You are on page 1of 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328767249

The Role of Source Characteristics and Message Appeals in Public Service


Advertising (PSA) Messages: An Application of Fishbein’s Expectancy-Value
Model and the Match-Up Hypothesis...

Article  in  Journal of Current Issues and Research in Advertising · November 2018


DOI: 10.1080/10641734.2018.1503112

CITATION READS

1 465

3 authors, including:

Hye-Jin Paek
Hanyang University
132 PUBLICATIONS   2,965 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Hye-Jin Paek on 06 January 2019.

The user has requested enhancement of the downloaded file.


Journal of Current Issues & Research in Advertising

ISSN: 1064-1734 (Print) 2164-7313 (Online) Journal homepage: http://www.tandfonline.com/loi/ujci20

The Role of Source Characteristics and Message


Appeals in Public Service Advertising (PSA)
Messages: An Application of Fishbein’s Expectancy-
Value Model and the Match-Up Hypothesis for
Anti-Binge-Drinking Campaigns Targeting College
Students

Ho-Young (Anthony) Ahn, Hye-Jin Paek & Spencer Tinkham

To cite this article: Ho-Young (Anthony) Ahn, Hye-Jin Paek & Spencer Tinkham (2018): The Role
of Source Characteristics and Message Appeals in Public Service Advertising (PSA) Messages:
An Application of Fishbein’s Expectancy-Value Model and the Match-Up Hypothesis for Anti-
Binge-Drinking Campaigns Targeting College Students, Journal of Current Issues & Research in
Advertising, DOI: 10.1080/10641734.2018.1503112

To link to this article: https://doi.org/10.1080/10641734.2018.1503112

Published online: 05 Nov 2018.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ujci20
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING
https://doi.org/10.1080/10641734.2018.1503112

The Role of Source Characteristics and Message Appeals in


Public Service Advertising (PSA) Messages: An Application
of Fishbein’s Expectancy-Value Model and the Match-Up
Hypothesis for Anti-Binge-Drinking Campaigns Targeting
College Students
Ho-Young (Anthony) Ahna, Hye-Jin Paekb, and Spencer Tinkhamc
a
Communications Division, Pepperdine University, Malibu, California, USA; bAdvertising & Public Relations,
Hanyang University, Ansan, Republic of Korea; cAdvertising and Public Relations, University of Georgia,
Athens, Georgia, USA

ABSTRACT
Guided by a match-up hypothesis and Fishbein’s expectancy-value theory,
this research examines the effects of message sources and appeals in anti-
binge-drinking public service ads (PSAs) on college students’ binge-
drinking attitudes and behavioral intention. College students (N ¼ 251)
participated in a 2 sources (expertise vs. similarity)  2 appeals (expectancy
vs. valuative) factorial experiment. Results show that there were conditional
impacts of a similar source (peer) on favorable attitudes toward the PSAs.
Messages were more effective when there was consistency between source
characteristics and appeals (i.e., expert/expectancy and peer/valuative).
The usefulness of Fishbein’s expectancy-value theory and the alternative
framework for PSA developers are discussed.

Binge-drinking behavior among college students is one of the most significant and challenging
issues on campuses (Pedersen and LaBrie 2007; Hingson et al. 2009; National Institute on
Alcohol Abuse and Alcoholism 2012; Wolburg 2016). According to the 2014 National Survey on
Drug Use and Health, 60% of college students aged 18–22 consumed alcohol in the previous
month. On average, two out of three of them had experience with binge drinking (SAMHSA
[Substance Abuse and Mental Health Services Administration] 2014). The National Institute
on Drug Abuse reported that 36% of college students have engaged in binge drinking. Binge
drinking is defined as “a pattern of drinking that brings a person’s blood alcohol concentration
to 0.08 grams percent or above” (Centers for Disease Control and Prevention 2012). Binge drink-
ing typically occurs “when a man consumes five or more alcoholic drinks in a row, and a woman
consumes four or more alcoholic drinks in a row, within about 2 hours.”
While binge drinking is considered an integral part of life as a college student (Boekeloo,
Novik, and Bush 2011; Wolburg 2016), it is also recognized as a nationwide problem, causing
concerns for student health (i.e., compulsiveness, depression, anxiety, etc.), safety issues (i.e., driv-
ing under the influence, injuring themselves, harming property, sexual assault, death, etc.), aca-
demic problems (i.e., missing classes, lower grades, etc.), or related social matters (i.e., losing
friends, violent behavior, etc.) (Johnsson, Leifman, and Berglund 2008; Hingson, et al. 2009;
NIAAA 2012; SAMHSA 2014).

CONTACT Ho-Young (Anthony) Ahn Hoyoung.Ahn@pepperdine.edu Communication Division, Pepperdine University,


24255 Pacific Coast Hwy, Malibu, CA 90263, USA.
This article was accepted under the previous editor, Eric Haley.
ß 2018 American Academy of Advertising
2 H.-Y. (A.) AHN ET AL.

With a clear issue in binge drinking and its related problems, health communication practi-
tioners and scholars have pursued efforts to reverse this trend among college students through a
variety of mass media campaigns (Dejong 2002). Among these media health campaigns, public
service announcements or public service ads (PSAs) are often employed in health campaigns
(DeJong and Atkin 1995), seeking to reduce binge-drinking rates.
The goal of a typical PSA is to influence people’s beliefs about a topic and their decisions to
engage in a behavior. By affecting people’s beliefs, feelings, and intentions, PSAs can influence a
target’s attitude toward engaging in particular behaviors. Some studies focusing particularly on
alcohol and other drugs, however, showed an effect of PSA on increasing knowledge and aware-
ness, but the PSA appeared to have an unsubstantial impact on actual behavior (i.e., Wechsler
and Isaac 1992; Stephenson 2003; Hill 2004). This apparent inconsistency and mixed results have
sparked much debate about how useful PSAs are at influencing attitude and behavior (Lynn
1974; Treise, Wolburg, and Otnes 1999; Snyder 2001; Wolburg 2001).
It is a challenging yet important task for PSA practitioners to determine how to enhance the
effectiveness of health messages. Varying types of messengers and messages have been studied as
potentially important determinants of effectiveness (Atkin 1994). Along with knowing who the
audience is, “who says” (i.e., message sources), and “how tells” (i.e., message appeals) in the
health messages are the key components of creating a persuasive health message in an effort to
make the messages effective. For example, particular types of messengers (expert, peer, celebrity,
etc.) have been adopted into a variety of health campaigns. However, none has been necessarily
more effective than the others in all conditions (Atkin 2001). Additionally, researchers and practi-
tioners have argued that the intended influence on the audience strongly depends on using
appropriate message appeals (i.e., Fishbein and Ajzen 1981; Atkin and Freimuth 1989; Zhang and
Gelb 1996; Lepkowska-White, Brashear and Weinberger 2003; Hoeken et al. 2007; Zhang 2010).
That is, what the messengers say and how they address it is of great importance. For that reason,
the roles of both message source and message appeal should be examined in terms of the goal of
the messages, typically attitudinal and behavioral change within the target audience.
While many studies have examined the relative effectiveness of the messenger sources (i.e.,
Haas 1981; Perse, Nathanson, and McLeod 1996; Andsager et al. 2006) or message appeals (i.e.,
Fishbein and Ajzen 1981; Dillard 1994; Stephenson and Witte 1998; Smith and Stutts 2003;
Zhang 2010) on audiences’ behavior and attitudinal change, there is a relative paucity of research
that examines the relative effectiveness of PSA messenger sources, message appeals, and the inter-
actions between source types and message appeals in terms of the persuasiveness of PSAs. In
terms of research on binge drinking among college students, much is known about the types of
factors affecting heavy drinking (i.e., individual characteristics such as drinking motivation, per-
sonality, race and ethnicity, or Greek membership), and little is known of how to address college
drinking and provide appropriate solutions in order to reduce it (Wolburg 2016), especially
through PSAs.
The purpose of the present study is to explore the effectiveness of the messenger sources and
message appeals, as components of persuasive communication in a context of a salient health
issue: binge drinking among college students. This study employs the match-up hypothesis
(Kamins and Gupta 1994) and the Fishbein expectancy-value theory (Fishbein and Ajzen 1975)
to examine the relative effects of two message sources (expert: expertise vs. peer: similarity), two
message appeals (expectancy appeal vs. valuative appeal), and the interactions. Based on the
insights of these theories, we generally expect that the relative effectiveness of the PSA will
depend upon the message source characteristics being matched with the message appeals. For
example, we predict that expectancy appeals focusing on the audience’s subjective beliefs regard-
ing the consequences of binge drinking will be more appropriate for delivery by an expert source
than a peer source. In addition, we expect that the valuative appeal, focusing on the shared, felt
experience of binge drinking, will be better matched with a similar source, a peer.
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 3

The novelties and major findings of this study provide both practical and theoretical implica-
tions for PSA practitioners and communication researchers in terms of tailoring and targeting
messages and more nuanced evaluation for interventions designed to reduce risk drinking.
First, considering the understudied nature of relative effectiveness of message source character-
istics on persuasion, examining the role of message sources (expert vs. peer) specifically targeting
to college students is important in terms of the development of target-tailored (i.e., peer-based
PSA) health interventions. Beyond the conventional fear-arousing appeals, unconventional appeals
(expectancy vs. valuative appeals) are created based on the concepts of the Fishbein’s expectancy-
value theory. The crux of the appeal is to capture college students’ subjective probability of a con-
sequence occurring and their evaluation of the outcome resulting from binge drinking. Thus, we
expect that the findings of this study may encourage PSA practitioners to develop intervention
messages tailored to the predispositions and abilities of college students.
Second, while many studies have dealt with matching effects in the areas of marketing commu-
nication, our study examines the relatively less studied interaction effects with persuasion varia-
bles such as message source characteristics and message appeals in the context of PSA. In
company with knowing “who says” and “how tells” in persuasion, understanding “who says how”
and ferreting out the relative effectiveness of possible matches between message source and
appeals can contribute to PSA practitioners getting one more step closer to creating persuasive
messages and achieving the ultimate attitudinal and behavior change of the targets.
Finally, this study is unique in that it scrutinizes the effectiveness of PSA in terms of two
different dimensions of attitudinal change. Specifically, this study attempts to use the unconven-
tional, diagnostic dimensions of Fishbein’s expectancy-value attitude model in a public health
area along with the global attitude model. We expect that this attitude model will be beneficial in
understanding health-related attitude research more precisely.

