You are on page 1of 30

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Gen van Heerden is Assistant Director Specialised Degree Programmes, Rotterdam School of Management, Erasmus university, Rotterdam, The Netherlands. She recently completed her PhD degree in Marketing at the Lule university of Technology, in Lule, Sweden. Leyland F. Pitt is Professor of Marketing and the Dennis F. Culver EMBA Alumni Chair of Business at the Segal Graduate School of Business at Simon Fraser university, Vancouver, Canada, and is also a Senior Research Fellow of the Leeds university Business School in the united Kingdom. Lydia Windisch is currently completing her doctoral thesis in psychology and is a Research Fellow in the Department of Marketing at Monash university. Address correspondence to Peter Steyn, Division of Industrial Marketing, eCommerce and Logistics, Lule university of Technology, Lule, Sweden. Email: peter.steyn@ltu.se

160

The persuasive impact of YouTube public service announcement producers


Hye-Jin Paek, Thomas Hove, Hyun Ju Jeong
Michigan State University

Peer or expert?

Mikyoung Kim
Ewha Womans University

To promote prosocial concerns and call attention to social problems, public service advertising practitioners are increasingly trying to involve laypeople in creating and delivering persuasive campaign messages. An emerging media channel for these efforts is websites that feature user-generated content (uGC), particularly the video-sharing website YouTube. However, despite this trend, little is known about the extent to which a public service announcement (PSA) video will be more effective depending on who produced it. Accordingly, this study empirically tests the degree to which the persuasive impact of a video differs depending on whether the producer is a layperson or an expert. We draw theoretical rationales from several areas to compare the impact of a perceivably similar producer and an expert producer on attitudes towards video, issue importance and behavioural intention. We also analyse how issue involvement moderates these producer effects. Implications for consumer educators, policy makers and marketers are discussed with specific reference to social media.

Introduction
The increasing popularity of websites that feature user-generated content (uGC) has triggered a new phase in public service advertising campaigns (Croft 2008; Todi 2008; Vance et al. 2009). One website that has rapidly become a key hub for sharing user-generated videos is YouTube (Loechner 2008). Researchers who study new media have noted that YouTube and other uGC websites are creating new viewing patterns, social interactions and communication dynamics between message producers and receivers (Jenkins 2006; Bian et al. 2008; Shirky 2008).
International Journal of Advertising, 30(1), pp. 161188 2011 Advertising Association Published by Warc, www.warc.com  DOI: 10.2501/IJA-30-1-161-188

161

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

By facilitating Internet users engagement in the creation and delivery of prosocial messages, uGC websites can strengthen professionals efforts to promote social causes and issues among key target audiences. One strategy is to supplement existing grassroots concern by hosting viral video contests and providing opportunities for laypeople to create cause-related messages for the general public. For example, as part of an effort encouraging Americans to actively protect themselves from the H1N1 flu virus, the uS Department of Health and Human Services (DHHS) recently sponsored a public service announcement (PSA) contest on YouTube (DHHS 2009). Informing this plan was the assumption that a PSA video created by someone from a perceivably similar group will be more effective than one produced by an expert or an organisation. However, despite the increasing trend to involve individual Internet users in creating and delivering persuasive campaign messages, little is known about the extent to which a PSA video will be more effective depending on who produced it. The purpose of the current study is to empirically test how the persuasive impact of a public service YouTube video differs depending on whether the producer is a layperson peer or an expert professional. Because the impact of media message producers on persuasion has not been extensively investigated, we draw theoretical rationales from several areas of psychological and persuasion research: social influence, social comparison theory and source characteristics (Moscovici 1985; McGuire 2001; Perloff 2003; Paek & Gunther 2007). Based on these rationales, we predict that a perceivably similar peer producer will have a relatively stronger impact than an expert producer on attitudes towards video, issue importance and behavioural intention. In addition, we expect that issue involvement will moderate the producer effect, based on several theoretical rationales such as the Elaboration Likelihood Model (Petty & Cacioppo 1986a, 1986b; Wilson & Sherrell 1993) and exemplification theory (Zillmann & Brosius 2000; Andsager et al. 2006). Relying on the insights of these theories and models, we expect that PSA videos that are perceived to be produced by peer exemplars will be more likely to cue a positive response in a lay audience. This experimental study was conducted among college students who viewed three types of child abuse prevention PSA videos on YouTube. Its findings provide both practical and theoretical implications for public service advertising practitioners and researchers, particularly issue advocates

162

PEER OR EXPERT?

who want to explore how video sharing and other types of uGC website can be used to increase the engagement of individual laypeople in their causes.

