|
Using Celebrity Endorsers to Increase Publicity Effects of Marketing
Communications
By
Xinshu Zhao, Ph.D.
Associate Professor
The University of North Carolina at Chapel Hill
School of Journalism and Mass Communication
CB # 3365, Carroll Hall
Chapel Hill, NC 27599-33645
Hyun Seung Jin, Ph.D. & Soontae An, Ph.D.
Assistant Professors
Kansas State University
School of Journalism and Mass Communication
105 Kedzie Hall
Manhattan, KS 66506-1501
Please address all correspondence to:
Hyun Seung Jin, Ph.D.
Assistant Professor
Kansas State University
School of Journalism and Mass Communication
105 Kedzie Hall
Manhattan, KS 66506-1501
Phone: 785-532-3959 Fax: 785-532-5484 E-mail: [log in to unmask]
The manuscript submitted to the Association for Education in Journalism and
Mass Communications
Using Celebrity Endorsers to Increase Publicity Effects of Marketing
Communications
Abstract
This paper investigates the effects of celebrity endorsement on the
effectiveness of a televised commercial in terms of 1) the publicity that
the commercial receives from the media, 2) TV viewers' attitude toward the
commercial and 3) TV viewers' memory of the commercial. The researchers
recorded the commercials aired during five Super Bowl broadcasts;
1992-1996, interviewed randomly selected viewers for their reactions toward
the commercials, and analyzed the news coverage of the commercials. Results
show that celebrity endorsement ads are more likable, more memorable and
draw more media attention. Beyond the main effects of celebrity
endorsement, news coverage and ad liking were identified as mediating
variables that increase brand recall.
Using Celebrity Endorsers to Increase Publicity Effects of Marketing
Communications
Featuring celebrities has long been a popular advertising technique. In the
1970s Gallup and Robinson (cited in Forkan, 1975) reported that 15 percent
of prime-time advertising featured celebrities. In the 1980s advertisers
used celebrities even more often (Alsop, 1987; Kaikati, 1987; Kamins,
Brand, Hoeke, and Moe 1989; Sherman, 1985), and this trend continued in the
early 1990s (Shani and Sandler, 1991). A recent estimate indicates that
about one-quarter of all commercials involve celebrity endorsements (Shimp,
2000).
Despite the popularity of using celebrities, there are many potential
hazards that have been documented. High costs are one reason why some
advertisers question the use of celebrities (Shani and Sandler, 1991). An
endorser's image can fail to mesh with the intended image of a product or
fail to appeal to the taste of the target audience (e.g., McCraken, 1989).
Celebrities may lose popularity or may disappear from the media spotlight
(Ziegel, 1983). Celebrities can negatively influence consumers' perceptions
of endorser credibility and attitude toward the ad when they endorse too
many products, thereby becoming overexposed (Tripp, Jensen, and Carlson
1994). Scandals can cheapen celebrities' images and subsequently damage the
endorsed products (Klebba and Unger, 1982; Shimp, 2000; Till and Shimp,
1998). After having bad experiences with celebrity endorsers, some
corporate executives consider the use of celebrities to be a risky business
and have vowed not to enter the celebrity arena again (Swenson, 1987; Ward,
1989).
In spite of such disadvantages, ads involving celebrity endorsements
continue to increase. One reason is the heightened attention given such ads
from the audience. Due to the celebrities' high profile, the endorsed ads
stand out from the commercial clutter, thus improving their ability to be
recalled by the audience. In addition, consumers like celebrity endorsement
ads (Atkin and Block, 1983; Sherman, 1985). Because of these perceived
advantages, advertisers who are unable to afford celebrities use
look-alikes and sound-alikes, despite considerable legal risks (Blanket,
1981; Borchard, 1985; Light and Tilsner, 1993; Major, 1989). Some
advertising practitioners and scholars consider celebrity endorsement "a
powerful marketing tool" (Swenson, 1987) or "the ultimate marketing ploy"
(Miller, 1989).
Celebrity figures can also generate extensive publicity for the endorsed
brands by capturing both consumer and media attention, leading to double
exposure to consumers of the same brands through both news stories and
commercials (Erdogan, Baker, and Tagg, 2001; Harris, 1998; Pasadeos,
Phelps, and Lamme, 2000). The announcement of a celebrity endorsement
contract widely publicized in the press is used as information by market
analysts to evaluate the potential profitability of endorsement
expenditures, thereby affecting the firm's expected return. Agrawal and
Kamakura's (1995) analysis showed that, on average, the impact of these
announcements on stock returns was positive; hence the cost of celebrity
endorsement should be viewed as worthy investment.
While there have been many speculations and anecdotal examples of media
publicity associated with celebrity endorsement (Agrawal and Kamakura,
1995; Harris, 1998), empirical evidence is scant. Also, most studies on the
effects of celebrity endorsers have been conducted in experimental
laboratories (cf. Agrawal and Kamakura, 1995). Experimental controls and
manipulations produce an unnatural viewing environment and psychological
conditions, and often require specially created stimuli that are very
different from real commercials. Generalizations of laboratory findings to
the general public should be viewed as tentative until they are
cross-verified with studies using naturalistic designs (McQuarrie, 1998;
Wells, 1993; Zhao, 2002).