Message source effects: Expertise versus similarity


On occasion, whether an audience accepts a message positively or negatively rests on the charac-
teristic of the spokesperson. A large body of literature indicates that attributes of message sources
endorsing persuasive messages play an important role in terms of message evaluation (Sternthal,
Phillips, and Dholakia 1978; Friedman and Friedman 1979; Haas 1981; Perse, Nathanson, and
McLeod 1996). Among multiple dimensions of source credibility, previous research has identified
two important message source characteristics: source expertise and source similarity. Researchers
have suggested that both characteristics have resulted in enhanced persuasive effects (McCroskey
1966; Ohanian 1991; Andsager et al. 2006). In general, high expertise and highly similar sources
are more persuasive than less expertise and less similar sources (i.e., Cialdini 2001; Andsager
et al. 2002; Perloff 2008).
Source expertise refers to the extent to which a speaker is perceived to be capable of making
correct assertions (Hovland, Janis, and Kelley 1953, p. 22), relevant thoughts, and skills (Perloff
2008). Many studies have found that people are more prone to accept an expert’s message than a
nonexpert source’s (Herr, Kardes, and Kim 1991; Feick and Higie 1992; Bone 1995). According
to Salmon and Atkin (2003), experts in health-related areas heighten response efficacy, defined as
the audience’s belief as to whether the recommendation presented in the message is effective in
preventing or eliminating the threat (Tay et al. 2001). Thus, people tend to follow experts’ recom-
mendations because they are viewed to be of a higher level of qualification on a particular subject
(Gilly et al. 1998).
Source similarity refers to the extent to which individuals consider themselves similar to the
speaker, based on certain characteristics (Brown and Reingen 1987), including shared values or
ideas (Salmon and Atkin 2003). Many theories such as the source-attractiveness model (Kelman
1961), reference group theory (Sherif and Sherif 1964), and social comparison theory (Festinger
4 H.-Y. (A.) AHN ET AL.

1954) have explained why perceived source similarity might raise the power of the information
conveyed. Also, several studies have indicated how similarity can lead to persuasion (Anderson
and McMillion 1995; Hass 198; Myrick 1998; Perloff 2008; Silvia 2005). For example, Salmon and
Atkin (2003) revealed that even if famous people, leaders in religion, and leaders in politics are
influential, in general, peer models are viewed as more persuasive than others in terms of younger
viewers.
One question of interest in this study is which message source characteristics—expertise or
similarity—will be more effective in the context of college students’ binge-drinking behaviors.
Either type of message sources can be more persuasive than the other. For example, college stu-
dents may perceive the expert to be more knowledgeable about the negative outcomes of binge
drinking. Then, it is expected that the PSA will be more persuasive than a PSA that uses a peer
messenger who may be perceived as having little or no formal expertise on the negative outcomes
of binge drinking.
On the other hand, the peer as a message source shares the value of similarity with college stu-
dents. Atkin (1994) argued that providing messengers who share similarities with the audience
tends to increase the likelihood of the message claims being accepted. Then, a peer message
source may be viewed as an experienced person who actually knows the negative feelings that
may result from binge drinking.
Although expert (expertise) or peer (similarity) is an expected contributor to persuasion effect-
iveness, the relative effects in different situations appear inconsistently, and no one source (expert,
peer, celebrity, etc.) has been necessarily greater than the others in terms of message source
effects (Atkin 2001, 1994). Also, few studies have dealt with the direct comparison between source
expertise and source similarity on the relative effectiveness on persuasion. Taking into account
the relative effectiveness of expertise and similarity, we address the following research question:
RQ1: Which source characteristic—expertise or similarity—is more effective on the (a) attitudes toward the
anti-binge-drinking PSA, (b) attitudes toward binge drinking, and (c) binge-drinking intentions?

Expectancy versus valuative message appeals: Applying expectancy-value theory


Health campaign messages often connect undesired behaviors with negative outcomes (Atkin
2001). While an appeal refers to “the basic motivational or persuasive technique used in
advertising” (Leiss, Kline, and Jhally 1986, 220), reaction to threatening messages in health cam-
paigns depends upon the message’s general tone or appeal (Pallak, Murroni, and Koch 1983;
Petty, Cacioppo, and Schumann 1983). Instead of merely urging individuals to act a certain way
by using conventional fear-arousing appeals with mixed results (Dillard 1994; Stephenson and
Witte 1998), Fishbein and Ajzen (1975) and Salmon and Atkin (2001) have provided measures of
the components of the attitude structure and suggested that the different components of the atti-
tude structure can be the basis for developing different types of persuasive appeals. Specifically,
Atkin (2001) suggested that message appeals in a health campaign can stress either of the two
basic components in Fishbein’s expectancy value formulation: the subjective probability of a con-
sequence occurring, or the degree of positive or negative valence of that outcome.
Fishbein’s expectancy-value theory was initially intended to predict attitudes toward objects or
actions. This theory posits that attitudes are built up and adjusted according to judgments regard-
ing beliefs and values (Fishbein and Raven 1962; Ajzen and Fishbein 2008). Fishbein and Ajzen
(1975) argued that persons first react to new information about the objects and actions by con-
structing a belief about the objects and actions. Given that a belief is previously present, it is
likely to be altered by novel information. Next, individuals allocate a value to each aspect where
belief is located. Third, the psychological computation of beliefs and values leads to a creation of
an expectation of attitudes or attitude development (Fishbein and Ajzen 1975). Essentially, one’s
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 5

attitude is formed by a factorial function of beliefs (b) and evaluated values (e): all beliefs (b)
about the objects or actions, weighted by the “evaluative aspects” (e) of those beliefs (Areni and
Lutz 1988). Fishbein and Ajzen (1975) characterized the theory with the following equation:
Pn
i¼1 AB ¼ bi  ei

where AB is attitude toward the behavior, bi is likelihood or probability of consequence i, and ei


is evaluation of consequence i.
As pointed out earlier, either of the two attitude determinants can be stressed as message
appeals: (1) an appeal that focuses on the subjective probability of a consequence occurring and (2)
an appeal that focuses on the degree of positive or negative evaluation of that outcome. In this
perspective, attitude formation and change may be viewed as dependent on a combination of the
judged likelihood that a claimed outcome is true (expectancy response) and/or a valuative assess-
ment of the valence of that outcome (valuative response). To borrow O’Keefe’s arguments, attitude
might be influenced either by cognitive (belief-related) considerations or by affective (feeling-
related) considerations (O’Keefe 1990). O’Keefe points out that to understand attitude fully, non-
cognitive elements (such as feelings) should also be recognized as a basis of attitude formation.
Although suggestions for developing the different types of persuasive appeals using these two
components (Atkin 2001) were made, few studies have investigated the relative impact of the
components of the attitude structures when incorporated into persuasive appeals. For that reason
in this study, two distinct types of appeals, expectancy appeals (assertions about outcome likeli-
hood) that focus on the subjective probability of a consequence occurring (i.e., how likely is it
that binge drinking can cause blackouts) and valuative appeals (stressing the affective valence of
the outcomes) that focus on the degree of positive or negative evaluation of that outcome (i.e.,
how bad or good it is to experience blackouts) were employed.
Because there are insufficient theories and studies that directly explain how expectancy and
valuative appeals work for persuasion (attitudes and behavioral intentions), a second research
question was developed.
RQ2: Which message appeal—expectancy or valuative appeal—is more effective on the (a) attitudes toward
the anti-binge-drinking PSA, (b) attitudes toward binge drinking, and (c) binge-drinking intentions?

Match-up hypothesis: Matching the message source with the message appeal
The assumption that persuasion will be enhanced when the audience perceives that there is a fit
between the messenger and the messages has been termed the match-up hypothesis (Kamins 1990;
Lynch and Schuler 1994; Kamins and Gupta 1994; Perse, Nathanson, and McLeod 1996; Moore
2004). This hypothesis is also explained by several theoretical postulates. Among them, social
adaptation theory explains that when a persuasive message provides an adaptive informational
cue that matches with people’s expectations, receivers are more likely to accept the message, and
thus persuasion would be more likely to occur (Kahle and Homer 1985) because “individuals’
beliefs about the adaptive utility (the message’s potential to offer socially useful information)
determines the level of impact that messages will have” (Perse et al. 1996, 173). Many advertising
effectiveness studies associated with the influence of a famous spokesperson propose that the
spokesperson’s attributes (i.e., expertise) can boost effective communication as long as the famous
endorser has a congruency with the product or brand (i.e., Moore 2004). This aspect of the
match-up hypothesis formalizes this idea of a celebrity fit with the product or brand (Kahle and
Homer 1985; Kamins 1990; Solomon et al. 1992; Lynch and Schuler 1994).
Continued research has indicated evidence for the use and effectiveness of a match-up hypoth-
esis, relating it to source credibility and attitude formation (Kamins 1990; Kamins and Gupta
1994). For example, in an advertisement, Kamins (1990) pointed out that a good-looking celebrity
6 H.-Y. (A.) AHN ET AL.

(i.e., Tom Selleck) was more effective on the credibility of the endorser and the advertising
evaluation than a physically unappealing model when endorsing a sports automobile or other
attentiveness-related goods. But for advertising unconnected to attractiveness, such as one for a
computer, the spokesperson (i.e., Telly Savalas) did not yield a high evaluation on the spokesper-
son credibility and attitude toward the advertising. From the study, Kamins (1990) found that the
fit between spokesperson and characteristics of the product clarified the visual cue of the person
determining the values of the product. On the whole, augmenting the impact of advertising or
public service announcements relies on good matches between the product representative and the
item for consumption (Perse, Nathanson, and McLeod 1996).
In applying the match-up hypothesis to the context of drinking, anti-binge-drinking messages
are more likely to be persuasive if source characteristics match message appeals. Salmon and
Atkin (2003) argued that belief (expectancy) appeals that emphasize the likelihood of consequen-
ces tend to make the messages more believable than evaluative (valuative) appeals for the audi-
ence because of the messenger’s perceived expertise. Thus, it follows that the belief-oriented
message (expectancy appeals) may be congruent with a messenger who has perceived expertise.
By contrast, the valuative appeals focusing on the perceived feeling and value of negative out-
comes resulting from binge drinking may be a better fit with source similarity because message
recipients may share similar feelings and values with the peer as a source of information about
negative consequences. Based on the empirical studies and theoretical arguments, the following
hypotheses were proposed:
H1: A PSA featuring an expert messenger in expectancy appeals (matched condition) will be more effective
than a PSA featuring an expert messenger in valuative appeals (mismatched condition) (H1a) as well as
a PSA featuring a peer in expectancy appeals (mismatched condition) (H1b).

H2: A PSA featuring a peer messenger in valuative appeals (matched condition) will be more effective than
a PSA featuring a peer messenger in expectancy appeals (mismatched condition) (H2a) as well as a
PSA featuring an expert in valuative appeals (mismatched condition) (H2b).

Unlike the predictions that matched combinations between sources and appeals will be super-
ior to the mismatched cases in terms of persuasion effects, there is little theory or conclusive evi-
dence regarding the relative effectiveness of the two congruent cases, expectancy/expert versus
valuative/peer, or the two incongruent conditions, expectancy/peer versus valuative/expert. In
other words, it is difficult to hypothesize which source–message combination will be more effect-
ive when both are congruent or both are incongruent. Thus, the following research questions
were generated.
RQ3: When the source and message appeal are congruent (expert/expectancy vs. peer/valuative), which
source-message appeal matching is more effective on the (a) attitudes toward the anti-binge-drinking
PSA, (b) attitudes toward binge drinking, and (c) binge-drinking intentions?

RQ4: When the source and message appeal are incongruent (expert/valuative vs. peer/expectancy), which
source-message appeal matching is more effective on (a) attitudes toward the anti-binge-drinking
PSA, (b) attitudes toward binge drinking, and (c) binge-drinking intentions?

