Professional Documents
Culture Documents
Womens Fashion Opinion Leaders
Womens Fashion Opinion Leaders
Womens Fashion Opinion Leaders
ÉRIC VERNETTE Preparing a media plan aimed at opinion leaders requires accurately identifying and
University of
describing the attributes of this target as well as measuring its affinities with different
Toulouse 1, France
vernette@univ-tlse1.fr media. Our research findings on women’s fashion, particularly magazines, reveal that
a media plan targeted at opinion leaders can succeed, that these opinion leaders
tend to be positive toward and discuss advertising media, and that they read more
women’s fashion magazines and have significantly more affinities with such media
than nonopinion leaders.
THE INTEREST IN OPINION leadership was first in- terknit relationships; at the same time, leaders
vestigated by sociologists in the United States in could be viewed as “opinion brokers” who carry
the 1950s. Their studies showed how those opin- information across the social boundaries between
ion leaders who are more exposed to media pro- groups (Burt, 1999). The influence wielded by an
cess the information they receive and forward it opinion leaders will depend on its degree of “cen-
to their immediate circle of friends or relatives trality” in the group and on the strength of the
(Katz and Lazarsfeld, 1955). In marketing, the links that join members of the network (Reingen
opinion leader is someone who informally influ- and Kernan, 1986). In a postmodern context, con-
ences the attitudes of other individuals in an in- sumption is part of a world of symbols and im-
tended direction (Reynolds and Wells, 1977). A ages (Firat and Venkatesh, 1995). Consumption
great number of marketing researches conducted enacts a social code that reflects affiliation to a
in the 1970s and 1980s highlight the potential of given social network. The judgments of a group
opinion leaders as a media target because they on individual choices and on the appeal of a
provide a primary “word of mouth” source of product become more important than the product
information in interpersonal communications: their itself. Within the group, the opinion leader’s judg-
immediate environment (friends, colleagues, neigh- ments would be given greater importance and
bors, social contacts) seeks their views before or emulated by his immediate environment for ex-
after buying a product or service (Bearden and pressing support for the norms of the group.
Etzel, 1982; Dichter, 1966; Montgomery and Silk, While opinion leaders provide an attractive tar-
1971; Newman and Staelin, 1973). Research has get for advertising media, a media plan centered
also shown that information spread by word of on this target raises a number of questions: how
The present study has been sup- mouth has a greater impact on decisions to buy can opinion leaders be identified in a given seg-
ported by advertising space seller than other marketer-dominated sources of infor- ment? Do the media they read statistically differ
Interdeco Global Advertising, a mation such as publicity (Herr, Kardes, and Kim, from the media read by nonopinion leaders? This
subsidiary of the Hachette Fili- 1991; Price and Feick, 1984). article aims to address these questions by trying
pacchi Media. We are particu- More recently, the theory of social networks has to identify the specific characteristics of these opin-
larly grateful to its marketing revived interest in opinion leadership (Iacobucci ion leaders as a media target. We will first discuss
director Bruno Schmutz for his and Hopkins, 1992). A social network includes a opinion leaders as a potential target for a given
contribution and commitment. large number of players and a structure of in- media plan. We will then go on to make different
hypotheses on opinion leader attitudes to tion has been invalidated by King and negative attributes (Holmes and Lett, 1977;
advertising media. Finally, we will dis- Summers (1970). Mizerski, 1982).
cuss our research results on the women’s
clothing fashion market in France. Strengths of a media plan aimed
How to target opinion leaders
at opinion leaders
in a media plan
THE POTENTIAL OF MEDIA PLANNING Regardless of the nature and sphere of
This entails three stages: identifying a sub-
AIMED AT OPINION LEADERS influence occupied by the opinion leader,
group of opinion leaders in the chosen
a media plan benefits from the opinion
marketing target; selecting relevant vari-
The role of the opinion leader in leader in the following two ways.
ables (social and demographic, lifestyles,
interpersonal communications
attitudes) to distinguish opinion leaders
According to Engel, Blackwell, and Min- It leverages its audience. The “two-step
from the remainder of the target; ranking
iard (1995), the opinion leader directly flow” and “multistage interaction” mod-
types of publications according to the af-
(by word of mouth) or indirectly (by peo- els both regard the media as an essential
finity they elicit on the part of opinion
ple’s imitating his or her behavior) has a information source for opinion leaders.
leaders.
major impact on his immediate environ- The latter will then be overexposed to
ment. The theory of the diffusion of inno- advertising in their areas of interest. More-
Stage 1: Selecting a method to identify
vation stresses the dual role of opinion over, because opinion leaders have an
opinion leaders. The current method used
leaders as information transmitters (Feick enduring involvement with consumer
to identify opinion leaders measures the
and Price, 1987; Gatignon and Robertson, products or services (Venkatram, 1990),
volume of information they exchange and
1985; Johnson-Brown and Reingen, 1987) the opinion leader likes to discuss con-
the degree of influence they exert. Three
and influencers because their status im- sumer products with his/her immediate
methods are available to do this but only
parts social visibility to the product. In environment: much of the information that
the aforementioned method applies to the
his model, Rogers (1983) ranks opinion reaches this indirect audience will tend to
marketing of consumer products and
leaders as interfaces between innovators originate from messages contained in
services.
and early majority. advertisements.
Katz and Lazarsfeld (1955) had initially
formulated a two-step flow communica- It increases advertising persuasion. In- Sociometric. Respondents are asked to
tion model in which the opinion leader formation exchanged in interpersonal com- name the people they turn to for advice.
first interprets the information provided munication has a stronger impact on An American survey conducted among
by its source, then forwards it to his im- consumer purchasing than advertising 20,000 general practitioners and specialist
mediate environment. Current research has (Herr, Kardes, and Kim, 1991). The views physicians revealed that on average each
uncovered numerous ways in which opin- of the opinion leader are recognized as physician seeks medical advice from five
ion leaders and receivers interact (Aassel, authoritative, and his/her advice is fol- colleagues and friendly advice from three
1983; Yale and Gilly, 1995). Some authors lowed because it is deemed to be unbi- other colleagues (Collins, Hawks, and Da-
prefer the term “influential,” which they ased and impartial. Prior to purchase, the vis, 2000). This method is effective in in-
regard as forwarding information on con- judgments of the opinion leader on con- dustrial marketing because the parent
sumer products and brands, rather than sumer brands shape and fashion the be- populations are small; however, this is
the term “opinion leader,” which implies liefs of his immediate environment; these unwieldy when applied to mass con-
a dominant position in the exchange of judgments contribute to customer satisfac- sumer products.
