most positive, compared to 3.6 for the unbranded group. Given, however, that the range of means is almost identical for the two data sets, there are clearly differences in perception attributable to independent variables. Branded jeans were perceived differently according to the age and sex of the respondent, though there were no interaction effects. Unbranded jeans were viewed significantly more negatively by high self-monitors than low self-monitors, especially when mediated by age. The findings suggest that young high self-monitors arc likely to be positive towards branded jeans but much more notably negative towards unbranded. This strength of negative feeling directed towards unbranded jeans was not the expected finding.

Self-monitoring scores, which depend on the extent to which a person desires to stay in tune with other people, appear to have an effect on the perceptions of unbranded jeans in particular. Indeed, there are no significant differences between the three types of self-monitoring with regard to branded jeans. One might say that high self-monitors define jeans in terms of branded jeans, and thus unbranded jeans are regarded negatively whereas branded jeans are just jeans.

This negative effect is not confined to perceptions of image as one might expect. Interestingly, all personality types agree that branded jeans have more of an image and are classier than unbranded jeans. The difference in conviction occurs noticeably with regard to functional attributes. There is a tendency for unbranded jeans to appear less well cut, less hard wearing and of lower quality to high self-monitors. Only with regard to sexiness are branded jeans rated significantly more positively by high self-monitors, and even this appears to be the result of a strongly negative reaction by low seif-monitors to the idea of any jeans being sexy rather than a particularly positive one by high self-monitors.

The findings support DeBono's contention (1987) that attitudes fulfil different goals for high and low self-monitors and therefore attitudes that are apparently similar (eg 'branded jeans are high quality') may be held for very different reasons. While high self-monitors show no significant differences compared to low self-monitors in their attitudes towards the cut and durability of branded jeans, it is clear from the overall results that these are rated positively only because the brand is socially acceptable (their ratings on these attributes for unbranded jeans are much lower than low self-monitors'). Their attitudes appear to have been formed for what DeBono calls asocial adjustive function, that is, to allow them 'to fit into important social situations and behave in ways appropriate to various reference groups.' In contrast, low self-monitors form attitudes for a value expressive (utilitarian) purpose and thus show much smaller differences between their ratings of branded and unbranded jeans with regard to utilitarian attributes. However, it is worth noting that the branded jeans were consistently rated more positively by all groups, and the difference was particularly high for the perception of quality.

This finding suggests that for high involvement items of fashion which are actively used as code by wearers, such as jeans, sneakers (trainers) and watches, there is no need for advertisers to highlight functional quality in order to compete with weakly supported brands or private labels. High self-monitors seem to transfer a negative image to the functional attributes of unbranded goods, and do not let their opinion of a product's utility affect their overall appraisal of a branded item, while low self-monitors share their overall high appraisal of the brand. It is more important to fulfil the symbolic needs of consumers because their functional needs in this case are to a large extent dependent on their symbolic ones. In other words, if the jeans do not satisfy a consumer's image requirements, they will not be perceived to be well cut or comfortable either. This is in keeping with the findings of Sirgy, et al (1991) who note that self-congruity (that is, a match between the symbolic or 'value-expressive' attributes of a product and the self-concept of the consumer) 'biases functional congruity'.

The study supports a model of choice (in high self-monitors) by elimination of the unacceptable rather than positive selection. Branding becomes crucial in this model in that the product class may be defined in the terms dictated by the branding, and even extended to include attributes not 'covered' by the brand. If jeans are unconsciously benchmarked by high self-monitors against the most visible (that is, highly advertised) brand, then all other jeans including other brands will suffer by comparison. There will still be trade-offs in the actual decision such that lesser brands may be selected for a number of reasons, but for high self-monitors the gap between the top brand and a generic product will be such that the generic product is in a different product class and will not even be considered.

This process of choice by rejection of the unacceptable is very similar to Bourdieu's (1984) concept of the 'refusal of other tastes'. This suggests that the vital act of consumer choice may be not to choose that which is most pleasing, but to reject that which is most distasteful: 'when they have to be justified, [tastes] are asserted purely negatively, by the refusal of other tastes'. With fashion products which cross class boundaries, such as jeans and sneakers, there is no longer any 'taste of necessity' underlying the choice (jeans are not worn because they are more affordable than other garments); instead the ubiquity of jeans can be explained by a desire to avoid 'the conspicuous intention of aloofness'. In this model of choice only low-self monitors will risk expressing an indifference to the judgement and taste of their peers by wearing jeans with no social meaning. ne preference for branded jeans expresses the desire for 'distinction', for discrimination between themselves as 'belongers', who have chosen to interpret the social code implied by the brand, and those who remain aloof, in effect a threat to their social identity.