Method
Design
The experiment was conducted using a 2  2 between-subjects factorial design: 2 (source charac-
teristics: an expert vs. a peer)  2 (message appeals: expectancy vs. valuative). It was a posttest-
only randomized design.
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 7

Sample
Participants were recruited via e-mails sent to 2000 randomly selected (by the office of the regis-
trar) undergraduate students in a large Southeastern university. Once they agreed to take part,
they were asked to go to a website where they could participate in the study. The participants
were randomly assigned to one of the four cells to test the research questions and hypotheses
(N ¼ 251; females ¼ 68%, non-Hispanic White ¼ 77%, averaged age ¼ 21 years, academic year ¼
freshman 25%, second year 17%, third year and fourth year 21%). Participants were compensated
through a random drawing for cash prizes (total $450) as incentives. Institutional review board
(IRB) approval was obtained for this study.

Stimulus materials
Four different transcripts for an anti-binge-drinking PSA were created. Each transcript contains
different source and appeal combinations (expert/expectancy, expert/valuative, peer/expectancy, or
peer/valuative). A script, rather than a broadcast stimulus, was used to avoid the possible confound-
ing effects such as a spokesperson’s voice and gender (Ray, Ray, and Zahn 1991). To manipulate
message sources (expertise vs. similarity), an expert (a medical doctor) and a peer were featured in
the PSAs. Also, Pat Michaels was used as both messengers’ name to avoid gender specificity. In this
study, expectancy appeals mainly focused on the likelihood of undesirable outcomes of binge drink-
ing (i.e., you are eight times more likely than non-binge drinkers to miss class). The valuative
appeals created were predominantly focused on the degree of positive or negative evaluation of that
outcome (i.e., how bad it is to have to throw up). According to the National Institute on Alcohol
Abuse and Alcoholism book (NIAAA 2012), message content included the college students’ percep-
tion of the negative outcomes of binge drinking: short-term/long term physical health threats and
social threats (for the stimuli, see Appendix).

Manipulation checks
Prior to the study, a pretest was conducted among 136 undergraduate students from the same
university, who did not participate in the main study. The purpose of the pretest was to deter-
mine whether source characteristics and message manipulations were valid.
To measure message source manipulation, participants were asked to rate the similarity and
expertise of message sources using a 7-point Likert scale ranging from “strongly disagree” (1) to
“strongly agree” (7). Adapted from the study by Price, Feick, and Higie (1989), the question items
included: I think that the spokesperson in the PSA is “similar to me,” “similar to me in his/her
behavior” (2 items for similarity), “expert,” “experienced,” “knowledgeable,” “qualified,” and “skilled”
(5 items for expertise). All of these items were summed and then averaged to construct each vari-
able. Cronbach’s alpha reliability indicated acceptable internal consistency (.74 and .72, respectively).
To measure the message appeal manipulation, the participants were given definitions of
expectancy message appeal and valuative message appeal and were asked to match the stimulus
with the definitions of the message appeals (i.e., think that the type of the message appeals I just
read applies to). Further, after exposure to the randomly assigned treatments, participants were
asked to directly assess both the message and the source. These Likert items form a basis for con-
sidering whether or not the treatments were perceived by participants to be consistent with their
theoretical meaning.

Procedures
Using an online survey module, participants were first asked about their personal background,
such as Greek organization (fraternity, sorority) membership and past drinking patterns, as well
8 H.-Y. (A.) AHN ET AL.

as their knowledge, experiences, and perceptions of the PSAs in general. After reading the tran-
scribed PSA, each participant was asked to complete a questionnaire indexing the effectiveness of
the PSA and their binge-drinking intentions. The definition of binge drinking was presented in
the manipulated PSA messages as follows: For men, 5 or more drinks in two hours and for
women, 4 or more drinks in two hours is defined as binge drinking. In a coding procedure, the
subsample that had no binge-drinking experience was purposively screened out of the sample
(N ¼ 21), so that every participant has at least a minimal level of binge-drinking behavior in the
past month, which represents a sample of “experienced” drinkers.

Measure-dependent variables
The model included three dependent variables: (1) attitude toward the anti-binge-drinking PSA;
(2) attitude toward binge drinking; and (3) binge-drinking intention. In addition, attitude toward
binge drinking was assessed in two different ways: (a) global attitude toward binge drinking and
(b) expectancy-value attitude toward binge drinking. Except for the expectancy-value attitude
toward binge drinking, multiple question items for each variable were averaged to construct each
dependent variable.

Attitude toward the PSA


Attitude toward the PSA was measured with the following five items that were adapted from
existing literature and modified to fit the study context (7-point Likert scale items anchored by
“strongly disagree” [1] and “strongly agree” [7]): “likable,” “interesting,” “good,” “favorable,” and
“pleasant” (alpha ¼ .87) (MacKenzie and Lutz 1989; Madden, Allen, and Twible 1988).

Global attitude toward binge drinking


Global attitude toward anti-binge drinking was measured by asking participants “After reading
the anti-binge-drinking PSA, I think that BINGE DRINKING is” with four responses using
7-point semantic differential scales (harmful/beneficial, unpleasant/pleasant, bad/good, and worth-
less/valuable; alpha ¼ .91 (Madden et al. 1988; MacKenzie and Lutz 1989).

Expectancy-value attitude toward binge drinking


This variable was assessed with the expectancy-value model formula. This measure was built as
the summation of each subject’s attribute specific belief (expectancy) strength score for the likeli-
hood of each of the bad outcomes of binge drinking (b) multiplied by the respective evaluation
(valuative) of the outcome (e) (cf. Smith and Swinyard 1983). Each negative message component
from the message content was included in the calculation. The form of the belief (expectancy)
items was “After reading the anti-binge drinking PSA, I feel that if I participate in binge drinking,
it will cause the symptoms of vomiting (b1) and blackouts (b2), a pounding hangover (b3), liver
cancer (b4), missing class (b5), damaging my academic career (b6), losing family support (b7), and
losing my friends (b8).” The endpoints were labeled “extremely unlikely” (3) and “extremely
likely” (þ3).
The form of the evaluative (valuative) items was “In general, I feel that vomiting (e1) is, black-
outs (e2) are, a pounding hangover (e3) is, liver cancer (e4) is, missing classes (e5) is, damaging
my academic career (e6) is, losing family support (e7) is, and losing my friends (e8) is.” The end-
points were labeled “extremely bad” (3) and “extremely good” (þ3). Then the expectancy-value
attitude toward binge drinking variable was computed by multiplying each e (valuative) compo-
nent by the b (expectancy) component (bi  ei) and summing up (b1  e1 þ b2  e2 þ … þ b8 
e8). Because each expectancy-value component was coded respectively on a scale from 3 to þ3,
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 9

the expected interpretation was that anything with a negative value represents a negative attitude
toward binge drinking, which is the desirable response (in this case, the mean summed attitude
value was calculated, yielding a theoretical range of 9 to 9). The summed bi  ei measure is con-
sistent with the measure of attitude proposed by Fishbein and Ajzen (1975).
To examine possible redundancy of these two measures, the zero-order correlation between
global attitude and Fishbein’s attitude was examined. The result revealed a positive correlation
with only a moderate effect size (r ¼ .56, p < .01). Thus, the two measures were found not to be
redundant and were treated as separate dependent variables in the multivariate analysis.

Binge-drinking intention
The intention was measured with the following three items using a 7-point Likert scale (Sheth
1974): “After reading the anti-binge drinking PSA, I think that I intend to binge drink in the
forthcoming month” (1 ¼ “extremely unlikely” and 7 ¼ “extremely likely”); “I will try not to binge
drink in the forthcoming month” (1 ¼ “definitely true” and 7 ¼ “definitely false”); and “I plan to
binge drink in the forthcoming month” (1 ¼ “strongly disagree” and 7 ¼ “strongly agree;”
alpha ¼ .95). Descriptive statistics for the dependent measures are reported in Table 1.

Covariates
Subjects’ alcohol use in the past, their recent alcohol use, past experience of negative outcomes
from binge drinking, and message quality served as covariates because they could confound the
effects between message sources and message appeals on persuasion (Hair et al. 1998). The ques-
tion related to subject’s past alcohol use was “Think back over the past few months, how many
times did you have five or more drinks at a sitting?” and to their recent drinking patterns was
“In the last two weeks, how many times have you had five or more drinks in a row?” (Wechsler
et al. 1994). Also, subjects’ past experience of negative outcomes due to binge drinking was
assessed with “Did any of following occur as a consequence of your drinking?”(vomiting, black-
outs, a pounding hangover, liver cancer, missing classes, lower grades, losing family support, and
losing close friends). These eight consequences were the same as the message content used in
PSA message stimuli. The yes/no responses were coded as 1 and 0 and summed in to a negative
past experience index potentially ranging from 0 to 8. Finally, a question associated with PSA

Table 1. Mean and standard deviation values for the dependent measures within each experimental condition and across
each independent variable.
Experimental conditions
Source Message Interactions
Expert/ Peer/ Expert/ Peer/
Expert Peer Expectancy Valuative expectancy expectancy valuative valuative
Attitude toward PSAs M 3.91 4.30 4.11 4.10 3.99 4.24 3.83 4.37
SD 1.16 1.17 1.11 1.22 1.16 1.06 1.17 1.29
Global attitude toward M 2.30 2.21 2.16 2.35 2.06 2.25 2.55 2.16
binge drinking
SD 1.17 1.06 1.13 1.09 1.09 1.17 1.21 .94
Expectancy value attitude M 1.22 1.74 1.85 1.11 2.20 1.50 .25 1.98
toward binge drinking
SD 4.18 4.32 4.32 4.13 4.08 4.53 4.13 4.10
Intention to binge drink M 2.64 2.53 2.49 2.67 2.37 2.61 2.90 2.45
SD 2.04 2.03 2.08 1.98 2.03 2.14 2.04 1.93
Notes. Mean value for the PSA score and global attitude: 1 ¼ strongly disagree, 7 ¼ strongly agree. Mean value for the expect-
ancy-value attitude: expectancy appeal, 1 ¼ extremely unlikely, 7 ¼ extremely likely and valuative appeal, 1 ¼ extremely bad,
7 ¼ extremely good. Recode: 1 to 3 and 7 to 3. Mean value for the intention to binge drink: 1 ¼ extremely unlikely, defin-
itely false, and strongly disagree, 7 ¼ extremely likely, definitely true, and strongly agree.
10 H.-Y. (A.) AHN ET AL.

message quality was “After reading the anti-binge drinking PSA, how much do you agree the
message was high quality?” This covariate was measured using a 7-point scale ranging from
“strongly disagree” (1) to “strongly agree” (7).

Results
Manipulation checks prior to experiment
Two independent-samples t-tests revealed statistically significant differences between treatment
groups for both factors of source similarity (mean difference ¼ .54, t(249) ¼ 3.309, p < .001) and
source expertise (mean difference ¼ 1.68, t(235) ¼ 10.739, p < .001). That is, the peer condition was
evaluated to be significantly more similar to the participants themselves than the expert condition,
while the expert condition was evaluated to be more knowledgeable than the peer condition.
To measure the message appeal manipulation, the participants were given definitions of
expectancy message appeal and valuative message appeal and were asked to match the stimulus
with the definitions of the message appeals (i.e., think that the type of the message appeals I
just read applies to). Half of the participants assessed the expectancy appeals while the other half
evaluated the valuative appeals. In regard to expectancy and valuative appeal manipulation checks,
a cross-tab analysis revealed a statistically significant difference in matching between message
stimuli and appeal definitions (v2 ¼ 1.22, df ¼ 3, p < .001), meaning that most participants were
able to match the expectancy appeal with the definition of expectancy appeal (89.7%) and the
valuative appeal with the definition of valuative appeal (88.2%). Overall, manipulation checks
prior to implementation of the experiment were successfully supported.

