Source Similarity and Fashion Newness As Determinants of Consumer Innovation

ABSTRACT - Using a 4 x 4 x 4 fractional replication, the research combined 4 different fashion ideas, in each of 4 levels of newness, and four pictured sources rated from highly similar to dissimilar to the population from which the 48 subjects in the experiment were chosen. For each subject one idea from each newness level was attributed to each source immediately before the subject rated her agreement with the idea. The main effects of both source similarity and fashion newness were significant as hypothesized.


Karen Kaigler-Evans, Clark Leavitt, and Lois Dickey (1978) ,"Source Similarity and Fashion Newness As Determinants of Consumer Innovation", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 738-742.

Advances in Consumer Research Volume 5, 1978      Pages 738-742


Karen Kaigler-Evans, The University of Texas at Austin

Clark Leavitt, The Ohio State University

Lois Dickey, The Ohio State University


Using a 4 x 4 x 4 fractional replication, the research combined 4 different fashion ideas, in each of 4 levels of newness, and four pictured sources rated from highly similar to dissimilar to the population from which the 48 subjects in the experiment were chosen. For each subject one idea from each newness level was attributed to each source immediately before the subject rated her agreement with the idea. The main effects of both source similarity and fashion newness were significant as hypothesized.


The diffusion of innovations and the adoption or non-adoption of new products and ideas is an area that has been cultivated by several disciplines including home economics, rural sociology and marketing. The related topic of social influence has occupied the attention of sociologists particularly. Most of the research studies in these areas have been field studies that have structure variables that underlie diffusion such as social strata and homogeneity of membership groups or the role of key individuals in relaying information and influence: innovators and opinion leaders, for example.

At the same time social psychologists also have been interested in the adoption of new ideas under the guise of attitude change and persuasion. However, the research paradigm, unlike that of home economics, has included the laboratory rather than a field setting and has been concerned largely with the individual processes underlying the adoption of new attitudes and beliefs. Social psychologists, together with students of mass communication, have been interested in the role of mass media in social influence and have dealt in this context with two general classes of independent variables, those related to the source and to the message.

There has been remarkably little cross-fertilization between these two research traditions of innovation and attitude change. The home economists' concern with innovation or education is essentially an application problem in need of conceptual enrichment while attitude change research has been characterized by a great deal of theory, but is unseasoned by serious attempts at application. The research reported here is an attempt to combine elements of both traditions. It uses the laboratory paradigm of attitude change research while attempting to enrich the independent variables in a way most appropriate to the research tradition familiar to the home economist, particularly that of social influence and fashion adoption. Primarily we have attempted to meld variables from both disciplines and have taken over only the necessary minimum of conceptualization from either field. Before describing the experiment, we shall review the ideas from each field that might interact profitably with those from the other.

Social Influence and Fashion Adoption

Three kinds of studies may be mentioned briefly: the psychological significance of clothing and fashion; population trait distribution related to innovators and opinion leaders, and the relation of social structure to adoption behavior. Each provides evidence for the relevance of the variables used in this research.

Studies in the literature of home economics deal with the function of clothing as both a cue used by others to make social judgments and as a means of expressing the wearers own internal needs or values (Gibbons, 1969; Compton. 1962; Douty, 1963; Rosencranz, 1962; Connor, et al., 1975). This research is only indirectly relevant here: it makes clear that clothing is rich in personal meaning and that it can also be ambiguous as to what it can mean for the self.

There has been a sizeable body of research on social influence both in relation to fashion and to other products and ideas. Informal influences or word-of-mouth activities have long been of interest to social scientists. The two-step flow of influence proposed by Lazarsfeld and Merton, (1954), is a classical attempt to explain the lack of an immediate, obvious effect of mass media by invoking informal networks as leveraging agents. Most of this research has attempted to identify the kind of people who are influential in adoption and who adopt early in the life cycle of an innovation. In particular, King, (1963), has used such research to challenge the traditional concept of fashion as a vehicle for status differentiation that "trickles down" from higher to lower strata. Instead he emphasized the importance of peers or similar others as sources of fashion information (see Grindering, 1967).

More recently a different set of constructs has been invoked to examine the spread of ideas in informal communication. These are variables applied to communication networks, in particular the interconnectedness and the homogeneity of an individual's network. A person's network of social contacts can vary from homogeneous inter-connectedness to a state of more heterogeneous, looser connections. At some point between these two extremes, conditions are right for maximum influence.