The objective of this study is to empirically investigate how celebrity
endorsement for the advertisements aired during five Super Bowl games
(1992-1996) affected media attention to, viewers' attitude toward, and
viewers' memory of the commercials under natural viewing conditions.
Celebrity Effects on News Coverage of the Endorsed Ads
In America we live in a celebrity culture (Freeman, 1999; Golway, 2002).
People pay more attention to stories about famous people and news media
have catered to this public taste. Golway (2002) argues that "The nation's
news and entertainment executives believe that the surest way to readers,
and therefore to profits, is to tell the stories of the famous, over and
over again if necessary" and "there seems little doubt that whatever else
has changed, the American media remain transfixed – or at least believe
that the public remains transfixed – by celebrity" (p. 6). Brookhiser
(1998) declares that "Celebrity conquers America" (p. 30).
From a business point of view, there is a sure advantage when celebrities
appear in media content. Celebrities have tangible features that can be
advertised and marketed (e.g., body, face, voice, a certain kind of
personality, etc.), regardless of the type of media content that features
celebrities – news, entertainment, or ads. That's why celebrities are
widely regarded as vital capital and an investment, both in Hollywood
(Dyer, 1982) and Madison Avenue (Agrawal and Kamakura, 1995; Atkin and
Block, 1983; Sherman, 1985).
As the media responds to the audience, celebrity endorsement itself often
becomes news. Michael Jackson signing a contract with Pepsi made headlines.
So did Tiger Wood's $100 million deal with Nike. The "milk mustache"
campaign, introduced in 1995, garnered a huge windfall of publicity for the
ads, which featured several celebrities (Harris, 1998). Agrawal and
Kamakura (1995) found 110 news stories about celebrity endorsement
contracts in some selected media content from 1980 to 1992. Pasadeos et al.
(2000) also found that celebrity endorsers were one of the key elements
that major daily newspapers dealt with when the papers carried advertising
stories.
It is not unusual for advertising issues to be treated as business or
entertainment news. Regular columns on advertising have appeared in major
metropolitan daily newspapers. The New York Times carries about 700
advertising news stories per year (Pasadeos et al., 2000). On average,
advertising stories appear once a week in network television news (Keenan,
1995). Advertising practitioners realize the value of working with their
public relations counterparts; recognizing that publicity about ads can
increase the impact of their persuasive messages.
However, there is much information generated by advertisers about ad
campaigns and products that is ignored by the news media. Information
provided to news media, therefore, should have some news value or it likely
will be ignored. Pasadeos et al. (2000) suggested "interesting" as an
important criterion impacting the newsworthiness of an advertising
campaign. In light of the values used by news gatekeepers, an ad campaign
would be more newsworthy – more interesting – if it involved a celebrity
endorser. Many advertising practitioners agree that celebrity endorsers
generate "extensive PR leverage for brands" (Erdogan et al., 2001, p. 39).
Yates (1995) also notes that successful ad campaigns hit headlines and
celebrity endorsers made them headline-worthy. Based on the news value of
celebrities and their impact on media publicity, we generated the following
hypothesis:
H1: Celebrity endorsed commercials tend to generate more news coverage than
commercials without celebrities.
Celebrity Effects on Ad Attitude and Ad Recall
In celebrity effect studies, as in other studies of advertising effects,
consumers' attitude toward an ad often has been used as a main dependent
variable (Atkin and Block, 1983; Friedman, Termini, and Washington, 1976;
Kamins, 1989; Tripp, Jensen, and Carlson, 1994). Such studies, however,
have produced mixed results. Friedman et al. (1976) reported a more
favorable attitude toward celebrity ads than non-celebrity ads, but the
difference was not statistically significant. Atkin and Block (1983) found
statistically significant effects indicating a more favorable attitude
toward celebrity ads. While Freiden (1984) reported a favorable attitude
toward celebrity ads in some situations, he also reported that, in other
situations, non-celebrity endorsements fared better; however, the
differences were not statistically significant. Friedman and Friedman
(1979) reported a statistically significant interaction effect, under which
celebrity figures resulted in more favorable attitudes for some products. A
more recent experiment by Kamins (1989) reported a negative, albeit
statistically non-significant, effect of celebrity endorsement.
The empirical evidence appears contradictory and weak, and there is also
no consensus on what theories may sufficiently explain or predict the
effects of celebrity endorsement. Early researchers relied on source
credibility and source attractiveness theories (e.g., Atkin and Block,
1983; Friedman and Friedman 1979). According to those theories, a message
is more effective when a source, or an endorser in the case of advertising,
is perceived as credible (trustworthy and believable) and attractive
(familiar, likable and similar to a message receiver). McCracken (1989),
arguing that these two elements alone could not explain the whole process,
proposed a meaning transfer theory. According to this theory, the image of
a celebrity endorser carries a certain cultural meaning in the minds of
consumers. This image may be transferred to an advertised product in the
process of advertising. The celebrity, therefore, can help a marketer when
the meaning is desirable and the transfer successful.