Manipulation checks within the study


To see the consistency between how participants perceived the treatments and their theoretical meaning,
we asked participants to directly evaluate both the message and the source. Thirty-two message-related
items were employed, ranging from those descriptive of message content as well as perceived impact, to
perceptual items assessing positive and negative emotional impact as well as perceived credibility and
effectiveness. With respect to source assessment, 14 Likert items dealing with different aspects of per-
ceived credibility and similarity were included. Factor analyses of these two sets of items produced a four-
factor oblique solution for the message items and a two-factor oblique solution for source items. Message
factors were judged to represent usefulness/effectiveness, familiarity, positive emotions, and negative emo-
tions. A second-order solution produced two broader factors, judged to represent utilitarian attitude
toward the PSA, merging the usefulness and familiarity components, and hedonic attitude toward the
PSA, combining both emotional components. Standardized factor scores for each second-order factor
were saved to the data. The two-factor solution for source ratings clearly separated into an expertise com-
ponent and a similarity component. These also were saved to the data as standardized factor scores.
In order to test perceived source characteristics, treatments were entered in a multivariate ana-
lysis of variance (MANOVA) predicting the expertise and similarity factor scores. Results
show only main effects of the source treatment, in the expected direction. That is, the expert
source scored higher than the peer source on the perceived expertise factor (F ¼ 66.838, df ¼ 1,
p ¼ .000), while the peer source exhibited higher scores on the perceived similarity factor than
did the expert source (F ¼ 4.136, df ¼ 1, p ¼ .043).
The internal manipulation check for the message treatment exhibited a more complex pattern.
Research cited by Fishbein and Ajzen (1975, 496) found that appeals perceived to be “rational” or
logical contain belief statements that the subject can accept as true (similar to “expectancy”
appeals), whereas messages perceived to be “emotional” contain belief statements with attribute
evaluations of high polarity (similar to “valuative” appeals). Thus, it would be expected that the
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 11

expectancy message treatment would produce higher utilitarian attitudes toward the PSA than the
valuative treatment, and the valuative message treatment would produce more polarized hedonic
attitudes toward the PSA (i.e., higher in absolute value) than the expectancy appeal. Results of
MANOVA in which treatments predicted the utilitarian PSA attitude and the absolute value of
the hedonic PSA attitude support the first expectation that the expectancy appeal would produces
a more utilitarian attitude than the valuative appeal (F ¼ 5.306, df ¼ 1, p ¼ .022), but not the
second expectation (although within the sample the polarity of the hedonic attitude was indeed
higher for the valuative appeal than for the expectancy appeal). With respect to attitudinal polar-
ity, however, a significant interaction effect was observed, characterized by the “matched” condi-
tions (expert–expectancy appeal and peer–valuative appeal) exhibiting higher hedonic attitudinal
polarities than did the “mismatched” conditions (F ¼ 4.091, df ¼ 1, p ¼ .044). Thus, hedonic polar-
ity seems to be inextricably linked to both the appeal and the source. This observed disordinal
interaction confounds the possibility of observing within the experiment a significant hedonic
polarity effect of the valuative message treatment in isolation.

Research questions/hypothesis testing


A multivariate analysis of covariance (MANCOVA) was performed to determine the effect of the
type of message sources (expert vs. peer) and message appeals (expectancy vs. valuative) on (a)
attitudes toward the anti-binge-drinking PSA, (b) attitudes toward binge drinking, and (c) binge-
drinking intentions. MANCOVA seemed the most appropriate statistical test because there are
four multiple dependent variables, along with two independent variables and four covariates.
Multivariate and univariate effects are reported in Table 2.
Prior to investigating treatment effects with MANCOVA, various assumptions were tested.
Foremost, the assumption of normality was successfully verified by skewness and kurtosis statis-
tics in terms of the dependent variables and covariates. Second, the homogeneity of variance was
satisfied among the covariance matrices by a nonsignificant Box’s M test, F(30, 16376) ¼ .642,
p ¼ .935. Third, in a covariate analysis, covariates must be correlated with the dependent variables
(Hair et al. 1998). A correlation matrix suggested that all of the covariates were correlated with a
least one of the dependent variables. Fourth, before using MANCOVA it is important to identify

Table 2. MANCOVA: Effects of sources and appeals on dependent variables.


Multivariate
effects Univariate effects
F(1, 243)
Global attitude Expectancy-value
Attitude toward attitude toward Intention to
Wilks’s k F(4, 240) toward PSA binge drinking binge drinking binge drink
Covariate
Drinking pattern 1 (past) .94 3.73 2.16 11.58 1.42 11.33
Drinking pattern 2 (recent) .89 7.38 .00 1.22 3.19 24.34
Consequences of drinking .96 2.02 .90 1.64 6.02 4.46
PSA quality .49 63.78 255.38 18.37 13.55 12.78
Fixed factors
Sources (expert vs. peer) .95 3.49 13.59 .55 1.09 .33
Appeals (cognitive .99 .89 .01 2.64 2.29 .99
vs. affective)
Sources  appeals .96 2.28 1.93 5.61 6.39 3.50
Note. Reported significance levels are based on two-tailed tests with the exception of the significant interactions, which are
interpreted as one-tailed tests of directional hypotheses (H1 and H2).
p < 0.05.
p < 0.01.
p < 0.001.
12 H.-Y. (A.) AHN ET AL.

any outliers that affect the level of type I error and distort the results (Hair et al. 1998). An inves-
tigation of standardized residuals across the dependent variables indicated four cases as outliers.
Thus, the sample size was reduced to 251 observations.

RQ1 main effects: Expert versus peer


The first research question explored the effectiveness of message sources (expert vs. peer) on the
dependent variables. Multivariate tests indicated a significant difference among message sources
on the dependent measures (Wilks’s k ¼ .95, F(4, 240) ¼ 3.49, p < .05, partial g2 ¼ .06). Follow-up
analyses of variance (ANOVA) tests indicated that the source was significantly related to attitude
toward the PSA (F(1, 243) ¼ 13.59, p < .01, partial g2 ¼ .05), but not to the global belief-based
attitude or intention to toward binge drinking (Mpeer ¼ 4.30, Mexpert ¼ 3.91; F(1, 243) ¼ 13.59,
p <. 01). For summary of all descriptive statistics in the analysis, see Table 1.

RQ2 main effects: Expectancy appeal versus valuative appeal


The second research question explored the role of message appeals in participants’ responses to
the anti-binge-drinking PSA. A MANCOVA indicated that, after controlling for covariates, effects
of message appeals on dependent variables were not significant (Wilks’s k ¼ .99, F(4, 240) ¼ .89,
p ¼ .47, partial g2 ¼ .02), meaning that there was no significant mean difference of four dependent
variables between expectancy and valuative appeals. Thus, no further analysis was conducted.

Hypotheses testing
Interaction effects: Matching between message sources and message appeals
A MANCOVA was conducted to determine the effect of matching (the interaction) between mes-
sage sources and message appeals on the four dependent variables. A significant interaction was
found indicating Wilks’s k ¼ .96, F(4, 240) ¼ 2.28, p < .05, partial g2 ¼ .04. Follow-up analysis of
covariance (ANCOVA) on the dependent variables revealed that significant interaction effects
were found on the global attitude toward binge drinking (F(1, 243) ¼ 5.61, p < .01, partial
g2 ¼ .02), the expectancy-value attitude toward binge drinking (F(1, 243) ¼ 6.39, p < .01, partial
g2 ¼ .03), and the binge-drinking intention (F(1. 243) ¼ 3.50, p < .05, partial g2 ¼ .02), but not on
the attitude toward the PSA (F(1, 243) ¼ 1.93, p ¼ . 17, partial g2 ¼ .01). Figure 1 reports the form
of the disordinal interaction for the expectancy-value attitude. This form is typical of the other
significant interaction effects (global attitude and binge-drinking intention).
Next we report the results of specific hypotheses. In order to test hypotheses 1 and 2, message
and source treatments were recorded into four categories representing various combinations of
matched and mismatched sources and appeals. A one-way MANCOVA was employed, consistent
with the prior two-way MANCOVA in the sense that the same set of covariates was entered into
the model. Hypotheses 1 and 2 were tested using pairwise comparisons.
In terms of interpreting the data results, one of the dependent variables, attitude toward
the PSA, is different from the other three dependent variables. For example, when it comes to
the attitude toward the PSA items (e.g., after reading the anti-binge drinking PSA, I think
that the message in the PSA is “likable,” etc.), higher scores mean a favorable attitudinal response.
On the contrary, the lower scores of the global attitude toward binge drinking, Fishbein’s attitude
toward binge drinking, and intention to binge drink all represent a positive attitudinal and behav-
ioral response (e.g., after reading the anti-binge drinking PSA, I think that binge drinking is
harmful/beneficial, unpleasant/pleasant, bad/good, and worthless/valuable). From these points of
views, the interpretation of the mean difference between source and appeal fit was expected that,
regarding attitude toward the PSA, the positive value of the mean difference between the source
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 13

Figure 1. Disordinal interaction effect for the expectancy-value attitude.

and appeal combination illustrates positive relations between them. In regard to the rest of the
dependent variables, the negative value of the mean difference between the source and appeal
match also describes positive relations among the fits.

H1: The expert–expectancy match


In order to test hypothesis 1, post hoc comparison tests reported specific hypothesis testing
results. Hypothesis 1(a) predicted that a PSA featuring an expert messenger in an expectancy
appeal would be more effective than a PSA featuring an expert messenger in a valuative appeal.
The results exhibited that the participants in the expert/expectancy message condition reported a
significantly higher mean than those in the expert/valuative message in terms of global attitude
toward binge drinking (MExpert/Expectancy ¼ 2.06, MExpert/Valuative ¼ 2.55, p < .01), expect-
ancy-value attitude toward binge drinking (MExpert/Expectancy ¼ 2.20, MExpert/
Valuative ¼ 0.25, p < .01), and binge-drinking intention (MExpert/Expectancy ¼ 2.37, MExpert/
Valuative ¼ 2.90, p < .05). However, there was no significant mean difference of attitude toward
PSA observed. Thus, three out of four sets of hypothesis 1(a) were supported.
Hypothesis 1(b) predicted that the expert/expectancy message would be more effective than a
PSA featuring a peer in an expectancy appeal (peer/expectancy). An expert/expectancy message
for attitude toward PSA yielded inferior results compared to peer/expectancy (M Expert/
Expectancy ¼ 3.99, M Peer/Expectancy ¼ 4.24 p < .05). In regard to global attitude toward binge
drinking, expectancy-value attitude toward binge drinking, and binge-drinking intention, there
were no significant differences between expert/expectancy and peer/expectancy. Therefore, this
hypothesis was not supported.