information. However, this term is some- tion (or dissatisfaction). A consumer brand
what ambiguous. In effect, being an influ- endorsed by the opinion leader will tend Key informants. This method, applied in
encer implies having a “strong personality” to increase its appeal among its target. ethnography and psychosociology, uses
(Weiman, 1991) and feeling different from Conversely, a poorly rated brand may be participant observation: an observer (group
other people and seeking to act differ- swiftly rejected by the opinion leader’s member) appoints one or more indi-
ently (Chan and Misra, 1990). If this were environment because in interpersonal com- viduals who act as opinion leaders. This
true, the opinion leader would influence munication the individual receiving the approach is relevant to organizational re-
all consumer products, but this assump- information gives greater weight to the search (sales forces, industrial buyers, etc.)
but unsuitable for quantitative marketing leadership (Richins and Root-Shaffer, 1988; tion is fraught with consequences: if it
studies. Venkatram, 1990). The opinion leader’s were to be confirmed, the affinity of me-
expertise must be perceived by his imme- dia audiences will be identical within the
Self-designating. Respondents self-assess diate environment but this is not enough same product category for opinion lead-
their influence in a given category of con- of a prerequisite to make him or her an ers and nonopinion leaders alike. To be
sumer products or services by answering undisputed opinion leader (Kelman, 1961). worthwhile, a media plan targeted at opin-
a series of standard questions. This is a In terms of personality, opinion leaders ion leaders must establish specific affinity
suitable solution for quantitative studies. share little. Research has revealed that indices.
By using measurement scales with verbal opinion leaders are more sociable (Johnson-
ratings, opinion leaders can be identified Brown and Reingen, 1987), more active METHODOLOGY
within a given population. Different scales (Piirto, 1992), more innovative (Gold- The above discussions involve defining a
have been developed and compared in smith and Desborde, 1991) and have more range of product categories in order to
North America (Childers, 1986; Flynn, self-confidence (Summers, 1970) than non- identify and measure the attitude of opin-
Goldsmith, and Eastman, 1996; Gold- opinion leaders. The latter claim, how- ion leaders toward the media.
smith and Desborde, 1991; King and Sum- ever, is not corroborated by Goldsmith
mers, 1970). Although the results of this and Desborde (1991). Summers (1970) con- Choosing the product category
research confirm that these scales con- cludes that female fashion opinion lead- and media
verge and are reliable, the King and Sum- ers are more emotionally stable and suffer We have chosen the women’s clothing
mers (1970) scale raises methodology issues less from depression than the average fashion for several reasons. First, fashion
due to the way responses are recorded population. has often been a benchmark because of
and the fact that one item reduces Cron- Sociodemographic variables (age, in- the interest it draws from opinion leaders
bach’s alpha value. In addition, scale di- come level, gender, occupation) provide (Dawson and Ridgway, 1987; Goldsmith
mensionality has not been clearly resolved: little or no explanation as to the role of and Stith, 1992; Gutman and Mills, 1982;
Goldsmith and Desborde (1991) identify the opinion leader (Myers and Robertson, King, 1969; Summers, 1970). According to
two dimensions, whereas Flynn, Gold- 1972). These variables depend on differ- King (1969), fashion gets disseminated not
smith, and Eastman (1996) and Childers ent product categories (Engel, Blackwell, in a descending manner as the trickle-
(1986) have identified just one. Finally, and Miniard, 1995). For example, research down theory claims but interactively.
the external validity of the method is ques- shows that opinion leaders who favor the Lipovetsky (1987) goes further when
tionable because opinion leaders charac- internet are mainly young men, students, he maintains that since the explosion of
teristics differ from one culture to another and senior managers (Vernette, 2002). Chanel’s Mademoiselle in 1920, the fashion
(Marshall and Gitosudarmo, 1995). world has lost its epicenter: fashion, he
Stage 3: Measuring media affinity among claims, no longer belongs to a social elite
Stage 2: Identifying the specifics of opin- opinion leaders. Targeting opinion lead- but spans all social classes. While today’s
ion leaders. In the late 1960s, some au- ers is only of practical relevance if media world of fashion does not erase differ-
thors argued that opinion leaders were mainly utilized by opinion leaders can be ences in social ranking, it reduces such
polymorphic, i.e., that they embraced many identified and if such media differ signif- differences by giving greater importance
categories. Today, it is agreed that an opin- icantly from those chosen by nonopinion to individual expression. Such ephemeral
ion leader is rather monomorphic and leaders. However, according to Engel, signs are dictated by a desire to seduce
that only a small percentage (13 percent) Blackwell, and Miniard (1995, p. 732), this and be different (within marginal groups).