When a celebrity figure endorses a brand in an advertisement, one may
speculate that the meaning is transferred not only from the endorser to the
brand, but also transferred from the endorser to the advertisement and from
the advertisement to the brand. As the advertisement acts as the medium
during a transfer, it should be easier to detect the impact on ad attitude
than to detect the impact on brand attitude. If there is only a small
effect on attitude toward an ad, there might be an even smaller effect on
attitude toward a brand.
The weak empirical evidence does not necessarily point to a lack of
celebrity effects, but may instead be the result of methodological
limitations. The earlier experiments (e.g., Buhr, Simpson, and Pryor, 1987;
Friedman and Friedman, 1979; Kamins, 1990; Misra and Beatty, 1990; Mowen
and Brown, 1981) used print ads as the stimulus, and the ads were produced
or altered particularly for the purposes of the experiments. The typical
design employed one brand or one product type per experimental condition
and studied students as subjects. Textbook writers and other researchers,
including those who have studied celebrity effects (e.g., Fletcher and
Bowers, 1991; Friedman and Friedman, 1979; Haskins and Kendrick, 1993) have
often cautioned against generalizing results from such designs to the
"real" television ads in the "real" world.
If we are to accept the premises of the meaning transfer theory, then it is
important to further test for celebrity effects using a more naturalistic
design. The transfer of attitude may have a better chance to succeed when
professionally made, high-quality advertisements are used, and the
unusually high involvement level in a laboratory room may artificially
facilitate or hinder the transfer. Also, the student subculture may nurture
meanings that are different from those perceived by the general population
(Wells, 1993).
The following hypothesis is based on the above discussion and two
additional assumptions. First, most advertisers during Super Bowl games
were able and careful enough to select spokespeople who were attractive,
popular and likable. Second, those advertisers were able to produce
effective advertisements that enabled positive attitudes to be transferred
from the celebrities to the advertisements. Hence our hypothesis is:
H2: Celebrity endorsement commercials tend to generate a more favorable
attitude than
commercials without celebrities.
Among about two dozen empirical studies of celebrity endorsement, only a
few have examined the impact on consumer awareness or memory. Kamen, Azhari
and Kraph (1975) reported that consumers' awareness of Amoco's advertising
campaign increased after the company introduced John Cash as a spokesman.
In Friedman and Friedman's (1979) experiment using print ads, subjects
demonstrated a higher recall for celebrity ads than for non-celebrity ads.
Ray (1982) cited several Gallup and Robinson studies showing a 12 percent
greater recall rate for ads endorsed by celebrities, although little is
known about the designs, measurements or other details about those
unpublished studies.
Additional study is needed to elucidate the effects of celebrity
endorsement on ad recall, if only because of the scarcity of empirical
evidence. The lack of recent evidence is particularly disturbing given the
significantly intensified competition in the past decade among media,
vehicles and advertisements (Townsend, 1988). Practitioners and researchers
often cite celebrities' power to attract audience attention, which would
help a commercial stand out in an increasingly cluttered environment
(Erdogan et al., 2001; Kaikati, 1987; Ward, 1989). We will call this line
of thinking "attention attraction." When a popular person appears in an
advertisement, the audience is more likely to be exposed to the
advertisement. If the meaning transfer is successful, the advertisement
should carry some of those meanings; therefore, viewers should be more
likely to cognitively process the message.
For better generalizability, it would be desirable to use multiple
products and brands, and to test the effects in a natural viewing
environment. That is particularly important because, according to
"attention attraction" theory, attention is a mediating variable. The often
used controlled experiment and print stimuli tend to fix subjects'
attention at a certain (typically very high) level, which could reduce or
even eliminate the celebrity effects. Based on the above discussion
regarding attention attraction by the use of celebrity endorsers, we
developed the following hypothesis:
H3: Celebrity endorsement commercials tend to generate higher brand memory
than
commercials without celebrities.
Method
Research Design: Some researchers (e.g., Pieters and Bijmolt, 1997; Youn
et al., 2001; Zhao, 1997; Zhao 2002) argue that, aside from the typical
laboratory experiments and questionnaire surveys, many gigantic experiments
are run everyday – with tens of thousands of television ads as the stimuli,
millions of audience members as the subjects and thousands of advertisers
as the experimenters. All three major elements of television advertising –
advertisements, audience and viewing environment – are natural. Most of the
costs, from those associated with advertising production to fees for media
placement, are paid for by the advertisers. This is not, of course, a
controlled experiment. The subjects are not randomly assigned into
different conditions; nor are the stimuli counterbalanced. This is,
therefore, a quasi-experiment (Cook and Campbell, 1979) and various field
study techniques can be used to take advantage of its strengths and address
its weaknesses.
In this study, the annual Super Bowl game is considered as a
quasi-experiment. The Super Bowl is probably the most visible advertising
event of the year. Not surprisingly, the games have become a point of
interest for advertising researchers (Pavelchak, Antil, and Munch, 1988;
Youn et al., 2001; Zhao, 1997). Some national advertisers use the game to
release new ads, which are often very high quality. Local advertisers buy
"spot spaces" from a network's local affiliates. These spaces are usually
in the pre-game broadcast, half-time entertainment show, and post-game
interviews. The quality of the ads tends to be much lower than that of the
national ads broadcast during the four quarters of a game.