H2: The peer–valuative match


Hypothesis 2(a) predicted that a PSA featuring a peer messenger in a valuative appeal (peer/
valuative) would be more effective than a PSA featuring a peer messenger in an expectancy
14 H.-Y. (A.) AHN ET AL.

appeal (peer/expectancy). When it comes to peer/valuative and peer/expectancy combination,


there were no statistically significant differences in all dependent variables, leading to no support
for hypothesis 2(a).
Hypothesis 2(b) predicted that a PSA with the peer/valuative message would be more effective
than one featuring a doctor in a valuative appeal (expert/valuative). The results indicated that the
peer/valuative message yielded significant superiority over the expert/valuative message across all
the dependent variables, yielding full support for hypothesis 2(b). Specifically, there were signifi-
cant differences between peer/valuative and expert/valuative regarding attitude toward the PSA
(MPeer/Valuative ¼ 4.37, MExpert/Valuative ¼ 3.83, p < .001), global attitude toward binge drink-
ing (MPeer/Valuative ¼ 2.16, MExpert/Valuative ¼ 2.55, p < .05), expectancy-value attitude toward
binge drinking (MPeer/Valuative ¼ 1.98, MExpert/Valuative ¼ 0.25, p < .01), and binge-drink-
ing intention (MPeer/Valuative ¼ 2.45, MExpert/Valuative ¼ 2.90, p < .05).
The final research questions aimed to analyze the relative effectiveness of the two congruent
conditions (RQ3), expert/expectancy versus peer/valuative, and two incongruent cases (RQ4),
peer/expectancy versus expert/valuative. The results found that the peer/valuative congruent com-
bination resulted in a more positive attitude toward the PSA than expert/expectancy mix
(MExpert/Expectancy ¼ 3.99, MPeer/Valuative ¼ 4.37, p < .01), and the peer/expectancy incongru-
ent mix resulted in a more positive attitude toward the PSA than expert/valuative combination
(MExpert/Valuative ¼ 3.83, MPeer/Expectancy ¼ 4.24, p < .01). However, there were no significant
differences in the rest of the dependent variables. See Table 3 for a summary of the results of
hypotheses and research questions 3 and 4.

Covariate effects
Some significant effects of the covariates were observed. Subjects’ extent of alcohol use in the past
indicated that 57.4% had five or more drinks in one sitting in the past few months, while 35%
had five or more drinks in a row twice per week or three to five times per week (past binge

Table 3. Interaction hypotheses and research questions testing.


Mean difference (estimated marginal means)
Expectancy-Value
Attitude Global attitude toward attitude toward Intention to binge
toward PSA binge drinking binge drinking drink
Hypothesis 1(a), .16 .49 1.95 .52
expert/expectancy—expert/valuative
Hypothesis 1(b), .25 .20 .70 .23
expert/expectancy—peer/expectancy
Hypothesis 2(a), .14 .09 .49 .16
peer/valuative—peer/expectancy
Hypothesis 2(b), .54 .38 1.74 .45
peer/valuative—expert/valuative
Research question 3, .38 .11 .22 .07
expert/expectancy—peer/valuative
Research question 4, .405 .292 1.251 .291
expert/valuative—peer/expectancy
Note. All significance levels reported are based on one-tailed tests of directional hypotheses. Regarding interpretation of all
mean difference scores, a negative sign indicates that the first term in the difference score has a higher value than the first
term in the algebraic expression, while a positive sign indicates that the second term in the expression has a higher value.
Thus, a positive sign for the mean difference for the PSA score indicates greater effectiveness of the second term (combin-
ation of source and appeal) in the algebraic expression, whereas for all the other dependent variables, a negative sign for
the mean difference indicates greater effectiveness for the first combination. All hypotheses predicted greater effectiveness
for the first combination of source and appeal represented in the difference scores.
p < 0.05.
p < 0.01.
p < 0.001.
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 15

drinking experience, M ¼ 1.71, SD ¼ 1.00). Past binge-drinking experience as a covariate had


significant multivariate effects in the model (Wilks’s k ¼ .94, F(4, 240) ¼ 3.73, p < .05, partial
g2 ¼ .06). The past binge-drinking experience was significantly associated with two attitudes
toward binge-drinking measures and intention to binge drink dependent variables (r > .36,
p < .01), while relatively weakly correlated with attitude toward the PSA (r ¼ .12, p < .05).
Most recent binge-drinking experience as a covariate showed significant multivariate effects
(Wilks’s k ¼ .89, F(4, 240) ¼ 7.38, p < .01, partial g2 ¼ .11). Most recent drinking patterns in the
past 2 weeks revealed that 63% of them recently had five or more drinks in a row (defined as a
binge drinker), which means the majority of subjects have had binge-drinking experiences
(M ¼ 1.77, SD ¼ 1.17). This covariate was significantly associated with all (r > .37, p < .01), but
not with attitude toward the PSA measure.
Subjects perceived that the PSA message in the experimental study had a neutral quality
(M ¼ 3.87, SD ¼ 1.407), while as a covariate it had significant multivariate effects (Wilks’s k ¼ .49,
F(4, 240) ¼ 63.78, p < .01). The perceived message quality was significantly associated with atti-
tude toward the PSA (r ¼ . 69, p < .01), meaning that the higher the quality of the message was
perceived to be, the more favorable was the attitude toward the PSA identified. Also, it was min-
imally associated with global and Fishbein’s attitude toward binge-drinking measures (r < .21,
p < .01), but not with the intention to binge drink.
Finally, even if only a few were detected as having one or more experiences of these negative
outcomes, the majority of subjects (75%) did not have the experiences. The negative experience
of binge-drinking outcomes as a covariate had marginally significant multivariate effects (Wilks’s
k ¼ .96, F(4, 240) ¼ 2.02, p < .10). Also, a significant difference on two attitudes toward binge
drinking and intention to binge drink was observed (r > .32, p < .01). Somewhat surprisingly,
reporting past negative experiences was associated positively with PSA attitude and intention. The
results of covariate effects are presented in Table 2.

Discussion
While several studies in the communication field have examined the relative effectiveness of the sour-
ces or message appeals on message recipients’ behavior and attitudinal change, there is a relative lack
of research that investigates the relative effectiveness of PSA messenger sources, message appeals, and
the interactions between source types and message appeals in terms of the persuasiveness of PSA. This
study examines the direct and joint effects of message sources (expert vs. peer) and message appeals
(expectancy appeal vs. valuative appeal) on persuasion in the context of anti-binge-drinking PSA.
Given the fact that binge drinking is increasing among college students (Johnsson, Leifman,
and Berglund 2008), it is critical for communication researchers and PSA practitioners to develop
effective message strategies for preventing binge drinking. The major findings of this study
include the following: (1) Regardless of matching between sources and appeals, a peer was gener-
ally a more effective source in terms of attitude toward the anti-binge-drinking PSA. (2)
Messages were more effective when there was matching between source characteristics and
appeals (e.g., expert’s expertise/expectancy and peer’s similarity/valuative). Overall, the order of
relative effectiveness of matching of sources and message is expert/expectancy, peer/valuative,
expert/valuative, and peer/expectancy. (3) The inclusion of the expectancy-value attitude as a
dependent variable was observed as having diagnosing message effects, which can be of great
benefit in understanding health-related attitude research more precisely.

Interpretations of the findings


First, in comparing expert and peer messengers, a peer (perceived as having greater similarity to
the viewer than an expert) was more effective on attitudes toward the PSA than an expert source
16 H.-Y. (A.) AHN ET AL.

(as long as past and recent binge-drinking experience, negative experience of binge-drinking out-
comes, and message quality perception were controlled). This finding is consistent with prior
studies that found similar communicators were found to be more effective than dissimilar ones in
persuasion (Feick and Higie 1992; Andsager et al. 2006) and argued that young people acting as
peer models in health campaigns are more positively regarded than conventional sources (Atkin
1994). Even though no theory has been proposed to explain the superiority of similarity over
expertise, our findings are also consistent with the past findings suggesting that people are likely
to be more persuaded by similar sources because the similar source will be assumed to have simi-
lar preferences (Brock 1965; Simons, Berkowitz, and Moyer 1970; Naylor, Lamberton, and
Norton 2011). Thus, we can conclude that college students are likely to favor a similar messenger
who might share the value of binge drinking, the similar binge-drinking preferences, or even
negative feelings from binge drinking as a natural characteristic of young people, leading to favor-
able attitudes toward the anti-binge-drinking PSA. Thus, this result may suggest that there are
some contexts where experience-, observation-, and insight-based knowledge can substantially
outweigh conventional expertise.
The college students’ preference to a peer source is also pronounced in the significant inter-
action effects. That is, even if college students perceived a match between expert and expectancy
message appeals, they still tended to favor the PSA delivered by the peer, regardless of the mes-
sage appeals. Indeed, the peer/expectancy combination (mismatched condition) yielded a more
positive attitude toward a PSA than expert/expectancy (matched condition). Also, for the congru-
ent condition of peer/valuative versus expert/expectancy, the peer/valuative combination showed
a better effect on PSA attitude than did the expert/expectancy combination (RQ3). Similarly, the
two incongruent conditions, when compared, exhibited the same pattern; that is, the peer/expect-
ancy treatment resulted in a more positive PSA attitude than did the expert/valuative treatment
(RQ4). These results strongly support the idea that a similar source may be more effective on
favorable attitudes toward PSA than an expert source, even when mismatched with expectancy
appeals, at least among college students.
The fact that source similarity (peer) produced no significant main effects on the other
dependent variables, such as attitudes toward binge drinking and binge-drinking intention, indi-
cates the difficulty of persuading college students regarding binge drinking, particularly among
those who already have experience with binge drinking. Drinking is a social behavior deeply
ingrained in college life. Health practitioners should find multiple pathways through which col-
lege students can understand and are encouraged to change their perceptions regarding the social
mandate that encourages binge drinking. It may not be coincidental that many campus-wide
anti-binge-drinking campaigns use social appeals to shape more sound social norms and cultural
values (i.e., Social Norm Campaigns).
In terms opf another main effect, this study also found that there were no significant differen-
ces between expectancy and valuative appeals on all the dependent variables. One potential
explanation for the null main effects of message appeals may be ascribed to the fact that the mes-
sages were designed to focus solely on the negative outcomes and negative valences. In other
words, the message content used in these message stimuli was based largely on the content of
negative health and social threats. It is possible that the negatively framed threats offset the effects
of message appeals. Another potential explanation is that the covariate, perceived message quality,
may fully mediate the direct impact of the message treatment. Based on Areni and Lutz’s (1988)
argument quality model, argument quality is the product of perceived strength (expectancy) and
valence (valuative). That is, the null main effects of message appeals may imply equivalence of
two appeals regarding perceived quality in terms of their roles in persuasion.
With respect to the second major finding, message-matching effects, the results suggest that
when it comes to valuative appeals (focusing on negative evaluation of bad outcomes), a similar
source is a better match than an expert source. Considering the manipulated source
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 17