has a leadership in more than four catego- is impossible: “Even if they can be iden- However, Lipovetsky (1987) notes that
ries (King and Summers, 1970). However, tified, their media exposure patterns often equally, “people are more informed but
opinion leadership may be shared be- do not differ in any meaningful ways more unstructured, more adult but more
tween similar product categories (Myers from those of receivers. Hence, it may be unstable, more open but more open to
and Robertson, 1972). On the other hand, impossible to mount strategies that reach influence.” As a result, so the argument
the opinion leader tends to have endur- only this segment . . . Use of general me- goes, there is not one fashion trend but
ing involvement with the product cat- dia will be wasted resulting from the large many delocalized and contrary trends
egory in which he exercises his opinion number of non-prospects.” This assump- driven by a whole host of groups orga-
nized around the opinion leaders. A sim- vey. The aim of this survey was to collect Because of the scope of our study, we
ilar view is held by Bourdieu (1979) who information from the same individuals on decided to make use of a self-assessment
notes that fashion leads to asserting a their consumer product buying patterns scale on opinion leadership. We chose the
“difference,” according to aesthetic codes and on the publications they read. This final version of the Childers’s scale (1986)
and measures that result in form prevail- type of survey polls readership of indi- that recommends eliminating the conten-
ing over function: fashion objects re- vidual magazine titles and is widely used tious item from the scale. We therefore
garded as “legitimate works of art” in media planning. By analyzing its find- have chosen five items measured on a
produce subtle “differences” that give rise ings, we can match opinion leadership 5-point semantic scale. The five questions
to ranking between and within social with a host of sociodemographic, attitudi- that make up the leadership scale are
groups. nal, and behavioral variables. presented in the Appendix. In practical
While opinion leadership in the fashion terms, the total leadership score for each
market is not significantly correlated with Measuring opinion leadership individual is obtained by adding together
the need for uniqueness, the opinion leader Each method of identification has its pros the scores that the individual has ob-
tends to prefer fashion brands that it views and cons (Rogers, 1983). While identify- tained on the five questions: the higher
as unique (Dawson and Ridgway, 1987). ing leaders by the sociometric method the total score, the greater the individu-
Finally, because our investigations center has the merit of being more objective, it is al’s opinion leadership in women’s fash-
on France, a country famed for fashion nonetheless admitted that this method is ion wear. The reliability and dimensionality
flair and expertise, the fashion and gar- only feasible for delimited and uniform of this scale have been verified in a French
ment industry is a sound choice cultur- social networks, provided all members of cultural context by Ben Miled and Le
ally and economically. the networks are mutually acquainted Louarn (1994). The latter have validated a
We have selected a single media chan- while remaining accessible to the re- two-dimensional factor structure (commu-
nel because a comparison between differ- searcher. For example, the method can be nication and influence) with a coefficients
ent media plans is more consistent when usefully applied to networks of profes- ranging between .68 and .82 depending
conducted within similar media vehicles sionals or communities with mutual inter- on the dimensions and consumer prod-
to leverage the effectiveness of a given ests (associations, sports clubs, trade ucts concerned.
advertising budget. Magazines have been unions, retail channels, scientific, commu-
chosen for two reasons. First, they occupy nities, universities, colleges) (Weiman, HYPOTHESES
a large part of the media market in France: 1994). [Weiman (1994) uses this method The above considerations and comments
some 20 percent of total advertising in- within the context of an Israeli kibbutz; prompt the following hypotheses:
vestments are allocated to them, more than Collins, Hawks, and Davis (2000) have a
in any other European country. Second, file of general practitioners listed under H1: Women’s clothing fashion opin-
this medium is particularly favored by professional services and suppliers.] Fi- ion leaders have a more favor-
fashion advertisers: consumer magazines nally, as Rogers points out (1983, p. 278), able attitude toward media
capture 51 percent of French advertising this method is not applicable in a sam- advertising than nonopinion
budgets in this industry. pling plan. On the other hand, in an open leaders.
marketing environment such as in the
Sampling world of mass consumer products, self- Individual affinity with the message of
Our sample is based on 10,000 usable designating methods are more flexible, an advertisement is a particularly useful
questionnaires representative of the 47 are easier to use, reliable, and a sound variable in media planning (Smit and
million French people over 15 years old indicator of attitudes on information Neijens, 2000). In the population at large,
(quota method) recruited by a French searching and product purchasing (e.g., the perception of advertising clutter var-
subsidiary of the Taylor Nelson Sofres Childers, 1986; Goldsmith and Desborde, ies according to the type of media con-
group. The questionnaires were fielded 1991). Moreover, as revealed in the con- cerned: television and magazines exhibited
and mailed to respondents during the sec- clusions of the Jacoby (1974) comparative the highest level of advertising-related
ond quarter of 2000. Our questionnaire study, the three main methods (key infor- communication problems (Elliot and Speck,
was included in the annual SIMM-Scanner mant, sociometric, and self-designating) 1998). Since product knowledge is corre-
survey, France’s main media market sur- satisfactorily converge. lated (r 5 .46) with opinion leadership
(Goldsmith and Desborde, 1991), opinion H4: The degree of quantitative affin- is so, the appeal of a given magazine
leaders need advertising to obtain infor- ity that women’s clothing fash- advertisement could have a higher im-
mation (Gutman and Mills, 1982). There- ion opinion leaders have with pact on leaders in identically perceived
fore, they should feel less inconvenienced different media differs signifi- magazines.
than nonopinion leaders. cantly for the women’s maga-
zines segments concerned. RESULTS
H2: Women’s clothing fashion opin-
ion leaders discuss media ad- According to Engel, Blackwell, and Min- Calibrating scores on the opinion
vertising more than nonopinion iard (1995), this assumption is unfounded leadership scale
leaders. because it presupposes that there are no How must an individual score on the
differences in affinity to explain the low in- opinion leadership scale to be regarded as
In accordance with the “two-step flow” terest on the part of advertisers for this opin- an opinion leader? Answering this ques-
or “multistage interaction” models, the ion leadership concept. This hypothesis tion is compounded by the fact that set-
views of opinion leaders are frequently requires further validation because consul- ting a threshold is a basically arbitrary
sought by their immediate environment tant research indicates that the opposite is exercise. However, our choice has been
on a given category of products. While true: according to Erdos and Morgan-MPG dictated by three types of considerations.
advertising is a source of information for (quoted by Advertising Age), the Wall Street First, in purely theoretical terms, we
the opinion leader, the latter would have Journal is “the publication most widely read assumed that the opinion leadership is
to refer to these products, if only to con- by opinion leaders in the USA.” Finally, not a dichotonomous condition but re-
vey his or her views on them. women’s consumer magazines are aimed flects a continuum of varying intensity
at carefully targeted audiences: some pub- (Weiman, 1994, p. 246). Consequently, with
H3: Women’s clothing fashion opin- lications will probably be more trendy than the allocation of scores on a metric scale,
ion leaders read more magazines others, and their editorial content more or we can calibrate the degree of leadership
than nonopinion leaders. Opin- less appealing to opinion leaders. For ex- from the usual indices used in descriptive
ion leaders also read more each of ample, “young” and “upscale” women’s statistics (median, quartile, decile): an in-
the categories of women’s maga- magazines are supposed to attract wom- dividual with a score in the initial upper
zines than do nonopinion leaders. en’s clothing fashion opinion leaders decile will be regarded as a “more pow-
more than do “people” or “mass consumer” erful” leader than a leader in the second
Opinion leaders are assumed to be ac- women’s magazines. or third decile.