Of course no single event is representative of all viewing situations. But
among single events, the Super Bowl game probably has the most
representative audience and presents a wide range of advertising qualities.
Super Bowl viewing ratings are usually high, allowing for cost-efficient
telephone interviews to be conducted. In this study, advertised brand
recall and attitude toward the ads were measured through survey interviews;
results were then aggregated across respondents after the games.
Telephone interviews were conducted on Monday through Thursday evenings
following each Super Bowl from 1992 through 1996 (total of five games).
Graduate and undergraduate students enrolled in research classes at a major
university conducted telephone surveys of local residents in Orange County,
North Carolina, which includes the cities of Chapel Hill, Carrboro
Hillsborough, Mebane, and Pittsboro. Random digit dialing was used to
include unlisted numbers. The interviewers asked for the person who had the
next birthday in the household. If a call yielded no answer, that number
was re-dialed at least three times before being discarded. In some years we
also interviewed respondents who did not see the Super Bowl broadcast;
however, the analysis reported in this paper is based only on responses
from those who watched at least part of the broadcast. The total sample
included 2,225 adults: 254 for 1992, 327 for 1993, 547 for 1994, 590 for
1995 and 507 for 1996. The survey response rates were 56.9 percent for
1992, 65.3 percent for 1993, 65.6 percent for 1994, 57 percent for 1995,
and 65 percent for 1996.
Demographic characteristics of the respondents were representative of the
local population. Each year, more than half, sometimes more than 70
percent, of the respondents watched the game, supporting our speculation
that the Super Bowl is not a regular sporting event – non-sports fans also
watch it. There were no significant demographic differences between those
who watched the games and those who did not. No respondents knew beforehand
that we would be conducting the interviews, so the viewing situation was
completely natural.
Dependent Variables: For advertised brand recall, interviewers asked each
respondent whether he or she had watched the Super Bowl game and how much
of it was watched. Those who watched any part were then asked to list all
advertisements they remembered seeing during the game. Two coders
separately coded the responses, which had been recorded verbatim during the
interviews. The two sets of results were in agreement in all but one case
(more than 99 percent). The unaided recall rates were then calculated as a
percentage of the people who mentioned an advertised brand.
Lutz (1985, p. 46) Defined attitude toward an ad as "a predisposition to
respond in a favorable or unfavorable manner to a particular advertising
stimulus." For this study, attitude toward an ad was assessed by measuring
the overall degree of liking. Respondents were given a list of the ads that
have been advertised during the Super Bowl broadcast, and those who
remembered seeing an ad were asked how good or poor they thought the ad
was. Liking was rated on a scale of 0 to 100, with 0 being the worst and
100 the best. Experimental studies and conventional surveys generally use
multiple items to measure attitudes. However, the limited length of a
telephone interview and the relatively large number of brands involved in
this study forced us to use only this single-item measure. This means that
an inter-item reliability could not be calculated. Reliability should not
be a major issue since liking was aggregated across hundreds of
respondents. That is, the large number of respondents should minimize the
impact of any individual errors.
To assess news coverage of the ads, we conducted content analysis on
selected news content. Included in the sample were three local newspapers;
Chapel Hill News, News & Observer of Raleigh, and Herald Sun of Durham, and
three national newspapers; USA Today, The New York Times, and the Wall
Street Journal, each of these three enjoys a relatively large circulation
(Marketer's guide to media, 1996) as well as national respect as a news
source. Based on circulation and probability of news coverage of the Super
Bowl ads, seven national magazines were selected: Reader's Digest, TV
Guide, Time, Newsweek, U.S. News & World Report, People, and Sports
Illustrated. Also included in the sample were news programs of three
television networks, ABC, CBS, and NBC, and one cable newsnetwork CNN.
In each of the five years, all news content in the selected media for the
two weeks before each Super Bowl Sunday was analyzed. For national
newspapers and magazines, the search was conducted using the Lexis/Nexis
data base and the hard copies of the magazines. For the local newspapers,
CD ROMs produced by the local papers were searched. For the three TV
networks and CNN, a news database from "Television News Archive" at
Vanderbilt University (http://tvnews.vanderbilt.edu/) was used. The coding
unit for the content analysis was "frequency of brand names mentioned in
news stories on the Super Bowl ads." Therefore, every brand advertised
during each of the five Super Bowl broadcasts had its own numeric value.
Independent Variable: In this study, a celebrity figure was operationalized
according to Friedman and Friedman's (1979, p. 63) definition: an
individual who is known to the public for his or her achievement in areas
other than that of the product class endorsed." By analyzing the
commercials taped during the games, coders placed the advertised brands
into two categories: Celebrity endorsed (coded "1" for later analysis) and
non-celebrity endorsed (coded "0").