Any particular person is more likely to be exposed to new ideas when interacting with a person who is different, but at some point the difference will become so great that communication will break down or fall off. The balance point of greatest effectiveness between these competing tendencies has been called the point of optimal heterophily, a term originally introduced by Merton, (1957) and made current by Rogers, (1976). Similar concepts underlie other field studies. Laumann, (1973), found that men in interlocking networks were less open-minded with respect to non-conformity than those in radial networks. Interlocking networks are more homogeneous in respect to religious and occupational status while the radial networks represent weaker ties and greater heterogeneity. The idea that we learn more or are similar to us has been the subject of other field studies. Granovetter, (1973), and Liu and Duff, (1972), have attempted to demonstrate what they called the "strength of weak ties" for the diffusion of innovations. People are more likely to accept suggestions from others with whom they have only weak ties.

Studies of Source and Message in Attitude Change

Much of the research mentioned above has been concerned with specific products or fashions. Most studies also are concerned with the characteristics of the person who influences others. These two variables are paralleled by message and source determinants in persuasion studies. We shall not review message research here except to point out that innovations and fashions are characterized by newness which is a frequently used variable in psychology, both in terms of familiarity and novelty seeking. It is often regarded as an underlying determinant of cognitive arousal.

Source variables have been approached in two ways that are relevant here. First, basic perceptual dimensions of the source of communication have been studied as interpersonal perceptions. Typically three dimension appear: expertise or competence, trustworthiness or safety, and attraction or dynamism (Giffin, 1967). (A similar approach has been used with message variables in which case novel and useful seem to emerge as basic dimensions.)

The second source variable of relevance here is similarity of the source to the receiver. Two contrasting interpretations have been made regarding the role of similarity. Byrne and others have demonstrated a relation between similarity and attraction which lends credibility to the "myth of super-representativeness" (Simons, 1973). This refers to the popularly held notion that influential speakers are seen as being similar to the audience -- "one of us." Clearly, however, an expert source would not be similar to the receiver -- the more expert, the less similar. Alpert and Anderson, (1973), found a curvilinear relation between similarity and influence -- not surprising in light of these considerations.

Whichever the direction of relation in any particular case, the variables of similarity of source and of newness of the idea are highly relevant to the problem of describing the underlying dynamics of consumer innovation. This is especially the case in view of the recent trend toward examining social influence and communication in terms of peer groups and homogeneous networks. The experimental procedure followed here not only used these variables but attempted to operationalize them in as representative a manner as possible in a laboratory setting.

In an experiment done to find the point of "optimum heterophily," Alpert and Anderson,(1973) used 3 students as presumed sources of statements about air travel. They were presented in photographs and varied from normal appearing to deviant and were shown to student subjects drawn from the same population (business administration). The pictured source that was most influential (evoked most agreement with statements attributed to him) was the one that was intermediate in magnitude of similarity to the subjects as measured by profile similarity on a series of rated personal traits.

The present study extends these findings to fashion suggestions and also varies the degree of newness of the suggestions. More important, it defines similarity in a more general way using self/similarity ratings of a group of independent judges to preset the source values. To the extent that this method is judged to be successful the results raise the question of whether similarity to the self is the critical variable or whether it is an artifact of group representativeness stemming from group similarity ratings.

We hypothesize that the relation between source similarity and influence or agreement with suggestions made by the source will be curvilinear with middle level sources being most influential (the so-called inverted U curve) as would be predicted by the notion of optimum heterophily.


While the experiment to be reported here was straightforward, several auxiliary studies were done beforehand to prepare the experimental materials and some additional measures were gathered on the subjects themselves in a later session. We shall first describe the preparation for each of the two experimental manipulations: the source of the innovations and the message about the innovation itself.

Selection of Persons for Sources

In an effort to maintain some of the naturalism of field studies, we used as sources of the fashion suggestions snapshots of actual coeds taken on the campus of a large Midwestern university where the experiment itself was to be carried out. The photographs had a casual look as though taken impulsively by a friend, although the size, background and camera angles were relatively uniform. Forty photographs were taken in as unsystematic a way as possible using different times and places and types of subjects.

From this pool, sixteen pictures were selected that seemed to represent the range of women typical of home economics undergraduates. A group of 63 home economics undergraduates, none of whom knew the women in the snapshots, were asked to judge each photographed woman on a five point scale, ranging from "extremely similar to you" to "not similar at all." Using the method of summated ratings, the sixteen photographs were scaled and four were chosen covering equal intervals of the scale (3.53, 2.85, 2.42 and 1.98 were the actual values of the mean similarity ratings).

The scale represents group similarity because the self ratings of similarity to each person in the group, when combined, become an index of the similarity of the photographed person to the entire group. (We felt that a direct rating of similarity to the group would be a meaningless task.)