characteristics (peer similarity), it is possible that valuative appeals enhance the effectiveness of
source similarity. With regard to attitude toward binge drinking and intention to binge drink, the
results show that an expert/expectancy match is significantly more persuasive than the expert/
valuative combination. This result coincided with the existing literature on the argument of
Salmon and Atkin (2003), which states that a medical authority may be a better source for belief-
oriented messages (expectancy appeals) because his or her expertise may enhance the believability
of the messages that are related to negative consequences.
Finally, in this study, Fishbein’s expectancy-value attitude criterion yielded results statistically
similar to those of the global attitude measure, but generated a higher F value than did the global
attitude measure. Also, a zero-order correlation between global attitude and Fishbein’s attitude
indicated only a moderately positive correlation (r ¼ .56). The observed correlation supports the
argument that Fishbein’s attitude structure, when based on persuasive message claims, may not
be equivalent to a corresponding measure of the global attitude. Thus, including Fishbein’s atti-
tude as a dependent variable in addition to global attitude is reasonable since each measure taps
a somewhat different aspect of the attitude construct. Indeed, Ajzen and Fishbein (2008) argued
that the claim-based structural equation has potential diagnostic power for assessing the effective-
ness of specific message claims. For example, among a number of message claims, it would be
possible to identify which claims work in terms of the level of belief and/or valence generated.
In order to explore the diagnostic potential of the expectancy-value model (Ajzen and Fishbein
2008), we analyzed separately the impacts of each health-related claim in our test messages. In
this study, eight parallel message claims were selected for each message treatment: symptoms of
vomiting, blackouts, pounding hangover, liver cancer, missing classes, damaging academic career,
losing family support, and alienating friends. All of these outcomes are negative and they include
both short- and long-term physical effects as well as personal and social effects. When each of
these eight expectancy-value components of the Fishbein model was calculated and analyzed as a
separate communication effect, it was observed that the eight attitudinal components showed
markedly different post-test values. Short-term physical claims were associated with the most
negative binge-drinking attitude (greatest effectiveness), followed by long-term physical effects,
personal effects, and social effects, respectively. This generally would suggest that persuasive mes-
sages emphasizing short-term negative physical outcomes of binge drinking possess greater
impact than other types of negative claims. When the eight claim-based attitudinal components
were compared statistically using paired-sample t-tests, each differed from every other component
at p < .001. This substantial variation in the relative effectiveness of individual claims has diagnos-
tic value in that it helps us identify which claims “worked” and which did not, which require
additional persuasion effort and which claims are already accepted. Additionally, parallel analysis
within message source treatments could provide insights into how different message sources
influence the ranking of attitudinal components. Further diagnostic insight could be obtained by
regressing the eight attitudinal components against global attitude toward binge drinking. In this
way, we could identify which claims were most salient as attitudinal determinants.

Implications
These findings have several practical and theoretical implications for communication researchers
and PSA practitioners. First, if their primary goal is to produce a positive attitude toward the
PSA among college students, the peer source is suggested. Regardless of matching between sour-
ces and appeals, a peer was generally perceived as a more effective message source than an expert
in producing a favorable attitude toward an anti-binge-drinking PSA. Given Atkin’s (1994) claim
that message recipients tend to accept the message claimed by similar sources, it appears that
college students perceive a peer source to be a person who shares their values, experiences, and
feelings. In the context of college life, binge drinking is such a social expectation that a peer
18 H.-Y. (A.) AHN ET AL.

source may yield a sense of expertise to students, rather than that of a medical spokesperson,
who, despite formal training, may be viewed as inexperienced or as a nonexpert in regard to
sharing the negative experiences resulting from binge drinking. This bears resemblance to
O’Keefe’s (2002) note that “the effects of perceived similarities and dissimilarities on judgments
of communicator expertise depend on whether, and how, the receiver perceives these as relevant
to the issue at hand” (203).
Second, if the primary goal were to influence college students’ attitude toward the behavior
and/or behavioral intention, the study would suggest that the most effective message/source com-
bination to deliver is an expectancy message delivered by an expert. With respect to the match-
up between source and message appeals, the matched combinations of expert/expectancy and
peer/valuative were perceived to be a strong match as long as the message has the same quality,
while the least effective is the mismatched condition, expert/valuative; peer/expectancy was only
moderately effective. These interaction findings provide evidence that effectiveness depends upon
the sources being matched with or related to the message appeal. In applying Salmon and Atkin’s
(2003) argument that the presence of experts in health-related fields increases the audience’s
belief in the recommended claims (i.e., Tay et al. 2001), it is notable that the expectancy appeal
focusing on the audience’s subjective beliefs regarding the consequences of binge drinking is
appropriate for delivery by an expert source, resulting in greater response efficacy. Also, a similar
source tends to heighten the shared felt experience component contained in the valuative appeal,
which also enhances response efficacy. PSA message developers should consider using these
matching effects in terms of the “who says how” perspective.
Finally, the findings showed that the expectancy-value attitude model, which is used infre-
quently to assess the effectiveness of PSAs and health campaigns, can be of great benefit in
understanding health-related attitudes more precisely. In addition to this diagnostic advantage,
assessment of attitudinal components may possess greater measurement validity than global atti-
tudinal responses. In areas of social research dealing with sensitive issues such as substance abuse,
it is possible that some participants in an experiment or survey tend to avoid revealing their true
attitudes toward the issues. Thus, the global attitude measure (i.e., “Do you think smoking is
beneficial/not beneficial, good/bad?”) might generate biased feedback from participants due to
their unwillingness to express their true feelings directly. In contrast, since the expectancy-value
attitude model enables participants to specify the likelihood of encountering the consequences of
the undesired behavior and assesses the valence of these outcomes, this may yield less biased
feedback. Accordingly, the attitude structure models can be used to achieve diagnostic goals and
to enhance measurement validity (Wilkie and Pessemier 1973).

Future research directions


Despite these findings, there remain several limitations inherent in this study. First, although the
focus of the study was to compare two kinds of source characteristics (expertise vs. similarity)
and their matching effects with message appeals (expectancy vs. valuative), there are many other
types of message sources applicable to the development of effective PSA messages. Expertise and
similarity source characteristics are known to be relatively effective in changing health percep-
tions. There is still much to be learned about the effects of the various dimensions of source cred-
ibility (i.e., source competence, trustworthiness, goodwill) (Phua 2016) and source attractiveness
(i.e., source attractiveness, likability, and familiarity) in the context of anti-binge-drinking PSAs.
Hence, it is important that future research examines how different messenger types would influ-
ence college students’ attitudes and behavioral intentions.
Second, the PSA message appeals explored in this study included only a list of negative out-
comes (as compared to positive), such as physical threats and social threats. Although threats or
negative appeals are known to be relatively effective in changing health perceptions, positive
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 19

appeals might serve to encourage change by linking desired behaviors with positive images (Atkin
2001). Thus, various kinds of sources and message appeals and their match-up should be
explored in future research. Such exploration will help develop more effective health messages to
specific target audiences.
Third, the types of negative binge-drinking outcomes used in this study were selectively adopted
from the College Drinking Fact Sheet (NIAAA, 2012). However, it remains to be seen whether the
messages adopted for this study fully reflect college students’ current salient beliefs and experience
with binge drinking. Ajzen and Fishbein (1980) argued that it is important to identify the target’s
common prominent beliefs to develop effective communications. They also suggested different types
of beliefs, such as primary beliefs, proximal beliefs, target beliefs, inferential beliefs, external beliefs,
and so on. Wolburg (2016) also noted that to reach a better understanding of how to curb college
students’ binge drinking, researchers should attempt to apprehend what truly influences their risky
behavior at the individual level and then create PSA messages matching with the college students’
genuine experience. For instance, when a college student holds an incorrect belief such as
“pluralistic ignorance,” meaning that the student mistakenly believes that he or she drinks less than
other students (Miller and McFarland 1987), or “false consensus,” meaning that the amount of
drinking of peers is similar to his or her own (Ross, Greene, and House 1977), PSAs can address
an authentic story (as opposed to the PSA message from the administration of a college) of how
excessive drinking behavior can result in losing one’s best friends. Applying these suggestions, PSA
messages can focus on college students’ different levels of beliefs and experiences associated with
binge drinking, such as social anxiety (Wolburg, 2016), drinking as a ritual behavior (Treise,
Wolburg, and Otnes 1999), or the consequences of losing self-control (Piquero, Gibson, and
Tibbetts 2002). As an example, a PSA can show an unsuccessful binge drinker suffers from poor
self-control and foolish decision making in the short term, leading to arrest, rape, and fatal acci-
dents. Also, the message scenarios could be different based on the genders’ perceived risks (i.e., for
female students: being raped, being pregnant by an unknown man; for male students: getting the
girl pregnant, being charged with sexual assault).
Fourth, this study used transcribed PSAs, rather than a broadcast stimulus, in order to avoid
possible biases, including PSA spokesperson’s voice and gender. Although the content of the four
PSAs used in this study could be applied to various PSA message contexts, it is important to
acknowledge that college students are more likely to spend time using social media than trad-
itional media (Burst Media 2007). Future studies should attempt to examine anti-binge-drinking
campaign issues through college students’ primary media, namely, social media (Maderazo 2008;
Lee and Ahn 2013), interactive media (i.e., word-of-mouth communication, consumer comments,
etc.) (Kareklas, Muehling, and Weber 2015), or personalized media (i.e., mobile) (Hovland, 2015).
Fifth, although the present study offered a contribution to the literature concerning how to
develop effective anti-binge-drinking PSAs, it would be a serious oversight to conclude that PSAs
or health prevention interventions alone will solve the issue. Hovland (2015) pointed out that
alcohol ads and related marketing strategies and tactics have been more sophisticated due to the
rise of new and cross media, and the interactive nature of those media platforms, and the
increased abilities for marketers and advertisers to send precise messages to target audiences.
Mart (2011) found that social media platforms (i.e., Facebook, YouTube, etc.) have offered a var-
iety communication channels for alcoholic beverage marketing. In this context, it is possible that
individuals can be exposed to both an anti-binge-drinking PSA and alcohol-promoting advertising
(Ahn, Wu, Kelly and Haley 2011) within the same social media page, or two messages across dif-
ferent media channels. Clearly, more research is needed to illuminate how college students would
deal with these conflicting messages in advertising and media.
Last but not least, although we used college students as the samples, our findings should be
replicated among college students in other regions, among noncollege students (i.e., minorities
including Hispanics or youth), and on various health topics other than binge drinking.
20 H.-Y. (A.) AHN ET AL.

Conclusion
Despite some limitations, this study contributes to the literature on persuasive PSA messages by
demonstrating the diagnostic effects of message sources and appeals and moves one step closer
toward the efforts of reducing college students’ binge-drinking behavior. Along with the fact that
individuals rarely change their personal behaviors (Hill 2004), it is important to understand that
drinking, especially excessive drinking, is a social behavior embedded in communities and cul-
tures. College students have a growing and higher rate of binge drinking and the context in
which they are currently involved will not easily change. Even with these circumstances, we rec-
ommend that both academics and practitioners invite more applicable and promising theories to
determine the varying types of effective messengers and message appeals.
Among other points, this study urges more investigation on a variety of important compo-
nents, such as potential competing messages against PSAs, appropriate message relevancy, and
adequate message exposure and its timing through effective vehicles that are related to the tar-
geted audience. These efforts are an important step in achieving the main goal of increasing audi-
ence knowledge and awareness, as well as changing attitudes and behavior through public service
advertising messages (Atkin and Freimuth 2001).