Control Variable: The frequency, namely the number of ads a brand places in
a Super Bowl broadcast, may affect the recall of the brand (Zhao, 1997). At
the same time, a brand that places more ads may have more chances of having
celebrity endorsers than a brand airing fewer ads, creating a confounding
effect when trying to assess the effect of celebrity on recall. That is, a
celebrity-endorsed brand might be remembered better simply because it was
advertised more often, not necessarily because of the celebrity
endorsement. Also, as mere exposure theory (Zajon, 1968) posits, repeated
exposure of an ad could be positively related to ad liking, creating a
similar confounding effect on ad liking. Ad frequency, therefore, was
controlled in our data analysis.
Analysis and Findings
There were more than 300 brands advertised in the five Super Bowls; only
newly telecasted national brand advertisements were selected for analysis.
Old ads re-run by some advertisers and ads for network television program
promotions were excluded from the data analysis. Brand memory and ad liking
may be affected by the fact that the commercials had been previously used,
and therefore possibly viewed, before the Super Bowl games. Also,
promotional messages from the network hosting the Super Bowl telecast were
not considered as a typical brand advertisement. Advertising Age, USA
Today, and other sources were used to determine which ads were new and
nationally televised. The final sample consisted of 136 advertised brands;
all of which appearing for the first time during the Super Bowl games.
Among them, 28 commercials used celebrity endorsers.
Descriptive statistics are presented in Table 1. The amount of news
coverage varied significantly from brand to brand. About 73 percent (99 out
of 136 brands) of the brand ads were mentioned at least once in a news
story, whereas 27 percent of the ads did not have any news coverage. The
highest number of news mentions for a brand was 144; the average across all
brands was 8.6. This wide range and skewed distribution shows that some ads
drew much more media attention than others. The response range for ad
liking (0-100 scale) was much narrower, with the most likable ad averaging
91.6 and the least likable ad 41.2. Overall, average ad liking was 61.2.
Average ad recall rate (0-100 scale) was 5.9 percent. The most memorable
brands had a recall rate of 56 percent; about 30 percent of brands were not
recalled by any respondent.
Examination of univariate normality indicated that brand recall and news
coverage had relatively high kurtosis values. Due to the fact that group
mean comparisons can be severely biased by large numbers in each group,
data transformation was necessary to reduce the problem. Thus, logarithm
transformations were performed for the two variables and transformation
substantially reduced both skewness and kurtosis. Both transformed and
non-transformed data were used for testing hypotheses.
Hypotheses Testing: One way multivariate analysis of covariance (MANCOVA)
was run on the set of dependent variables (brand recall, ad liking, and
news coverage) with group membership (celebrity endorsement ads vs.
non-celebrity ads) as the independent variable and ad frequency as the
control variable. Results from the MANCOVA showed that celebrity
endorsement had significant effects on the set of dependent variables after
controlling for ad frequency (Wilks' ? = .78; p < .001). Separate
univariate ANCOVAs were subsequently run to better elucidate the effects of
celebrity endorsement on each of the dependent variables.
Celebrity effects on all three dependent variables were significant (Table
2). Average news coverage score of the celebrity endorsement ads was
significantly higher than that of the non-celebrity ads (12.96 vs. 6.19; F
= 5.31; p < .05). Likewise, celebrity endorsement ads were more likable
(68.23 vs. 59.40; F = 20.55; p < .001) and memorable (17.62 vs. 2.84; F =
34.82; p < .001) than non-celebrity ads. Similar patterns were observed for
both transformed and non-transformed data. Thus, all three hypotheses were
supported by the data.
Relationships among Dependent Variables: The three dependent variables in
the study are interrelated. Brand recall showed a positive relationship
with news coverage (r = .56; p <.001) and ad liking (r = .54; p <.001).
Variations in a particular variable (e.g., brand recall) can be directly
influenced by news coverage or ad liking beyond the presence of celebrity
in ads. For example, with regard to the role of news coverage, Jin (2000)
showed a causal link between news coverage and brand memory in a study
examining how pre-exposure to publicity about ads affected recall of the
subsequent ads. He found that subjects who were exposed before the Super
Bowl game to a news story about upcoming Super Bowl ads had significantly
higher post-game recall of the ads than control subjects who did not read
the news story. The positive relationship between ad liking and memory has
been well documented in the advertising literature (e.g., Lutz, 1985;
MacKenzie, Lutz, and Belch, 1986). An ad that is liked creates a positive
feeling, which in turn improves recall of the ad because more attention is
given to the target ad at the time of encoding. On the other hand, viewers
may avoid cognitive processing when they do not like ads. Figure 1
summarizes the causal relationship between the major variables in question:
Celebrity endorsement directly affects news coverage, ad liking and brand
recall; in turn, news coverage and ad liking affect brand recall.
MANCOVA and univariate ANCOVAs are designed to disentangle the nature and
process of the interrelated mediating or dependent variables. Thus,
step-down analysis was conducted. By examining dependent variables in a
predetermined order, the step-down analysis assesses the role of mediating
variables. As presented in Figure 1, a dependent variable, brand recall,
was tested in univariate ANCOVA by adding news coverage and ad liking as
covariates. The results suggest that news coverage (F = 14.4; p <.001) and
ad liking (F = 27.9; p <.001) both had a significant impact on brand
recall, indicating that those two variables are mediating in nature (see
Table 3).