tive information seekers and to read mag- The second consideration is based on
azines massively, especially those that deal H5: The degree of “qualitative” affin- pragmatism: a low selection limit (e.g.,
with current issues: “Leaders are more ex- ity among the women’s fashion the median) yields a marketing target that
posed to a variety of media source, espe- wear opinion leaders differs sig- is relevant in size but with little distinc-
cially news and information program” nificantly for magazine press titles tiveness while, on the other hand, a high
(Sheth, Mittal, and Newman, 1999). Accord- that belong to the same segment. limit (e.g., the initial centile) produced a
ing to Piirto (1992) “influentials’ heavy use high differentiated target but with too lit-
of print media makes them relatively easy This is a stricter assumption than the tle marketing potential. Consequently, the
to reach with advertising.” Moreover, re- two previous one, but it is necessary so as choice of limit requires a trade-off be-
search on fashion opinion leadership has draw the interest of marketing managers. tween these two extremes.
shown that female opinion leaders read While it is important to check that the Third, we have observed the choices
more fashion magazines than other opin- opinion leaders read categories of the made by researchers faced with this type
ion leaders (Reynolds and Darden, 1971; printed press than nonleaders, it is equally of decision. Faced with a similar issue
Summers, 1970). However, this point re- important to check that some titles are when having to determine a threshold in
quires confirmation because according to regarded as having more appeal than oth- an innovation scale, Goldsmith and Stith
Chan and Misra (1990), exposure to mass ers. In other words, women’s magazines (1992) have argued that scores in the up-
media produces a weak correlation with the should reveal significantly different im- per 12 percent range of the scale would
degree of opinion leadership. age profiles among opinion leaders: if this encompass the innovators. This limit does
not seem to be really selective because, 10 percent of the U.S. online adult items should have led to elimination of
according to Rogers (1983), innovators fall population. this incomplete data on statistical grounds.
into the top 2.5 percent of a population These major differences suggest that the However, in theory, there is good reason
that adopts innovations (average adop- upper 10 percent limit is a satisfactory for keeping individuals who abandon the
tion time less two standard deviations). rule of thumb in marketing terms and scale half way through because this is
Moreover, opinion leaders would fit into fairly reasonable in theory. Consequently, probably strongly correlated with a lack
the next 13.5 percent (mean, less the in- we argued that a little over 10 percent of of opinion leadership in the category. Elim-
terval between one and two standard de- a given population are potentially opin- inating incomplete scores (generally lower
viations). In the political sphere, the ion leaders in a given category of con- than the mean) would upwardly bias the
findings of the 1970 Kingdom study sumer products (i.e., “Top 10”). We have results and hence overvalue the average
(quoted by Weiman, 1994, p. 56) reveal prudently defined a second group of more level of opinion leadership of the popula-
that only 10 percent of the American pop- moderate opinion leaders who belong to tion. On the other hand, individuals who
ulation exhibit opinion leader characteris- the next 15 percent of the scale (upper 25 did not reply to any of the five items
tics. Moreover, this 10 percent limit finds percent—upper 10 percent) that we call were eliminated because it is hard to say
a broad consensus among marketing prac- the “Top 25.” This group will be used to whether they did not reply because they
titioners: Keller and Berry (2003) maintain support certain hypotheses when necessary. overlooked the question or because they
that “10% of Americans determine how The average French score on the opin- were genuinely indifferent to the category
the rest consume and live by chatting ion leadership scale is 11.5 points (see of products. However, the impact of these
about their likes and dislikes.” Similarly, Figure 1); the limit for the first quartile is adjustments remains negligible: the Top
Cakim (2002) argue that the “E-fluentials,” 15 points (Top 25) and 18 points for the 10 threshold remains unchanged (18 points
defined according to their intensive use first decile (Top 10). Since the scale is in both options) and rises to 16 points
of email, chat rooms, messages boards, comprised of five items, the fact that re- (instead of 15) for the Top 25 in the sec-
company and opinion websites, make up spondents failed to answer one of these ond scenario.
Typical profile of the women’s clothing ing parties, the importance given to Verification of hypotheses
fashion opinion leader friends.
The typical profile of the “fashion” opin- Not surprisingly, fashion-related items H1: Women’s clothing fashion opin-
ion leader in France is a 15- to 35-year- highlight significant differences from the ion leaders have a more favor-
old female employee or student: the rest of the French population. In a nut- able attitude toward media
age, gender, and job variables are all shell, being hip and fashionable is above advertising than nonopinion
statistically significant (see Table 1). In all a way for opinion leaders to feel dif- leaders.
terms of lifestyle, the values traditionally ferent; much more than choosing a dress
The results in Table 3 show that opin-
associated with a young audience domi- code to show that one belongs to a given
ion leaders favor media advertising more
nate: open-mindedness, wanderlust, throw- social group (see Table 2).
than nonopinion leaders: the mean scores
of opinion leaders decline according to
the degree of opinion leadership on each
of the four items measuring attitude to-
ward advertising. The four series of one-
TABLE 1 way ANOVA (with three levels, i.e.,
degrees of leadership), conducted on the
Socio-demographic Profile of Women’s Clothing Fashion
attitude toward advertising, confirm the
Opinion Leaders existence of statistically significant differ-
Nonopinion ences according to the degree of opinion
Top 10 Leader x2 Significant leadership. Hypothesis H1 is therefore
Sociodemographic Variables (%) (%) Statistics Levels validated (see Table 4).
.............................................................................................................................................................
15–19 years old 17 6 H2: Women’s clothing fashion opin-
...........................................................................................................................
20–24 14 6 ion leaders discuss media ad-
...........................................................................................................................
vertising more than nonopinion
25–34 19 15 Age leaders.
...........................................................................................................................
x 2 = 295.5 p < .001
35–49 23 27
........................................................................................................................... On the item “advertising is a frequent
50–64 17 22 topic of discussion,” the Top 10 opinion
...........................................................................................................................
leaders scored 2.43, the Top 25 opinion
65+ 11 24
............................................................................................................................................................. leaders scored 2.24, and nonopinion lead-
Male 6 51 Gender ers scored 2.02. The results of a one-way
...........................................................................................................................
x 2 = 653.8 p < .001 analysis of variance with a 3-level factor
Female 94 49
.............................................................................................................................................................
(degree of opinion leadership) reveal that
Farmers 1 1
........................................................................................................................... such differences are statistically signifi-
Craftsmen, small businessmen 4 3 cant and support Hypothesis H2: opinion
...........................................................................................................................
leaders like discussing advertisements
Executives, senior managers 4 6
........................................................................................................................... more than nonopinion leaders (see Table 5).