References
Ahn, H.-Y. (Anthony), Lei Wu, Stephanie Kelly, and Eric J. Haley. 2011. “A Qualitative Study of College Student
Responses to Conflicting Messages in Advertising: Anti-Binge Drinking Public Service Announcements versus
Wine Promotion Health Messages.” International Journal of Public Health 56 (3):271–9.
Ajzen, I., and Martin Fishbein. 1980. Understanding Attitudes and Predicting Social Behavior. New York: Prentice
Hall.
Ajzen, I., and Martin Fishbein. 2008. “Scaling and Testing Multiplicative Combinations in the Expectancy–Value
Model of Attitudes.” Journal of Applied Social Psychology 38 (9):2222–47. doi:10.1111/j.1559-1816.2008.00389.x
Anderson, R. B., and Priscilla Y. McMillion. 1995. “Effects of Similar and Diversified Modeling on African
American Women’s Efficacy Expectations and Intentions to Perform Breast Self-Examination.” Health
Communication 7 (4):327–43.
Andsager, J. L., Erica W. Austin, and Bruce E. Pinkleton. 2002. “Gender as a Variable in Interpretation of Alcohol-
Related Messages.” Communication Research 29 (3):246–69.
Andsager, J. L., Victoria Bemker, Hong-Lim Choi, and Vitalis Torwel. 2006. “Perceived Similarity of Exemplar
Traits and Behavior.” Communication Research 33 (1):3–18.
Areni, C. S., and Richard J. Lutz. 1988. “The Role of Argument Quality in the Elaboration Likelihood Model.” In
Advances in Consumer Research, edited by Micheal J. Houston, vol. 15, 197–203. Provo, UT: Association for
Consumer Research.
Atkin, C. K. 1994. “Designing Persuasive Health Messages.” In Effective Dissemination of Clinical Health
Information, No. 95-0015, edited by Lee Sechrest, Thomas E. Backer, Everett M. Rogers, Timothy F. Campbell,
and Mary L. Grady, 99–110. Rockville, MD: Public Health Service, Agency for Health Care Policy and Research
Publication.
Atkin, C. K. 2001. “Theory and Principles of Media Health Campaigns.” In Public Communication Campaigns,
edited by Ronald E. Rice and Charles Atkin, 49–68. Thousand Oaks, CA: Sage.
Atkin, C. K., and Vicki S. Freimuth. 1989. “Formative Evaluation Research in Campaign Design.” In Public
Communication Campaigns, edited by Ronald Rice and Charles Atkin, 131–50. Newbury Park, CA: Sage.
Atkin, C. K., and V. S. Freimuth. 2001. “Formative evaluation research in campaign design.” In R. E. Rice and C.
K. Atkin, (Eds.), Public communication campaigns (3rd ed., pp. 125–145). Thousand Oaks: Sage.
Boekeloo, B. O., M. G. Novik, and E. N. Bush. 2011. “Drinking to Get Drunk among Incoming Freshmen College
Students.” American Journal of Health Education 42 (2):88–95.
Brock, T. C. 1965. “Communicator-Recipient Similarity and Decision Change.” Journal of Personality and Social
Psychology 1 (6):650–4.
Brown, J. J., and Peter H. Reingen. 1987. “Social Ties and Word-of-Mouth Referral Behavior.” Journal of Consumer
Research 12:1–16.
Burst Media. 2007. “Survey: To Reach College Students, Brands Need to Use Internet.” Accessed June, 2015. http://
www.marketingcharts.com/television/survey-to-reach-college-students-brands-need-to-use-internet-1389/burst-media-
college-time-spent-with-mediajpg/.
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 21

Centers for Disease Control and Prevention. 2012. “CDC Fact Sheets.” Accessed February, 2014. http://www.ded.
gov/vitalsigns/BingsDrinking/index.html.
Cialdini, R. B. 2001. “Harnessing the Science of Persuasion.” Harvard Business Review 79:72–81.
DeJong, W. 2002. “The Role of Mass Media Campaigns in Reducing High-Risk Drinking among College Students.”
Journal of Studies on Alcohol, Supplement 14 (suppl 14):182–92.
DeJong, W., and Charles K. Atkin. 1995. “A Review of National Television PSA Campaigns for Preventing
Alcohol-Impaired Driving 1987–1992.” Journal of Public Health Policy 16 (1):59–79.
Dillard, J. P. 1994. “Rethinking the Study of Fear Appeals: An Emotional Perspective.” Communication Theory 4
(4):295–323.
Feick, L., and Robin A. Higie. 1992. “The Effects of Preference Heterogeneity and Source Characteristics on Ad
Processing and Judgments about Endorsers.” Journal of Advertising 21 (2):9–24.
Festinger, L. 1954. “A Theory of Social Comparison Processes.” Human Relations 7 (2):117–40.
Fishbein, M., and Bertram Raven. 1962. “The AB Scales: An Operational Definition of Belief and Attitude.”
Human Relations 15 (1):35–44.
Fishbein, M., and Icek Ajzen. 1975. Belief, Attitude, Intention, and Behavior: An Introduction to Theory and
Research. Reading, MA: Addison-Wesley.
Fishbein, M., and Icek Ajzen. 1981. “Acceptance, yielding and impact.” In Expectancy Responses in Persuasion,
edited by Richard E. Petty, Thomas M. Ostrom, and Timothy C. Brock, 339–59. Hillsdale, MI: Lawrence
Erlbaum Associates.
Friedman, H. H., and Linda Friedman. 1979. “Endorser Effectiveness by Product Type.” Journal of Advertising
Research 19 (3):63–71.
Gilly, M. C., John L. Graham, Mary F. Wolfinbarger, and Laura J. Yale. 1998. “A Dyadic Study of Interpersonal
Information Search.” Journal of the Academy of Marketing Science 26 (2):83–100.
Hair, J. F., Rolph E. Anderson, Ronald L. Tatham, and William C. Black. 1998. Multivariate Data Analysis. Upper
Saddle River, NJ: Prentice Hall.
Haas, R. G. 1981. “Effects of Source Characteristics on Expectancy Responses and Persuasion.” In Expectancy
Responses in Persuasion, edited by Richard E. Petty, Thomas M. Ostrom, and Timothy C. Brock, 141–95.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Herr, P. M., Frank R. Kardes, and John Kim. 1991. “Effects of Word-of-Mouth and Product-Attribute Information
on Persuasion: An Accessibility-Diagnosticity Perspective.” Journal of Consumer Research 17 (4):454–62.
Hill, L. 2004. Alcohol Health Promotion via Mass Media: The Evidence on (in) Effectiveness. In Eurocare
“Bridging the Gaps” Conference, Warsaw.
Hingson, R. W., W. Zha, and E. R. Weitzman. 2009. “Magnitude of and Trends in Alcohol-Related Mortality and
Morbidity among U.S. College Students Ages 18–24, 1998–2005.” Journal of Studies on Alcohol and Drugs,
Supplement (s16):12–20.
Hoeken, H., Marianne Starren, Catherine Nickerson, Rogier Crijns, and Corine Van Den Brandt. 2007. “Is It
Necessary to Adapt Advertising Appeals for National Audiences in Western Europe?” Journal of Marketing
Communications 13 (1):19–38.
Hovland, C. I., Irving L. Janis, and Harold H. Kelley. 1953. Communication and Persuasion. New Haven, NY: Yale
University.
Hovland, R. 2015. “Alcoholic Beverage Advertising: An Integrative Interdisciplinary Review.” Journal of Current
Issues & Research in Advertising 36 (1):88–114. doi:10.1080/10641734.2014.912598.
Johnsson, Kent O., Anders Leifman, and Mats Berglund. 2008. “College Students’ Drinking Patterns: Trajectories
of Audit Scores during the First Four Years at University.” European Addiction Research 14 (1):11–8.
Kahle, L. R., and Pamela M. Homer. 1985. “Physical Attractiveness of the Celebrity Endorser: A Social Adaptation
Perspective.” Journal of Consumer Research 11 (4):954–61.
Kamins, M. A. 1990. “An Investigation into the ‘Match-Up’ Hypothesis in Celebrity Advertising: When Beauty
May Be Only Skin Deep.” Journal of Advertising 19 (1):4–13.
Kamins, M. A., and Kamal Gupta. 1994. “Congruence between Spokesperson Image and Product Type: A ’Match-
Up’ Hypothesis Perspective.” Psychology and Marketing 11 (6):569–88.
Kareklas, I., Darrel D. Muehling, and T. J. Weber. 2015. “Reexamining Health Messages in the Digital Age: A
Fresh Look at Source Credibility Effects.” Journal of Advertising 44 (2):88–104.
Lee, Yoon-Joo, and Ho-Young (Anthony) Ahn. 2013. “Interaction Effects of Perceived Sponsor Motives and
Facebook Credibility on Willingness to Visit Social Cause Facebook Page.” Journal of Interactive Advertising 13
(1):41–52.
Leiss, W., Stephen Kline, and Sut Jhally. 1986. Social Communication in Advertising: Persons, Products and Images
of Well-Being. Toronto, Canada: Methuen.
Lepkowska-White, E., T. G. Brashear, and M. G. Weinberger. 2003. “A Test of Ad Appeal Effectiveness in Poland
and the United States: The Interplay of Appeal, Product, and Culture.” Journal of Advertising 32 (3):57–67.
Lynn, J. R. 1974. “Effects of Persuasive Appeals in Public Service Advertising.” Journalism Quarterly 51 (4):622–30.
22 H.-Y. (A.) AHN ET AL.