In sum, the results showed that celebrity endorsement ads drew more media
attention, were more likable, and more memorable than non-celebrity ads.
Beyond the main effects of celebrity endorsement, news coverage and ad
liking were mediating variables that helped increase brand recall.
Discussion
Employing a naturalistic quasi-experimental design, this study found
evidence in support of the hypotheses that celebrity endorsement during
five Super Bowl games (1992-1996) generated greater media attention to, and
more favorable attitude toward, commercials shown during the game, as well
as contributing to greater recall of the endorsed brand. Celebrity figures
help to increase brand memory, as attention attraction theory would
predict; and the celebrity endorsement led to more favorable attitude, as
meaning transfer theory would predict. It is apparent that celebrity
endorsement helps the commercials capture consumer attention as well as
media attention. Considering the news values employed by reporters and
editors, an ad campaign would be more newsworthy if it is associated with a
celebrity endorser. In turn, exposure of an ad via a news story further
increases brand recall (Jin, 2000).
During the Super Bowl games, commercials with celebrity endorsers brought,
on average, about three times more news coverage than non-celebrity ads.
The data support the observations of scholars and practitioners that
celebrity endorsement has substantial news value. In addition, the
celebrity endorsements ads had 15 percent higher ad liking scores and about
six times greater advertised brand recall rate than non-celebrity ads.
Compared with the previous findings, the estimations of ad liking and brand
recall appear large – particularly with regard to brand recall. As the
advertising environment has become increasingly crowded, celebrities appear
to be playing a larger role in helping advertisements stand out.
Most of the prior studies assessing celebrity endorsement effects were done
in an experimental environment using print ads; therefore, the subjects'
attention level could have been higher than in a natural TV viewing
situation. The fact that attention varied naturally in our study may
partially explain the larger effects we observed. More importantly, the
news stories about the celebrity endorsement ads also helped to increase
brand memory beyond the presence of celebrity in the ads, as seen in Figure
1. The data support those who believe in celebrity endorsement – it can,
indeed, be a powerful marketing tool.
The Super Bowl represents a unique sporting, media, advertising, and social
event in U.S. culture. Further investigation is needed to empirically
determine the effects of celebrity endorsement in other advertising
situations. Identification of situations in which celebrity endorsement is
particularly effective, or ineffective, will help advertisers decide when
the investment in celebrity endorsers is worth the cost and potential risk
of failure.
Field studies such as this one typically are weaker than controlled
experiments in making definitive causal inference; however, the study did
allow use of a large number of real ads, seen under natural viewing
conditions, and was relatively inexpensive to conduct. More convincing
evidence may be produced in future studies that employ different controls.
More studies conducted under less controlled viewing conditions are also
needed to study effects on attitude toward brand and buying intention, to
investigate whether the findings of this study can be generalized to a
national audience or non-Super Bowl/sports events, and to test different
kinds of effects on different audience groups or for different product
types. Such studies would add to our collective knowledge about the effects
of celebrity endorsement. Further, when such knowledge is based on evidence
from both the controlled experiments and field studies, it would have the
strengths of both and without the weaknesses of either.
References
Agrawal, Jagdish, and Wagner A. Kamakura. "The Economic Worth of Celebrity
Endorsers: An Event Study Analysis." Journal of Marketing 59, 3 (1995): 56-62.
Alsop, Ronald. "Does Sports Marketing Make Sense?" Wall Street Journal,
October 22, 1987.
Atkins, Charles, and Martin Block. "Effectiveness of Celebrity Endorsers."
Journal of Advertising Research 23, 2 (1983): 57-61.
Blanket, L. "Lookalike Bab Irks Admen, Reflects FACTS Guidelines."
Advertising Age, May 11, 1981.
Borchard, W.M. "Use Look-Alikes with Care, Court Warns." Advertising Age,
September 2, 1985.
Brookhiser, Richard. "Celebrity Conquers America." American Heritage 49, 1
(1998): 30-35.
Buhr, Thomas A., Terry L. Simpson, and Burt Pryor. "Celebrity Endorsers'
Expertise and Perceptions of Attractiveness, Likeability, and Familiarity,"
Psychological Reports 60 (1987): 1007-1309.
Cook, Thmoas D., and Donald T. Campbell. Quasi-Experimentation: Design and
Analysis Issues for Field Setting. Chicago: Rand McNally College Pub. Co.,
1979.
Dyer, Richard. Stars. London: BFI publishing, 1982.
Erdogan, B. Zafer. "Celebrity Endorsement: A Literature Review." Journal of
Marketing Management 15, 4 (1999): 291-314.
Erdogan, B. Zafer, Michael J. Baker, and Stephen Tagg. "Selecting Celebrity
Endorsers: The Practitioner's Perspective." Journal of Advertising Research
41, 3 (2001): 39-48.
Fletcher, Aland D., and Thomas A. Bowers. Fundamentals of Advertising
Research, 4th ed. Belmont, California: Wadsworth. 1991.
Forkan, James P. "Commercial Actors Squeezed by Stars, 'Real People'."