LIMITATIONS

In this study it was not possible to compare each respondent's perception of the branded versus the unbranded product, hence the findings are based on aggregate rather than individual perceptions. Further research is being carried out in which all respondents will receive three different stimuli: a primary brand, a secondary brand and a private label. This should make it possible to measure the effects of branding more precisely than the current study.

By presenting photographs rather than labels, perceptions concerning attributes such as fit might have been affected by the stimuli rather than being based on the respondents' existing attitudes. A possible problem with using labels, however, is that some respondents may not be familiar with one or more of them and therefore have no opinions on them.

CONCLUSION

This study suggests that advertisements do indeed create meanings in fashion items that are decoded with surprising consistency depending on differing levels of fashion involvement. Self-monitoring, which can be seen as a surrogate for fashion involvement, discriminates perceptions of unbranded products in particular. For young high self-monitors in particular, negative meanings appear to be more influential on attitudes than positive meanings. Our findings are in keeping with the school children in Elliott's 1994 study of the meaning of sneakers, for whom 'if a brand was not fashionable it was very definitely unfashionable with strong emotional valence' found in the unprompted adjectives elicited. These findings suggest that advertisers cannot afford to become lax in supporting their fashion brands: a large part of the target market forms attitudes for 'social-adjustive' purposes, and will be quick to change attitude if another brand becomes more visible and hence more socially acceptable.

REFERENCES

Belk, R, Mayer, R and Bahn, K (1981) 'The Eye of the Beholder: Individual Differences in Perceptions of Consump tion Symbolism' Advances in Consumer Research, 9, 523-530.

Bourdieu, P (1984) Distinction: A Social Critique of the Judgement of Taste, London: Routledge and Kegan Paul.

DeBono, K (1987) 'Investigating the Social-Adjustive and Value-Expressive Functions of Attitudes: Implications for Persuasion Processes, Journal of Personality and Social Psychology, 52, 2, 279-287.

Elliott, R (1994) 'Exploring the Symbolic Meaning of Brands' British Journal of Management, 5, S13-S19.

Elliott, R (in press) 'Existential Consumption and Irrational Desire', European Journal of Marketing. Financial Times (1995) 'Designers Who Want to Keep Jeans on the March', 22 October, Weekend, VI.

Hollander, A (1978) Seeing Through Clothes, New York: Viking Press.

Holman, R (1980) 'Apparel as Communication' in E. Hirshman and M Holbrook, eds, Symbolic Consumer Behavior, Proceedings of the Conference on Consumer Esthetics and Symbolic Consumption, New York, 7-15.

Johar, J and Sirgy, M J (1991) 'Value-Expressive Versus Utilitarian Advertising Appeals: When and Why to Use Which Appeal' Journal of Advertising, 20, 3, 23-33.

Lennon, S, Davis, L and Fairhurst, A (1988) 'Evaluations of Apparel Advertising as a Function of Self-Monitoring' Perceptual and Motor Skills, 66, 987-996.

McCracken, G and Roth V (1989) 'Does Clothing Have a Code? Empirical Findings and Theoretical Implications in the Study of Clothing as a Means of Communication' International Journal of Research in Marketing, 6, 13-33.

Meyers-Levy, J and Maheswaran, D (1991) 'Exploring Differences in Males' and Females' Processing Strategies' Journal of Consumer Research, 18, 63-70.

Meyers-Levy J and Sternthal, B (1991) 'Gender Difference in the Use of Message Cues and Judgements' Journal of Marketing Research, 28, 84-96.

Noesjirwan, J and Crawford, J (1982) 'Variations in Perception of Clothing as a Function of Dress Form and Viewers' Social Community' Perceptual and Motor Skills, 54, 155-163.

Park, C W, Jaworski, B and MacInnis, D (1986) 'Strategic Brand Concept-image Management' Journal of Marketing, 50 (Oct) 135-145.

Shavitt, S, Lowrey, T and Han, S-P (1992) 'Attitude Functions in Advertising: The Interactive Role of Products and Self- Monitoring' Journal of Consumer Psychology, 1, 337-364.

Sirgy, M J, Johar, J, Samli A and Claiborne, C (1991)'Self-Congruity Versus Functional Congruity: Predictors of Consumer Behavior' Journal of the Academy of Marketing Science, 19, 363-375.

Snyder, M (1989) 'Selling Images versus Selling Products: Motivational Foundations of Consumer Attitudes and Behavior' Advances in Consumer Research, 16, 306-311.

Snyder, M and DeBono, K (1985) 'Appeals to Image and Claims about Quality: Understanding the Psychology of Advertising' Journal of Personality and Social Psychology, 49, 586597.

Snyder, M and Gangestad, S (1986) 'On the Nature of Self-Monitoring: Matters of Assessment, Matters of Validity' Journal of Personality and Social Psychology, 51, 125-139.

Swartz, T (1983) 'Brand Symbols and Message Differentiation' Journal of Advertising Research, 23, 5, 59-64.

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