Lynch, J., and Drue Schuler. 1994. “The Matchup Effect of Spokesperson and Product Congruency: A Schema
Theory Interpretation.” Psychology and Marketing 11 (5):417–45.
MacKenzie, S. B., and Richard J. Lutz. 1989. “An Empirical Examination of the Structural Antecedents of Attitude
toward the Ad in an Advertising Pretesting Context.” Journal of Marketing 53 (2):48–65.
Madden, T. J., Chris T. Allen, and Jacquelyn L. Twible. 1988. “Attitude toward the Ad: An Assessment of Diverse
Measurement Indices under Different Processing Sets.” Journal of Marketing Research 25 (3):242–52.
Maderazo, J. W. 2008. “The Benefits and Pitfalls of Using Social Media for Reporting.” Mediashift. Accessed
September, 2015. http://www.pbs.org/mediashift/2008/01/the-benefits-and-pitfalls-of-using-social-media-for-
reporting017.html.
McCroskey, J. C. 1966. “Scales for the Measurement of Ethos.” Speech Monographs 33 (1):65–72.
Miller, D. T., and C. McFarland. 1987. “Pluralistic Ignorance: When Similarity Is Interpreted as Dissimilarity.”
Journal of Personality and Social Psychology 53 (2):298–305.
Moore, J. 2004. “Public Service Advertising and the Effects of Using the Actor Portrayal Label.” Annual convention
of the Association for Education in Journalism and Mass Communication, Toronto, Canada.
Myrick, R. 1998. “In Search of Cultural Sensitivity and Inclusiveness: Communication Strategies Used in Rural
HIV Prevention Campaigns Designed for African Americans.” Health Communication 10 (1):65–85.
National Institute on Alcohol Abuse and Alcoholism (NIAAA). 2012. “College Drinking.” Accessed September,
2015. http://www.niaaa.nih.gov/alcohol-health/special-populations-co-occurring-disorders/college-drinking.
Naylor, R. W., Cait P. Lamberton, and Davie A. Norton. 2011. “Seeing Ourselves in Others: Reviewer Ambiguity,
Egocentric Anchoring, and Persuasion.” Journal of Marketing Research 48 (3):617–31.
Ohanian, R. 1991. “The Impact of Celebrity Spokespersons’ Perceived Image on Consumers’ Intention to
Purchase.” Journal of Advertising Research 31 (1):46–54.
O’Keefe, D. J. 1990. Persuasion: Theory and Research. Newbury Park, CA: Sage.
O’Keefe, D. J. 2002. Persuasion: Theory and Research. 2nd ed. Thousand Oaks, CA: Sage.
Pallak, S. R., Enid Murroni, and Juliann Koch. 1983. “Communicator Attractiveness and Expertise, Emotional and
Rational Appeals, and Persuasion: A Heuristic versus Systematic Processing Interpretation.” Social Cognition 2
(2):122–41.
Pedersen, Eric R., and Joseph LaBrie. 2007. “Partying before the Party: Examining Prepartying Behavior among
College Students.” Journal of American College Health 56 (3):237–45.
Perloff, R. M. 2008. The Dynamics of Persuasion: Communication and Attitudes in the 21st Century. New York,
NY: Lawrence Erlbaum.
Perse, E. M., Amy I. Nathanson, and Douglas M. McLeod. 1996. “Effects of Spokesperson Sex, Public Service
Announcement Appeal, and Involvement on Evaluations of Safe-Sex PSAs.” Health Communication 8 (2):
171–89.
Petty, R. E., John T. Cacioppo, and David Schumann. 1983. “Central and Peripheral Routes to Advertising
Effectiveness: The Moderating Role of Involvement.” Journal of Consumer Research 10 (2):135–46.
Phua, J. 2016. “The Effects of Similarity, Parasocial Identification, and Source Credibility in Obesity Public Service
Announcements on Diet and Exercise Self-Efficacy.” Journal of Health Psychology 21 (5):699–708.
Piquero, A. R., Chris L. Gibson, and Stephen G. Tibbetts. 2002. “Does Self-Control account for the Relationship
between Binge Drinking and Alcohol-Related Behaviours?” Criminal Behaviour and Mental Health 12 (2):135.
Price, L. L., Lawrence F. Feick, and Robin A. Higie. 1989. “Preference Heterogeneity and Co-Orientation as
Determinants of Perceived Informational Influence.” Journal of Business Research 19 (3):227–42.
Ray, G., Eileen Berlin Ray, and Christopher J. Zahn. 1991. “Speech Behavior and Social Evaluation: An
Examination of Medial Messages.” Communication Quarterly 39 (2):119–29.
Ross, Lee, David Greene, and Pamela House. 1977. “The False Consensus Effect: An Egocentric Bias in Social
Perception and Attribution Processes.” Journal of Experimental Social Psychology 13 (3):279–301.
Salmon, C. T., and Charles Atkin. 2003. “Using Media Campaigns for Health Promotion.” In Handbook of Health
Communication, edited by Teresa L. Thompson, Alicia M. Dorsey, Katherine I. Miller, and Roxanne Parrott,
449–72. Mahwah, NJ: Lawrence Erlbaum Associates.
Sheth, J. N. 1974. “An Investigation of Relationships among Evaluative Beliefs, Affect, Behavioral Intention, and
Behavior.” In Consumer Behavior: Theory and Application, edited by John U. Farley, John A. Howard, and
L. Winston Ring, 89–114. Boston, MA: Allyn & Bacon.
Sherif, M., and Carolyn Wood Sherif. 1964. Reference Groups: Exploration into Conformity and Deviation of
Adolescent. New York, NY: Harper & Row.
Silvia, P. J. 2005. “Deflecting Reactance: The Role of Similarity in Increasing Compliance and Reducing
Resistance.” Basic and Applied Social Psychology 27 (3):277–84.
Simons, H. W., Nancy N. Berkowitz, and R. John Moyer. 1970. “Similarity, Credibility and Attitude Change: A
Review and a Theory.” Psychological Bulletin 73 (1):1–16.
Smith, R. E., and William R. Swinyard. 1983. “Attitude-Behavior Consistency: The Impact of Product Trial versus
Advertising.” Journal of Marketing Research 20 (3):257–67.
JOURNAL OF CURRENT ISSUES & RESEARCH IN ADVERTISING 23

Smith, K. H., and Mary Ann Stutts. 2003. “Effects of Short Term Cosmetic versus Long Term Health Fear Appeals
in Antismoking Advertisements on the Smoking Behaviour of Adolescents.” Journal of Consumer Behaviour 3
(2):157–77.
Snyder, L. B. 2001. “How Effective Are Mediated Health Campaigns?” In Public Communication Campaigns, edited
by Ronald F. Rice and Charles K. Atkin. Newbury Park, CA: Sage.
Solomon, M. R., Richard D. Ashmore, and Laura C. Longo. 1992. “The Beauty Match-up Hypothesis: Congruence
between Types of Beauty and Product Images in Advertising.” Journal of Advertising 21 (4):23–34.
Sternthal, B., Lynn W. Phillips, and Ruby Dholakia. 1978. “The Persuasive Effect of Source Credibility: A
Situational Analysis.” Public Opinion Quarterly 42 (3):285–314.
Stephenson, M. T. 2003. “Examining Adolescents’ Responses to anti-Marijuana PSAs.” Human Communication
Research 29 (3):343–69.
Stephenson, M. T., and Kim Witte. 1998. “Fear, Threat, and Perceptions of Efficacy from Frightening Skin Cancer
Messages.” Public Health Reviews 26 (2):147–74.
Substance Abuse and Mental Health Services Administration (SAMHSA). 2006. Results from the 2005 National
Survey on Drug Use and Health: National Findings, No. SMA 06-4194. Rockville, MD: Office of Applied Studies.
DHHS Publication.
Tay, R. S., Barry C. Watson, Olivia Radbourne, and Ben De Young. 2001. “The Influence of Fear Arousal and
Perceived Efficacy on the Acceptance and Rejection of Road Safety Advertising Messages.” In 2001: Road Safety
Research, Policing and Education Conference (Regain the Momentum), Melbourne.
Treise, D., J. Wolburg, and C. Otnes. 1999. “Understanding the “Social Gifts” of Drinking Rituals: An Alternative
Framework for PSA Developers.” Journal of Advertising 28 (2):17–31.
Wechsler, H., Andrea Davenport, George Dowdall, Barbara Moeykens, and Sonia Castillo. 1994. “Health and
Behavioral Consequences of Binge Drinking in College.” JAMA: The Journal of the American Medical
Association 272 (21):1672–7.
Wechsler, H., and Nancy Isaac. 1992. “‘Binge’ drinkers at Massachusetts colleges. Prevalence, Drinking Style, Time
Trends, and Associated Problems.” JAMA 267 (21):2929–31.
Wolburg, J. 2001. “The Risky Business of Binge Drinking among College Students: Using Risk Models for PSAs
and anti-Drinking Campaigns.” Journal of Advertising 30 (4):23–39.
Wolburg, J. 2016. “Insights for Prevention Campaigns: The Power of Drinking Rituals in the College Student
Experience from Freshman to Senior Year.” Journal of Current Issues & Research in Advertising 37 (1):80–94.
doi:10.1080/10641734.2015.1119770
Wilkie, W. L., and Edgar A. Pessemier. 1973. “Issues in Marketing Use of Multi-Attribute Attitude Models.”
Journal of Marketing Research 10 (4):428–41.
Zhang, Y., and Betsy D. Gelb. 1996. “Matching Advertising Appeals to Culture: The Influence of Products’ Use
Conditions.” Journal of Advertising 25 (3):29–46.
Zhang, J. 2010. “The Persuasiveness of Individualistic and Collectivistic Advertising Appeals among Chinese
Generation-X Consumers.” Journal of Advertising 39 (3):69–80.

Appendix: Stimuli—Public service advertising messages


PSA transcripts (Expert/Expectancy)
Doctor: Hi, I’m Doctor Pat Michaels. I’m the Director of the North Georgia Addiction Rehabilitation Center. So, I
have many years of experience in the medical field. I am here to discuss the dangers of binge drinking. For men, 5
or more drinks in two hours and for women, 4 or more drinks in two hours is defined as binge drinking.
Doctor: If you binge drink, you are likely to experience these symptoms related to binge drinking—vomiting,
blackouts, and a pounding hangover the next day. If binge drinking happens repeatedly over enough time, your
body will very likely suffer from irreversible damage such as liver cancer.
Doctor: Other problems? You are 8 times more likely than non-binge drinkers to miss class and eventually damage
your academic career. Binge drinking makes it more likely that you will lose family support or even your close
friends. It may feel good to be the “life” of the party, but it is terrible to actually destroy your own life.
For more information, visit w-w-w-dot-binge-drinking-prevention-dot-o-r-g or call 1-800-300-2222. A public ser-
vice message brought to you by the North Georgia Anti-Addiction Alliance.

PSA transcripts (Peer/Expectancy)


UGA student: Hi, I’m Pat Michaels. I’m a UGA undergraduate student. So, in many ways, I am just like you.
I am here to discuss the dangers of binge drinking. For men, 5 or more drinks in two hours and for women, 4 or
more drinks in two hours is defined as binge drinking.
24 H.-Y. (A.) AHN ET AL.

UGA student: If you binge drink, you are likely to experience these symptoms related to binge drinking—vomiting,
blackouts, and a pounding hangover the next day. If binge drinking happens repeatedly over enough time, your
body will very likely suffer from irreversible damage such as liver cancer.
UGA student: Other problems? You are 8 times more likely than non-binge drinkers to miss class and eventually
damage your academic career. Binge drinking makes it more likely that you will lose family support or even your
close friends. It may feel good to be the “life” of the party, but it is terrible to actually destroy your own life.
For more information, visit w-w-w-dot-binge-drinking-prevention-dot-o-r-g or call 1-800-300-2222. A public ser-
vice message brought to you by the North Georgia Anti-Addiction Alliance.

PSA transcripts (Expert/Valuative)


Doctor: Hi, I’m Doctor Pat Michaels. I’m the Director of the North Georgia Addiction Rehabilitation Center. So, I
have many years of experience in the medical field. I am here to discuss the dangers of binge drinking. For men, 5
or more drinks in two hours and for women, 4 or more drinks in two hours is defined as binge drinking.
Doctor: Think about how bad it is to have vomiting, blackouts, and a pounding hangover. Consider how devastat-
ing it is to have irreversible damage such as liver cancer.
Doctor: Other problems? Consider how terrible it is if missing classes damages your academic career in the end.
Think about how bad it is to lose family support or to lose friends.
It may feel good to be the “life” of the party, but it is terrible to actually destroy your own life. For more informa-
tion, visit w-w-w-dot-binge-drinking-prevention-dot-o-r-g or call 1-800-300-2222. A public service message
brought to you by the North Georgia Anti-Addiction Alliance.

PSA transcripts (Peer/Valuative)


UGA student: Hi, I’m Pat Michaels. I’m a UGA undergraduate student. So, in many ways, I am just like you. I am
here to discuss the dangers of binge drinking. For men, 5 or more drinks in two hours and for women, 4 or more
drinks in two hours is defined as binge drinking.
UGA student: Think about how bad it is to have vomiting, blackouts, and a pounding hangover. Consider how
devastating it is to have irreversible damage such as liver cancer.
UGA student: Other problems? Consider how terrible it is if missing classes damages your academic career in the
end. Think about how bad it is to lose family support or to lose friends.
It may feel good to be the “life” of the party, but it is terrible to actually destroy your own life. For more informa-
tion, visit w-w-w-dot-binge-drinking-prevention-dot-o-r-g or call 1-800-300-2222. A public service message
brought to you by the North Georgia Anti-Addiction Alliance.

View publication stats

You might also like