Middle managers 11 12
...........................................................................................................................
PCS H3: Women’s clothing fashion opin-
Office workers 24 15 x 2 = 360.7 p < .001 ion leaders read more categories
...........................................................................................................................
Manual workers 7 15 of female magazines than non-
...........................................................................................................................
opinion leaders. In general, wom-
Retirees 11 26 en’s clothing fashion opinion
...........................................................................................................................
Students 26 10 leaders also read more con-
...........................................................................................................................
sumer magazines than nonopin-
Unemployed 13 12
............................................................................................................................................................. ion leaders.
TABLE 2
Mean Values of Clothing Fashion Opinion Leaders
Nonopinion t Significant
Lifestyle and Values* Top 10 Leader Values Levels
................................................................................................................................................................................................................................................
“The same look allows me to identify with people
who are close to me and who think like me.” 1.98 2.03 −1.77 p < .10
................................................................................................................................................................................................................................................
“It is important to keep informed of world events.” 3.52 3.43 4.60 p < .01
................................................................................................................................................................................................................................................
“In my generation men and women shared a lot.” 3.19 3.07 5.20 p < .01
................................................................................................................................................................................................................................................
“I have a good relationship with my parents.” 3.36 3.33 1.41 n.s.
................................................................................................................................................................................................................................................
“I feel good about myself. I’m happy.” 3.18 3.16 .99 n.s.
................................................................................................................................................................................................................................................
“I am confident about my future.” 2.80 2.85 −2.01 p < .05
................................................................................................................................................................................................................................................
“I have excellent friends.” 3.52 3.32 9.05 p < .01
................................................................................................................................................................................................................................................
“It’s important to sow your oats before starting work.” 2.84 2.61 8.38 p < .01
................................................................................................................................................................................................................................................
“You have to live your passions.” 3.38 3.30 2.93 p < .05
................................................................................................................................................................................................................................................
“I feel original and different from other people.” 2.51 2.32 7.20 p < .01
................................................................................................................................................................................................................................................
“It is essential to be fashionable.” 2.24 1.98 10.21 p < .01
................................................................................................................................................................................................................................................
“It’s important to travel and discover the world.” 2.78 2.65 4.89 p < .01
................................................................................................................................................................................................................................................
*Each proposal is rated on a 4-point Likert scale (1 5 totally disagree, 4 5 totally agree).
“Life would be dull without advertising.” 2.39 2.25 2.10 ers” is noticeably higher among opinion
.............................................................................................................................................................
leaders than among nonopinion leaders
*Each proposal is rated on a 4-point Likert scale (1 5 totally disagree, 4 5 totally agree).
(Z 5 11.4, p , .001). The format of the
questionnaire does not make it possible to
directly determine whether opinion lead-
We have followed the classification ments (Table 6) are significantly higher ers read more consumer magazines than
adopted by women’s magazine profession- ( p , .001) among opinion leaders in the nonopinion leaders because this variable
als that divide women’s consumer maga- fashion industry than among nonopinion is measured on a nominal scale. However,
zines into “upscale,” “middle-market,” leaders, resulting in far higher affinity based on the midpoints in each range of
“popular,” and “youth and teenager.” The indices. This overweighting does not have values and on an estimated score of 10 for
penetration rates of these different seg- the same impact on different categories of the heavy reader segment, the average
TABLE 4
Analysis of Variance of Attitude toward Advertising
Sum of Degrees of Mean F
Attitudes toward Advertising Squares Freedom Squares Ratio
................................................................................................................................................................................................................................................
“I like advertising.”
Variance between groups 164.93 2 82.46 107.81*
................................................................................................................................................................................................................................................
Variance within groups 6,994.82 9,146 0.76
................................................................................................................................................................................................................................................
“I find advertising entertaining.”
Variance between groups 77.26 2 38.96 43.09*
................................................................................................................................................................................................................................................
Variance within groups 8,305.14 9,287 0.90
................................................................................................................................................................................................................................................
“I trust companies that advertise their products”
Variance between groups 71.95 2 35.98 64.57*
................................................................................................................................................................................................................................................
Variance within groups 5,137.65 9,287 0.56
................................................................................................................................................................................................................................................
“Life would be dull without advertising.”
Variance between groups 82.02 2 41.01 51.32*
................................................................................................................................................................................................................................................
Variance within groups 7,389.05 9,287 0.80
................................................................................................................................................................................................................................................
*p , .001.
TABLE 5
Analysis of Variance of Advertising Discussion and Opinion Leadership
Sum of Degrees of Mean F
Sources of Variance Squares Freedom Squares Ratio
................................................................................................................................................................................................................................................
“Advertising is a frequent topic of discussion.”
Variance between groups 165.47 2 82.77 110.56
................................................................................................................................................................................................................................................
Variance within groups 6,901.56 9,287 0.75 *
................................................................................................................................................................................................................................................
*p , 0,001.
TABLE 6
Penetration Rates of Magazine Segments
Top 10
Segments of Women’s Opinion Nonopinion Affinity Z
Consumer Magazines Leaders Leaders Indices* Test
................................................................................................................................................................................................................................................
Upscale and glossy women’s magazines 33% 15% 220 z = 14.6**
................................................................................................................................................................................................................................................
Middle market women’s magazines 58% 38% 152 z = 11.9**
................................................................................................................................................................................................................................................
People and popular women’s magazines 36% 20% 180 z = 10.7**
................................................................................................................................................................................................................................................
Youth and teenager women’s magazines 11% 6% 183 z = 6.28**
................................................................................................................................................................................................................................................
*(% opinion leaders Top 10/% nonopinion leaders) 3 100
**p , .001.
TABLE 7
Penetration Rates for Readership Segments
Top 10
Categories of Magazines Readers Opinion Nonopinion Affinity
(Immediate Readership) Leaders Leaders Indices*
................................................................................................................................................................................................................................................
Do not read consumer magazines (have not read any titles during the last reference period) 2.2% 4.9% 44
................................................................................................................................................................................................................................................
Rarely read consumer magazines (one to three titles read during the last reference period) 25.6% 37% 69
................................................................................................................................................................................................................................................
Read consumer magazines quite often (four to seven titles during the last reference period) 29.8% 33.2% 89
................................................................................................................................................................................................................................................