Advertising Age, November 17, 1975.
Freeman, Laurie. "Relating Publicly Has Great Value, Great Downside."
Marketing News, April 12, 1999.
Frieden, Jon B. "Advertising Spokesperson Effects: An Examination of
Endorser Type and Gender on two Audiences." Journal of Advertising Research
24, (1984): 33-41.
Friedman, Hershey H., and Linda Friedman. "Endorser Effectiveness by
Product Type." Journal of Advertising Research 19, 5 (1979): 63-71.
Friedman, Hershey H., Salvatore Termini, and Robert Washington. "The
Effectives of Advertisements Utilizing Four Types of Endorsers." Journal of
Advertising 6, 2 (1976): 22-24.
Golway, Terry. "Bold-Face Lives: Media Coverage of Celebrities and Ordinary
People." America, January 7, 2002.
Harris, Thomas L. Value-Added Public Relations: The Secret Weapon of
Integrated Marketing. Lincolnwood, IL: NTC Business Books, 1998.
Haskins, Jack B. and Alice Kendrick. Successful Advertising Research
Method. Lincolnwood, IL: NTC Business Books, 1993.
Jin, Hyun Seung. "Advertising and the News: Does Advertising Campaign
Information in News Stories Improve the Memory of Subsequent
Advertisements?" Unpublished dissertation, The University of North Carolina
at Chapel Hill, 2000.
Kaikati, Jack G. "Celebrity Advertising: A Review and Synthesis."
International Journal of Advertising 6, 2 (1987): 93-105.
Kamen, Joseph, Abdul Azhari, and Judith Kragh. "What a Spokesman Does for a
Sponsor," Journal of Advertising Research, 15, (1975): 17-24.
Kamins, Micheal A. "Celebrity and Non-Celebrity Advertising in a Two-Sided
Context." Journal of Advertising Research 29, 3 (1989): 34-42.
Kamins, Michael A. "An Investigation into the 'Match-Up' Hypothesis in
Celebrity Advertising: When Beauty May be Only Skin Deep." Journal of
Advertising 19, 1 (1990): 4-13.
Kamins, Michael A., Meribeth J. Brand, Stuart A. Hoeke, and John C. Moe.
"Two-Sided Versus One-Sided Celebrity Endorsements: The Impact on
Advertising Effectiveness and Credibility." Journal of Advertising 18, 2
(1989): 4-10.
Keenan, Kevin L. "Television News Coverage of Advertising: An Exploratory
Census of Content." In Proceeding of the 1995 Conference of the American
Academy of Advertising, 1995.
Klebba, Joanne M., and Lynette S. Unger. "The Impact of Negative and
Positive Information on Source Credibility in a Field Setting." In Advances
in Consumer Research, Vol. 10, R. P. Bagozzi and A. M. Tybout, eds. Provo,
UT: Association for Consumer Research, 1982.
Light, Larry, and Julie Tilsner. "Do You Wanna Dance with Lawyers?"
Business Week, October 18, 1993.
Lutz, Richard J. "Affective and Cognitive Antecedents of Attitude Toward
the Ad: A Conceptual Framework." In Psychological Processes and Advertising
Effects: Theory, Research, and Application, L. F. Alwitt and A.A. Mitchell,
eds. Hillsdale, NJ: Lawrence Erlbaum Associates, 1985.
MaCkenzie, Scott B. Richard J. Lutz, and George E. Belch. "The Role of
Attitude Toward the Ad as a Mediator of Advertising Effectiveness: A Test
of Competing Explanations." Journal of Marketing Research 23, 2 (1986): 130-43.
Major, Michael J. "Bette Wouldn't Sing, But Court Says Y&R Still Owes Her
Supper." Marketing News, November 6, 1989.
McCracken, Grant. "Who is the Celebrity Endorser? Cultural Foundation of
the Endorsement Process." Journal of Consumer Research 16, 3 (1989): 310-21.
McQuarrie, E. F. "Have Laboratory Experiments Become Detached from
Advertiser Goals? A Meta-Analysis." Journal of Advertising Research 38
(1998): 15-25.
Miller, Cyndee. "Celebs' Sweet Smell of Success Generates Dollars and
Scents," Marketing News, September 25, 1989.
Misra, Shekhar, and Sharon E. Beatty. "Celebrity Spokesperson and Brand
Congruence: An Assessment of Recall and Affect." Journal of Business
Research 21, September (1990): 159-73.
Mowen, John C., and Stephen W. Brown. "On Explaining and Predicting the
Effectiveness of Celebrity Endorsers." In Advances in Consumer Research Vol
8, K. B. Monroe, ed. Ann Arbor, MI: Association for Consumer Research, 1981.
Pasadeos, Yorgo, Joe Phelps, and Opdycke Lamme. "Newspaper Coverage of
Advertising: The Unfolding Story." In Proceedings of 2002 Conference of the
American Academy of Advertising, 2000.
Pavelchak, Mark A., John H. Antil, and James M. Munch. "The Super Bowl: An
Investigation into the Relationship among Program Context, Emotional
Experience, and Ad Recall." Journal of Consumer Research 15, 3 (1988):
360-367.