UGC websites as a new PSA channel


YouTube is a free website service that provides individual Internet users with easy opportunities to upload and share their original videos (Freeman & Chapman 2008). Often designated as a uGC website (Daugherty et al. 2008), YouTube draws approximately 20% of all HTTP traffic, which is nearly 10% of all traffic on the Internet (Cheng et al. 2007). Among Internet users aged 1829, 89% report that they view content on videosharing sites, and 36% do so on a typical day (Pew Internet & American Life Project 2009). YouTube has already been used by a variety of business entities, from entertainment companies to charitable organisations (Mediaco 2007). Among the videos now commonly available on the website are public service announcements (PSA) produced by non-profit organisations and governments. YouTube has increasingly served as a channel for disseminating PSAs ever since the website committed to the 2007/2008 Clinton Global Initiative, which used YouTube as an official outlet for delivering public service messages (YouTube 2007). Considering the growing popularity of uGC websites, our main purpose is to empirically demonstrate the degree to which an online PSA videos persuasive impact differs depending on whether the producer is a layperson or an expert. Two characteristics that distinguish these types of producer from one another are perceived similarity and expertise. In other words, a peer producer will be perceived as more similar to Internet users themselves, while an organisation or professional campaigner may be perceived as an expert on the PSAs featured issue. Noting that these characteristics of perceived similarity and perceived expertise have been most prominent in relation to source characteristics (Perloff 2003), the current study focuses on the perceived similarity or expertise that viewers attribute to the producers of YouTube videos.

163

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Producer effects: perceived similarity versus perceived expertise


In research on message source effects, perceived similarity and perceived expertise have been acknowledged as important characteristics that explain why certain messengers or human portrayals tend to be more effective than others (e.g. source effects; Ohanian 1990, 1991; McGuire 2001; Andsager et al. 2006). Perceived similarity refers to the extent to which individuals perceive a portrayal as realistically reflecting their own experiences (Austin & Meili 1994; Andsager et al. 2006), or as similar to themselves based on certain attributes. These attributes may include demographics, shared values, shared ideas and common experiences (Salmon & Atkin 2003). Several theoretical arguments outline how perceived similarity can lead to persuasion. First, regarding source similarity, the source-attractiveness model suggests that receivers identify more closely with sources who are similar to themselves (Kelman 1961). In turn, this identification process guides people in evaluating the messages that such sources deliver. Second, the social influence process and its underlying mechanisms (e.g. conformity) suggest that people, especially youths, tend to think and behave based on the ways others think and behave (for an overview, see Moscovici 1985). Third, according to reference group theory (Sherif & Sherif 1964; Hyman & Singer 1968), one of the important attributes that leads people to view others as belonging to their reference group is perceived similarity. Young people in particular are more likely to perceive people in their own age cohort as their significant reference group, and in turn this perception is more likely to influence their attitudes and behaviours. Perceived expertise is the degree to which message receivers perceive a communication to convey relevant knowledge, skill, or experience (Hovland et al. 1953; McGinnies & Ward 1980; Perloff 2003). Along with credibility and trustworthiness, perceived expertise refers generally to a communicator or messages positive characteristics. It generates a more favourable evaluation of the messages, and it influences the receivers beliefs, attitudes or behaviours (Haas 1981; Ohanian 1990; Perse et al. 1996). Few studies have compared whether perceived similarity or perceived expertise will have relatively stronger effects on persuasion. Instead,

164

PEER OR EXPERT?