Selecting of Fashion Suggestions

Although it would have been possible to bring actual fashion innovations into the laboratory, such a strong manipulation would have overwhelmed the source variable. In any case, influence is most often manifest in talking about innovations before the actual trial. To maintain realism in representing the fashion object we asked women from undergraduate classes to write down fashion suggestions that they actually had heard recently. The term "suggestion" was not given a limited interpretation but included all forms in which comments might be made such as "I like ...". A total of 78 women contributed 352 items. After editing and eliminating duplication there were 58 acceptable fashion comments.

Next we asked a panel of five buyers from local stores catering to students to rate each statement for the extent to which it was a new idea for local college women. They sorted fashion comments into four groups from extremely fashionable to not-so fashionable. The four levels of fashion newness included such statements as "Long dresses are fun to wear to parties" and "Shoes with rope trim are comfortable." Sixteen statements were selected, four from each newness level, for presentation to the subjects in the experimental treatments. All but one statement lacked unanimous agreement concerning the newness level.


The main experiment was carried out with 48 undergraduate women enrolled in home economics. None were used in any of the previous steps. Each was assigned to one of 4 groups in a Latin square design pairing each of the 4 items at each newness level with one of the 4 photographed sources using a different combination in each of the 4 groups. Thus each of the sixteen items was paired with each source an equal number of times, the order being counterbalanced in the Latin square design. The photographed source was shown using a 35 mm slide projected on a screen in front of the subjects.

When the picture of the source was projected on the screen the subject indicated on a five-point scale her agreement with the fashion opinions attributed to that source. Each subject made judgments about 16 items, four from each source.

A second session of the experiment required all subjects to rate each source on a series of trait descriptors. In this session, following the main experiment by one week, each subject was given a personality scale to measure open versus cautious information processing (Leavitt and Walton, 1975), a measure of affiliation or liking for each source (Hayes, 1972), and the series of adjective ratings to describe the sources (reported in detail elsewhere).


Complex analysis of variance was used to determine the significance of the results using Plan 12 of Winer (1962, page 571). This is a 4 x 4 x 4 balanced fractional replication Latin square in which the treatments are assumed to be fixed and subjects within groups are a random variable. The three main effects are the four sources, the four newness levels and the four replicated fashion suggestions within each newness level (assumed to be identical in the design of the study). The re-suits are contained in Table 1.

The main effects of sources and newness level were significant. Their interaction was not significant for all groups combined, but when the groups are considered separately they have a significant interaction. Replicates by groups interaction is also significant indicating that response to the individual items was determined not only by newness level and source pairing but also by their unique qualities. This failure to achieve uniform response to items within newness levels -- while not surprising -- makes the interactions in the study difficult to interpret.

The mean agreement level for the four sources is shown separately for each newness level in Figure 1. Even though the first order interaction is not significant, combining the newness levels may be misleading. It can be seen that the effect of similarity is most unequivocal for the two extreme newness levels while the two middle levels show the inverted U curve found in previous research.

The overall effect of the group similarity of the four sources is curvilinear in roughly the same manner as the two middle levels of newness. The mean agreement from most to least similar was 3.6, 3.4, 3.7, and 2.9. The exception to the trend was source II. In order to investigate this further, the scores were plotted separately for the thirteen subjects who scored highest and the eleven subjects who scored lowest on the open processing scale given in the post session. This is a measure of the extent to which the person adopts an open, constructive attitude toward innovation. The results are graphed in Figure 2. Clearly it is source II that distinguishes the two types of processors.

To explore this difference further, the average rating for all four sources on the 26 trait ratings made by all subjects in the post session was computed. The strong and weak traits for each source are listed in Table 2. It can be seen that source II and source III are almost mirror images. Source II is high on anxious and aggressive and low on conventional and trustworthy while source III is the reverse except for conventional.





This is interpreted as indicating that open subjects are less dependent on conventional credibility cues such as expertise and more likely to respond to a source who is somewhat controversial as a source per se, but more representative of the group with which the subject identifies.





What do the results tell us about the psychological processes underlying consumer innovation? They clearly demonstrate the importance both of characteristics of the source and of the innovation itself for acceptance of fashion ideas. The attempt to ground laboratory experimentation in the real world by using naturalistic manipulations -- snapshots and "overheard" fashion ideas --was only partially successful since the realistic flavor of the fashion items seemed to increase the "noise level" of the analysis: the four replicates within each newness level contributed significant interaction variance in the design. This could have been avoided by constructing sterile phrases designed to produce response only in terms of the one aspect of interest -- newness. Also, the effects of the photographs were more difficult to interpret than would they have been if only one dimension were varied -- hair color for instance. But, all-in-all, these individualistic variations give us more faith in the generalizability of the results than would have been the case if a less naturalistic stimulus variation had been used.