Pieters, R. M. and T. H. A. Bijmolt. "Consumer Memory for Television
Advertising: A Field Study of Duration, Serial Position, and Competition
Effects." Journal of Consumer Research, 23 (1997): 362-372.
Ray, Michael L. Advertising and Communication Management. Englewood Cliff:
Prentice-Hall, 1982.
Shanni, David, and Dennis Sandler. "Celebrity Alone Isn't a Sure Hit,"
Marketing News, August 8, 1991.
Sherman, Stratford P. "When You Wish Upon A Star." Fortune, August 19, 1985.
Shimp, Terrence. Advertising, Promotion: Supplemental Aspects of Integrated
Marketing Communications, 5th ed. Fort Worth, TX: The Dryden Press, 2000.
Swenson, Chester A. "Endorsements Are a Powerful Marketing Tool," Sales &
Marketing Management in Canada (MII) 28, 3 (1987): 8-10.
Till, Brian D., and Terence Shimp. "Endorsers in Advertising: The Case of
Negative Celebrity Information." Journal of Advertising 27, 1 (1998): 67-82.
Townsend, Bickley. "The Media Jungle." American Demographics, December, 1998.
Tripp, Carolyn, Thomas D Jensen, and Les Carlson. "The Effect of Multiple
Product Endorsements by Celebrities on Consumer Attitudes and Intentions."
Journal of Consumer Research 20, 4 (1994): 535-47.
Ward, Adrienne. "Toiletries & Beauty Aids: Passion Inflames Celebrity
Fragrance." Advertising Age, February 27, 1989.
Wells, William D. "Discovery-oriented Consumer Research." Journal of
Consumer Research 19, 4 (1993): 489-504.
Yates, Karen. "Making News: Press Coverage of Advertising Campaign."
Campaign, May 26, 1995
Youn, Seounmi, Tao Sun, William D. Wells, and Xinshu Zhao. "Commercial
Liking and Memory: Moderating Effects of Product Categories." Journal of
Advertising Research 41, 3 (2001): 7-13.
Zajonc, Robert B. "Additional Effects of Mere Exposure." Journal of
Personality and Social Psychology Monograph 9, 2 (1968): 1-28.
Zhao, Xinshu. "Clutter and Serial Order Redefined and Retested." Journal of
Advertising Research 37, 5 (1997): 57-73.
Zhao, Xinshu. "A Variable-Based Typology and a Review of
Advertising-Related Persuasion Research in 1990s." In The Persuasion
Handbook: Developments in Theory and Practice, James Price Dillard and
Michael Pfau, eds. pp. 495-512, Thousand Oaks, CA: Sage, 2002.
Ziegel, Vic. "Mark Spitz: Eleven Years and Seven Gold Medals Later."
Advertising Age, February 7, 1983.
Table 1
Descriptive Statistics for Dependent Variables
Variable
Min
Max
Mean
Median
S.D.
Skewness
Kurtosis
News Coverage
0.00
144.0
8.6
4.0
16.1
5.3
38.2
News Coverage
(log-transformed)
0.00
5.0
1.5
1.6
1.2
0.2
-0.7
Ad Liking
41.2
91.6
61.2
59.2
8.9
0.9
0.8
Recall (%)
0.00
56.7
5.9
0.8
11.8
2.6
6.1
Recall
(log-transformed)
0.00
4.1
1.0
0.6
1.2
1.1
0.1
Table 2
Comparisons of Means of Dependent Variables
Variables
Celebrity Ads
(N = 28)
Non-Celebrity Ads
(N = 108)
F Value
News Coverage
12.96 (27.84)
6.19 (10.22)
5.31*
News Coverage
(log-transformed)
2.29 (1.19)
1.32 (1.13)
7.87**
Ad Liking
68.23 (8.28)
59.40 (8.10)
20.55***
Recall (%)
17.62 (18.0)
2.84 (6.98)
34.82***
Recall
(log-transformed)
2.20 (1.42)
0.74 (0.89)
34.62***
1. *: p <0.05; **: p <0.01; ***: p <0.001
2. Standard deviations are in parentheses.
Table 3
Effects of Ad Liking and News Coverage on Brand Recall:
Step-Down Analysis
Sum of Squares
df
Mean Square
F
Sig.
?2
Covariates:
Ad frequency
20.51
1
20.51
44.64
.000
.25
Ad Liking
12.83
1
12.83
27.93
.000
.18
News
6.63
1
6.63
14.44
.000
.10
Independent Variable:
Celebrity
6.17
1
6.17
13.44
.000
.09
Error
60.19
131
.46
Total
335.89
136
R square = .680 (Adjusted R square = .671)
Notes:
1. The dependent variable, "brand recall" was used by the log-transformed
data.
2. A covariate, "news" was used by the log-transformed data.
3. Additional analysis with original data (without log-transformation)
showed similar
results. All variables except "news" were significant at
alpha level 0.001. "News" was
significant at 0.05.
Figure 1
Celebrity Endorsement Effects on News Coverage, Ad Liking, and
Brand Recall
|