researchers have stressed the persuasive impact of perceived similarity and perceived expertise as two important kinds of source characteristics. For example, in source effects research, sources whom audiences perceive as similar to themselves are more likely to produce persuasion than sources whom audiences perceive as dissimilar (Perloff 2003). In other cases, people tend to respond more favourably to experts rather than nonexperts because they view them as sources who are either higher in quality (Gilly et al. 1998) or more authoritative (Cialdini 2001; Perloff 2003). Although perceived expertise may have an impact, several theoretical rationales suggest that a video producers perceived similarity (e.g. membership in ones own age cohort) will play a greater role in college students evaluation of a YouTube video. First, based on the premises of social influence and social comparison, college students may form more favourable attitudes towards a video that has been produced by people from their own age cohort, as opposed to experts who work for child welfare organisations. In accordance with this reasoning, Ahn et al. (2009) tested a radio PSA on college students heavy drinking to compare the relative impact of peer sources (i.e. similarity) vs doctor sources (i.e. expertise) and their interaction with message appeals (i.e. cognitive vs affective). They found that the message with the peer source was more effective in terms of positive attitudes towards the PSA. When the message included peer source coupled with affective appeals, it was effective in terms of attitudes towards the PSA (more favourable), heavy drinking (less favourable) and intention not to drink heavily. The second rationale informing the current study is related to producer effects with respect to the nature of YouTube. Because YouTube is one of the largest uGC channels on which laypeople, or people like me, can create, upload and discuss available videos, one can reasonably assume that persuasion will be enhanced when the audience perceives a matchup between the producer and the channel. According to social adaptation theory, when a persuasive message provides an adaptive informational cue that matches peoples expectations, they will accept the message, and persuasion will be more likely to occur (Kahle & Homer 1985). To apply this argument to the current study, a video produced by a peer rather than an expert will be more likely to match what individuals expect to see on a uGC website such as YouTube, and therefore more likely to get their acceptance. In addition, because people may associate YouTube with

165

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

videos produced by people who are similar to themselves rather than by issue experts, they are more likely to respond positively to peer-produced videos delivered via uGC websites. Thus, we formulate the following main effect hypothesis of PSA producers to predict three major advertising effect variables: attitudes towards the video, issue importance (equivalent to attitudes towards the brand in commercial advertising contexts) and behavioural intention: H1: Individuals will respond more positively to child abuse prevention PSAs produced by a peer than by an expert in terms of (a) attitudes towards the video, (b) issue importance and (c) behavioural intention

Role of involvement in producer effects


Although it is expected that PSA videos produced by peers lead to persuasion more effectively than those produced by experts, such producer effects may not be uniform from one audience member to the next. An important moderating variable that might differentiate producer effects is involvement. While multiple definitions and types of involvement exist across various disciplines, issue involvement seems most pertinent to the current context. Issue involvement refers to personal relevance, the extent to which the attitudinal issue under consideration is of personal importance (Petty & Cacioppo 1990, p. 368). Several theories discuss the role of involvement in information processing and persuasion. In particular, the Elaboration Likelihood Model (ELM) has been used widely to explain how source characteristics such as similarity, credibility and expertise may affect peoples information processing and subsequent attitude change. According to the ELM, there are two routes to persuasion: the central and the peripheral. The central route involves thoughtful scrutiny of a messages claims, and people follow the route when they have the cognitive ability or motivation to process those claims. One such motivation would be issue involvement (Petty et al. 2008). By contrast, the peripheral route involves taking cognitive shortcuts (Petty & Cacioppo 1986a, 1986b). In conditions of low involvement, peripheral cues that lead to persuasion include source characteristics such as similarity, liking or cred-

166

PEER OR EXPERT?

ibility. As Wilson and Sherrell (1993) point out, seven out of the eight studies that examined source effects by involvement reported that effects such as credibility were more pronounced under low-involvement conditions. Similarly, in an experiment comparing the role of message argument versus source characteristics in persuasion, low-involvement participants changed their opinion based on a likeable communicator rather than message arguments, while the opposite was the case for high-involvement participants (Chaiken 1980). According to this reasoning, producer effects from perceived similarity may be more pronounced when people have low issue involvement. In addition, people often do not subject media messages to extensive cognitive processing (Harris 1987; Heath 2000). Instead, they rely on available information (e.g. exemplification theory; Zillmann & Brosius 2000). This means that, when people exposed to a child abuse prevention PSA have low involvement in the issue, they will evaluate the video according to the relevant and available information that it was produced by someone similar to themselves. By contrast, when people have high involvement in the issue, they will tend to evaluate the PSA more carefully using all the information available with respect to message quality, argument strength, video production quality, and so on. In such circumstances, the fact that the video is produced either by someone similar to themselves or by an issue expert will not affect how they evaluate and respond to the video. Informed by these rationales, we formulate the following moderation hypothesis: H2: The effects of a perceivably similar producer will be more pronounced among low-involved individuals than among highinvolved individuals in terms of (a) attitudes towards the video, (b) issue importance and (c) behavioural intention.