Our measure of similarity was more generalizable than Alpert and Anderson's, but the operation that makes it so -- independent judgments -- raises the question of whether the essential variable is similarity of source to the individual receiver or to the group as a whole (super-representativeness). Further research will be required to unconfound these two interpretations.


Mark Alpert and W. Thomas Anderson, "Optimal Heterophily and Communication Effectiveness: Some Empirical Findings,'' Journal of Communication, XXIII(September, 1973), 328-343.

Norma Compton, "Personal Attributes and Color Design Preferences in Clothing Fabrics," Journal of Psychology, LIV (1962), 191-195.

Barbara Connor, K. Peters and M. Nagasawa, "Person and Costume: Effects on the Formation of First Impressions," Home Economics Research Journal, IV(September, 1975), 32-41.

Helen Douty, "Influence of Clothing on Perception of Persons," Journal of Home Economics, LV(1963), 197-220.

Keith Gibbins, "Communication Aspects of Women's Clothes and Their Relationship to Fashionability," British Journal of Social and Clinical Psychology, VII(December, 1969), 301-312.

N. Giffin, "The Contributions of Studies of Source Credibility to a Theory of Interpersonal Trust in the Communication Process," Psychological Bulletin, 68(1967), 104-119.

M. Granovetter, "The Strength of Weak Ties," American Journal of Sociology, 78(May, 1973), 1360-80.

Margaret Grindering, "Fashion Diffusion," Journal of Home Economics, Vol. 59, No. 3, (March, 1967), 171-174.

Charles W. King, "Fashion Adoption: A Rebuttal of the 'Trickle Down' Theory," in Stephen A. Greyser, ed., Toward Scientific Marketing (Chicago: American Marketing Association, 1963), 108-125.

Edward O. Laumann, Bonds of Pluralism: The Form and Substance of Urban Social Networks (New York: John Wiley and Sons, 1973).

Paul F. Lazarsfeld and Robert K. Merton, "Friendship as Social Process: A Substantive and Methodological Analysis," in Monroe Berger, Theodore Abel and Charles H. Page, ed., Freedom and Control in Modern Society (New York: D. Van Nostrand Co., Inc., 1954).

Clark Leavitt and Karen Kaigler-Evans, "Mere Similarity Versus Information Processing: An Investigation of Source and Message Interaction," Journal of Communication, 2(1975), 300-6.

Clark Leavitt and John Walton, "Development of Scale for Innovativeness," in Mary Jane Schlinger, ed., Advances in Consumer Research, Vol. III (1975), p. 545.

W. T. Liu and R. W. Duff, "The Strength in Weak Ties," Public Opinion Quarterly, 36(Fall, 1972), 361-66.

Robert K. Merton, Social Theory and Social Structure, (Glenco, Illinois: The Free Press, 1957).

Everett Rogers, "New Product Adoption and Diffusion," Journal of Consumer Research, No. 4(1976), 290-306.

Everett Rogers and Dilip K. Bhomik, "Homophily and Heterophily: Relational Concepts for Communication Research," Public Opinion Quarterly, XXXIV(Winter, 1970-1971), 523-538.

Mary Lou Rosencranz, "Clothing Symbolism," Journal of Home Economics, LIV(1962), 18-22.

Herbert W. Simons, "Perception and Communication," in C. David Harper and Row, Publishers, ed., Advances in Communication, (1973), 215-240.

B. J. Winer, Statistical Principles in Experimental Design (McGraw Hill, 1962).



Karen Kaigler-Evans, The University of Texas at Austin
Clark Leavitt, The Ohio State University
Lois Dickey, The Ohio State University


NA - Advances in Consumer Research Volume 05 | 1978

Share Proceeding

Featured papers

See More


Remind Me of What I Have: Thinking about a Favorite Possession Mitigates the Negative Impact of Inequality on Subjective Well-being

(Joyce) Jingshi Liu, Hong Kong University of Science and Technology
Amy Dalton, Hong Kong University of Science and Technology
Anirban Mukhopadhyay, Hong Kong University of Science and Technology

Read More


C11. More of a Bad Thing: How Consumers Ignore Pollutant Levels in Healthiness Assessment

Aner Tal, Ono Academic College (OAC)
Yaniv Gvili, Ono Academic College (OAC)
Moty Amar, Ono Academic College (OAC)

Read More


J8. Exchange with The Rich, Concern with The Poor: The Effects of Social Class on Consumer Response to Brand Relationship

Bing Han, Shanghai Jiao Tong University
Liangyan Wang, Shanghai Jiao Tong University

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.