Frequently read consumer magazines (more than eight titles read during the last reference period) 42.4 24.6 172
................................................................................................................................................................................................................................................
Total 100 100
................................................................................................................................................................................................................................................
*(% of Top 10 opinion leaders/% of nonopinion leaders) 3 100.
number of titles is 6.11 for opinion lead- reading period. The reference period will not so great over a longer period of
ers versus 4.96 for nonopinion leaders— exceed the periodicity of the title, e.g., reading.
the difference is highly significant (t 5 over 8 days for a weekly magazine and
24.5, p , .001). These different findings over a month for a monthly magazine. H5: The degree of “qualitative” affin-
support Hypothesis H3. Table 8 lists the penetration rates for ity among the women’s fashion
immediate readership: the penetration in- wear opinion leaders differs sig-
H4: The degree of affinity that dices of different women’s consumer mag- nificantly for magazine press
women’s clothing fashion opin- azine titles are not strictly equivalent titles that belong to the same
ion leaders have with different between titles. A one-way ANOVA with segment.
media differs significantly for the four levels (segments) was run to see if
women’s magazines segments magazine titles significantly vary in terms We selected four common image items
concerned. of affinity indices for an immediate read- of reader’s magazine: overall image, trust,
ership situation. Results shown in Table 8 closeness, appeal. Table 10 compares
We have allocated the main women’s are highly significant at p , .001 (F ra- perceptions on the major four women’s
consumer magazine titles to either one of tio 5 20.9, with 3 and 24 degrees of free- fashion magazines among leaders and
the four abovementioned segments, in dom). The same analysis was conducted nonleaders.
keeping with the conventional media cat- with affinity indices calculated for the same A three-way ANOVA was conducted
egories used by professionals. Media plan- titles but over a longer reading period. on these image scores obtained for the
ners distinguish two major indices to assess Again, the F ratio is particularly signifi- different magazine titles. This analysis fully
the audience of a given medium. The cant (45.19, p , .001). This supports Hy- validates Hypothesis H5. Several interest-
first, called “immediate recent reader- pothesis H4. ing findings deserve mention:
ship,” encompasses the many readers who To estimate the dispersion within each
state that they held a copy of a weekly segment, we have calculated the varia- • First, there are no statistical differences
consumer magazine during the week fol- tion coefficients [(standard deviation/ in the way the four magazines are per-
lowing publication and during the month average) 3 100] of affinity indices for ceived by all readers ( p . .10), as sug-
following publication of a monthly mag- each segment and readership type (see gested by Table 10.
azine. The second index covers “long- Table 9). A coefficient above 25 expresses • On the other hand, the image scores
term” readers, i.e., the total number of a statistical series with highly dispersed attributed by leaders and nonleaders
individuals who have read an average observations: this is the case with imme- are very significantly different ( p 5
issue of a magazine, irrespective of the diate readership. The scatter of indices is .014).
TABLE 9
Variation Coefficients for
TABLE 8
Affinity Indices
Comparison of Penetration Rates (Immediate Readership)
According to Segments and Targets Long-
Immediate Period
% Penetration Segments Readership Readership
...........................................................................
for the % Penetration Affinity
Upscale 19.7 18.6
...........................................................................
Magazine Titles Segments Top 10 for Nonleaders Indices
.............................................................................................................................................................
Middle market 17.7 11.6
...........................................................................
Elle Upscale 12.3 4.2 293
.............................................................................................................................................................
Popular 21.8 11.5
...........................................................................
Madame Figaro Upscale 6.4 3.9 164
.............................................................................................................................................................
Teen and
Femme Upscale 2.1 1.1 184
............................................................................................................................................................. young adult 26.4 14.7
...........................................................................
Marie Claire Upscale 14.7 6.2 236
.............................................................................................................................................................
Marie France Upscale 7.3 3.7 197
.............................................................................................................................................................
Vogue Upscale 5.8 2.3 252
............................................................................................................................................................. • All the interactions are significant. The
Votre beauté Upscale 2.6 .9 276 most interesting finding is the very strik-
.............................................................................................................................................................
ing interaction between the magazine
Fémina Middle market 13.3 9.6 138
............................................................................................................................................................. titles and the leadership ( p 5 .001): in
Version femme Middle market 11.5 8.1 142
............................................................................................................................................................. marketing terms, this confirms that
Coté femme Middle market 2.0 1.6 125
............................................................................................................................................................. while the ways in which the magazine
Femme actuelle Middle market 31.0 18.0 174 are perceived are assumed to be similar
.............................................................................................................................................................
among all the respondents, the same is
Avantage Middle market 10.2 4.8 212
............................................................................................................................................................. not true among the leaders. Moreover,
Modes Travaux Middle market 15.2 7.7 196
............................................................................................................................................................. the interaction between the image items
Prima Middle market 13.2 7.8 168 and the leadership means that al-
.............................................................................................................................................................
though the leaders generally rate these
Gala Popular 9.0 4.0 225
............................................................................................................................................................. titles more positively than the nonlead-
Point vue Popular 3.2 1.8 174
............................................................................................................................................................. ers, the variances between both these
Maxi Popular 13.4 6.7 198
............................................................................................................................................................. targets vary according to the image items
Nous deux Popular 4.8 4.1 117 considered (see Table 11).
.............................................................................................................................................................
Voici Popular 15.6 6.7 231
............................................................................................................................................................. DISCUSSION AND IMPLICATIONS
France Dimanche Popular 6.3 4.8 131
............................................................................................................................................................. The fact that Hypotheses H1 and H2 have
Ici Paris Popular 7.6 4.3 176 been validated confirms the merit of de-
.............................................................................................................................................................
vising media plans that target “women’s
Girls Teenager and young adult 6.7 1.3 500
.............................................................................................................................................................
fashion” opinion leaders. Opinion leaders
Ok podium Teenager and young adult 3.8 1.0 365
............................................................................................................................................................. are, on the one hand, more exposed and
Salut Teenager and young adult 3.6 1.1 318
............................................................................................................................................................. more inclined to be in favor of advertis-
Star club Teenager and young adult 3.8 1.2 306 ing than nonleaders, and, on the other,
.............................................................................................................................................................
advertising is for them a more frequent
Jeune and Jolie Teenager and young adult 9.8 1.5 628
............................................................................................................................................................. topic for discussion than for nonleaders.
20 ans Teenager and young adult 9.1 1.5 590
............................................................................................................................................................. Our findings reveal how important it is to
Biba Teenager and young adult 5.8 1.5 394 leverage the audience exposed to an ad-
.............................................................................................................................................................
vertisement: hence, one single contact with
an opinion leader will naturally produce
Opinion leaders are, on the one hand, more exposed and gested that a media plan specifically geared
to opinion leaders was of little value, be-
more inclined to be in favour of advertising than nonlead- cause leader and nonleader targets alike
were equally reached by advertising
ers, and, on the other, advertising is for them a more messages.
Hypothesis H5, on validation, reveals
frequent topic for discussion than for nonleaders. that the opinion leaders have a far better
perception of top-of-the-range women’s
magazines than nonleaders. In other words,
“fashion wear” leaders not only read more
n (free) contacts in his or her area of foregone conclusion. But by far the most women’s magazines than nonleaders, but
influence. In concrete terms, the leader is encouraging result yet is that the affinity their appraisal is qualitatively superior:
more exposed to advertisements than the of leaders for magazines varies consider- they tend to trust them more, have more
nonleader because he reads more publica- ably from one title to another. For in- affinity with them, and find their articles
tions, such as men’s or women’s maga- stance, the leaders’ affinity for Elle more interesting than nonleaders.
zines; he naturally leverages his or her magazine is proportionally higher than In addition, the leaders have signifi-
contacts when talking to family and friends that for its competitors within the “up- cantly different qualitative affinities be-
about the advertisements he or she has scale segment”: 293 points as against 276 tween magazines, whereas the latter are
seen or read. Basically, whatever kind of points for its nearest competitor. It is there- aimed at an identical market segment.
influence model is used (two-step flow or fore possible to rank publications based For example, Elle magazine generates sig-
interactive), the fact remains that advertis- on the indices habitually utilized in me- nificantly greater qualitative affinities than
ing messages will induce interpersonal dia planning (coverage, effective audi- its competitors, both in terms of overall
communication between the opinion leader ence, affinity, etc.) so as to allocate budgets image (index 5 176) or on specific items
and his or her circle of friends and more effectively to various publications. (closeness: index 5 471, trust index 5 302,
acquaintances. In other words, although it is practically appeal index 5 352). Consequently, be-
Besides this first quantitative audience impossible to find one single title that is cause the leaders pay more attention to
lever, all messages targeted at opinion read by opinion leaders only (because 10 advertising than nonleaders, it may be
leaders are likely to be bolstered by a percent to 25 percent of a given popula- safely assumed that the “advertising re-
second lever, more qualitative in nature. tion may be termed opinion leaders), it is ceptivity” among the leaders varies sig-
Since information as conveyed during in- to be noted that the affinity index regard- nificantly between magazine titles. This is
terpersonal communication is deemed to ing the opinion leader target for two pub- an argument be prioritizing this target in
be more credible, advertising messages lications having the same readership varies media planning.
imparted by an opinion leader (who uses considerably. Our own findings moderate Moreover, although we have seen that
the message as one of his or her sources considerably the assumptions of Engel, overall the leaders read more women’s
of information) will carry with them stron- Blackwell, and Miniard (1995) who sug- magazines than the nonleaders, Table 12
ger conviction. Therefore, any advertise-
ment that the opinion leader believes in
will strengthen, in a first stage, his or her
confidence in the brand, then, in a second . . . [A]ny advertisements that the opinion leader be-
stage, will also strengthen that of his or
her circle of friends and family by word lieves in will strengthen, in a first stage, his or her con-
of mouth.
Validating Hypotheses H3 and H4 turns fidence in the brand, then, in a second stage, will also
out to be also of great interest. We have
first demonstrated that leaders read a strengthen that of his or her circle of friends and family
greater number of women’s magazines
than nonopinion leaders, which was a by word of mouth.
March 2004 JOURNAL OF ADVERTISING RESEARCH 101
TARGETING AND MEDIA PLANNING
TABLE 10
Comparison between Evaluated Perceptions on Women’s Magazines among Fashion Wear
Leaders and Nonleaders
Assessment Items Elle Madame Figaro Marie Claire Marie France
................................................................................................................................................................................................................................................
Strong image (population) 25.70 a 23.91 24.68 20.38
................................................................................................................................................................................................................................................
Leaders 42.23 30.68 35.79 25.57
................................................................................................................................................................................................................................................
Nonleaders 23.95 23.12 23.39 19.78
................................................................................................................................................................................................................................................
Affinity index 176.35 b 132.71 153.02 129.27
................................................................................................................................................................................................................................................
Trust (population) 5.33 5.76 5.66 4.14
................................................................................................................................................................................................................................................
Leaders 13.51 10.23 11.68 5.60
................................................................................................................................................................................................................................................
Nonleaders 4.47 5.24 4.96 3.97
................................................................................................................................................................................................................................................
Affinity index 302.54 195.26 235.59 141.00
................................................................................................................................................................................................................................................
Closeness (population) 2.79 2.97 2.73 1.54
................................................................................................................................................................................................................................................
Leaders 9.68 6.39 7.23 3.93
................................................................................................................................................................................................................................................
Nonleaders 2.05 2.57 2.20 1.26
................................................................................................................................................................................................................................................
Affinity index 471.45 248.83 328.18 311.01
................................................................................................................................................................................................................................................
Appealing (population) 6.20 6.48 4.94 2.88
................................................................................................................................................................................................................................................
Leaders 17.57 10.65 9.39 5.90
................................................................................................................................................................................................................................................
Nonleaders 4.99 5.99 4.42 2.53
................................................................................................................................................................................................................................................
Affinity index 351.90 177.90 212.32 233.26
................................................................................................................................................................................................................................................
a
% of individual who attribute the on-line item to the magazine.
b
(% of leaders who attribute the on-line item to the magazine/% of nonleaders who attribute the on-line item to the magazine) * 100.
reveals that all these titles do not have wear leader. For example, it is twice as and for the same titles (e.g., women, up-
the same appeal among the leaders: e.g., expensive to reach a leader in Madame per status, 25 to 49 year olds), we ob-
the effective audience varies by some Figaro than with Elle magazine. The two tained very similar costs.
five points between Elle magazine (27.60 challengers (Marie France and Marie Claire) Our findings tally with those published
percent) and Madame Figaro (22.6 per- are about 10 percent less expensive than in other studies that focused on other cat-
cent). Consequently, while it is logical to the leader Elle. (Readers interested in egories of opinion leaders in other coun-
assume that any women’s magazine title these issues should consult Vernette and tries (e.g., Vernette and Schmutz, 2003;
will necessarily reach a larger number fash- Schmutz, 2003.) Weiman, 1994): in generally, the magazine
ion wear leaders, a more selective choice It therefore is economically feasible to press enables accurate targeting of opinion
of these same titles secures substantial segment the titles according to power scales leaders.
readership gains on the opinion leaders and effective costs based on opinion lead- Figure 2 summarizes the average affin-
target. An inspection of advertisement ership. Finally, when the costs involved in ity gains in terms of the Top 10 leaders
insert costs confirms the merit of such contacting a fashion wear leader is com- target achieved by advertisement mes-
selectivity because there are substantial pared with those incurred for other tradi- sages inserted in the print magazine
variances in the cost of contact a fashion tional targets with equivalent potential press; we have followed the steps of
TABLE 11
Analysis of Variance of Title, Image Items, and Opinion Leadership
Sum of Squares
Source (Type I) df Mean Square F Sig.
................................................................................................................................................................................................................................................
Title magazines Hypothesis 4,669.651 4 1,167.413 4.857 .109
................................................................................................................................................................................................................................................
Error 741.371 3.085 240.341
................................................................................................................................................................................................................................................
Leadership Hypothesis 389.624 1 389.624 26.319 .014
................................................................................................................................................................................................................................................
Error 44.412 3 14.804
................................................................................................................................................................................................................................................
Image items Hypothesis 2,844.153 3 948.051 53.705 .001
................................................................................................................................................................................................................................................
Error 72.840 4.126 17.653
................................................................................................................................................................................................................................................
Leadership * title Hypothesis 78.313 3 26.104 16.423 .001
................................................................................................................................................................................................................................................
Error 14.305 9 1.589
................................................................................................................................................................................................................................................
Title * items Hypothesis 39.945 9 4.438 2.792 .071
................................................................................................................................................................................................................................................
Error 14.305 9 1.589
................................................................................................................................................................................................................................................
Leadership * items Hypothesis 44.412 3 14.804 9.314 .004
................................................................................................................................................................................................................................................
Error 14.305 9 1.589
................................................................................................................................................................................................................................................
The media planner can benefit from choosing opinion Childers, T. L. “Assessment of Psychometric
Properties of an Opinion Leadership Scale.”
leaders as a specific media target because this will indi- Journal of Marketing Research 23, 2 (1986): 184–88.
ion leaders. The media planner can benefit (namely New Developments and Approaches in Con- Dichter, E. “How Word-of-Mouth Advertising
from choosing opinion leaders as a spe- sumer Behavior Research, 1997, as coauthor). He is Works.” Harvard Business Review 44, November–
cific media target because this will indi- a member of the board of the French Marketing December (1966): 147–57.
(“Doctorat d’Etat”) at the University of Paris-Nanterre. Chan, K. K., and S. Misra. “Characteristics of Goldsmith, R. E., and R. Desborde. “A Validity
His research area is focused on consumer behavior, the Opinion Leader: A New Dimension.” Jour- Study of a Measure of Opinion Leadership.”
strategic marketing, and psychometrics. He has pub- nal of Advertising 19, 3 (1990): 53–61. Journal of Business Research 23 (1991): 362–71.
–——, and M. T. Stith. “The Social Values of ries.” Journal of Marketing Research 7, 1 (1970): –——, and W. D. Wells. Consumer Behavior.
Fashion Innovators.” Journal of Applied Business 43–50. New York: McGraw Hill, 1977.
Research 9, 1 (1992): 10–16.
Lipovetsky, G. L’empire de l’éphémère. Paris: Gal- Richins, M. L., and T. Root-Shaffer. “The
Gutman, J., and M. K. Mills. “Fashion Life- limard, 1987. Role of Involvement and Opinion Leadership
style and Consumer Information Usage: For- in Consumer Word-of-Mouth: An Implicit Model
mulating Effective Marketing Communications.” Marshall, R., and I. Gitosudarmo. “Varia-
Made Explicit.” Advances in Consumer Research
In Proceedings of American Marketing Association, tion in the Characteristics of Opinion Leader
15 (1988): 32–36.
B. Walker et al., eds., 1982. across Cultural Borders.” Journal of Inter-
national Consumer Marketing 8, 1 (1995): 5–12. Rogers, E. Diffusion of Innovations, 2nd ed. New
Herr, P. M., F. R. Kardes, and J. Kim. “Effects
York: The Free Press, 1983.
of Word-of-Mouth and Product Attribute Infor- McGuire, W. “Some Internal Psychological Fac-
mation on Persuasion: An Accessibility Diag- tors Influencing Consumer Choice.” Journal of Sheth, J. N., B. Mittal, and B. I. Newman.
nostic Perspective.” Journal of Consumer Research Consumer Research 2, 4 (1976): 302–19. Consumer Behavior. Orlando, FL: The Dryden
17, 4 (1991): 454–62. Press, 1999.
Mizerski, R. W. “An Attribution Explanation
Holmes, J. H., and J. D. Lett. “Product Sam- of Disproportionate Influence of Unfavourable
Smit, E. G., and P. C. Neijens. “Segmentation
pling and Word-of-Mouth.” Journal of Advertis- Information.” Journal of Consumer Research 9, 3
Based on Affinity for Advertising.” Journal of
ing Research 17, 5 (1977): 35–40. (1982): 301–310.
Advertising Research 40, 4 (2000): 35–43.
Consumer Behavior, James U. McNeal, ed. Ap- ple. New York: SUNY series, 1994.
Reynolds, F. D., and W. R. Darden. “Mutually
pleton, Century Crofts, 1969.
Adaptative Effects of Interpersonal Communi- Yale, L. J., and M. C. Gilly. “Dyadic Percep-
–——, and J. O. Summers, “Overlap of Opinion cation.” Journal of Marketing Research 8, 4 (1971): tions in Personal Source Information Search.”
Leadership across Consumer Product Catego- 449–54. Journal of Business Research 33, 3 (1995): 225–38.
APPENDIX
Measurement Scale of Women’s Clothing Fashion Leadership
(Adapted from Ben Miled and Le Louarn Scale, 1994)