Methods
Study design This study employs a 2 (producer type: peer vs expert) 2 (issue involvement: low vs high) 3 (types of child abuse prevention PSAs: fear PSA, emotional PSA and informational PSA) mixed factorial design. The pro-

167

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

ducer type was manipulated as a between-subjects factor. The PSA type was treated as a within-subject factor. Issue involvement was measured to serve as a moderator. Stimuli Producer type For the two types of PSA producer, we identified average college student (peer) and state-level non-profit organisation (expert). Regarding the peer type, college students seem highly likely to be producers of YouTube messages in that 31.5% of youths aged 18 to 24 participate in social media websites (Ellazar et al. 2009). Regarding the expert type, non-profit organisations are known to be official producers of YouTube PSA messages. We featured these two types of PSA producer in the introduction to a paper-and-pencil questionnaire with the following written statements: In this section, Id like you to watch and evaluate three different YouTube clips on child abuse prevention that were created and uploaded by a college student (peer) or by Michigan Chapter of Prevent Child Abuse America (expert). PSA type We selected and modified three child abuse prevention PSAs originally found on YouTube. These PSAs were selected based on the following criteria: (1) comparability of length i.e. about one minute, similar to typical PSAs; (2) different formats and appeals for testing whether our hypotheses are supported across various PSA types; (3) focus on the child abuse issue in general, rather than on specific child abuse themes such as sexual abuse, verbal abuse and physical abuse; and (4) absence of information about a specific locality (e.g. a child abuse call centre in Texas). The selected videos were further modified to make them similar in terms of length and call-to-action message. First, the fear PSA shows an acted-out scene of an adult male having dinner with his mother and father. It employs a fear appeal when the mother and father abruptly begin to beat him, and concludes with a caption that asks the audience how they would feel if the victim were an actual child. Second, the emotional PSA shows graphic images of children with bruised faces, black eyes, and other visible signs of injury, pain and sadness, intercut with caption screens that

168

PEER OR EXPERT?

appeal to emotion (e.g. No one to speak for them). Finally, the informational PSA shows images of happy children, intercut with caption screens that indicate the prevalence of the child abuse issue, usually through statistical information (e.g. Four children die from abuse every day). We designated these three PSAs with the fear, emotional and informational labels based on the results of participants evaluations. That is, participants were asked to judge whether the videos were rational (3)/emotional (3) and whether the videos relied on facts (3)/on images (3) (Liu & Stout 1987; Perse et al. 1996). Exploratory factor analysis (EFA) indicated a one-factor solution for the fear PSA (84% of total variance explained, inter-item correlation = 0.67), for the emotional PSA (85% of total variance explained, inter-item correlation = 0.93), and for the informational PSA (78% of total variance explained, inter-item correlation = 0.55). After these two items were averaged, a repeated-measures ANOVA was conducted. The results showed a significant mean difference among the three PSAs [ = 0.97, F (2, 227) = 81.20, p < 0.01]. A pairwise comparison test via the Bonferroni method revealed that the emotional PSA was evaluated to be more emotional than the informational PSA [Mean diff. = 0.33, p < 0.05]. In addition, a repeated-measures ANOVA test indicated significant mean differences among the three PSAs [Mean diff. = 0.59, F (2, 230) = 81.20, p < 0.001] in terms of the participants self-reported fear. A pairwise comparison test via the Bonferroni method reported that the fear PSA was rated as more fear-inducing than both the emotional PSA [Mean diff. = 0.97, p < 0.001] and the informational PSA [Mean diff. = 1.51, p < 0.001]. Data and samples For extra credit, 332 undergraduate students enrolled in various communication courses at a major Midwestern university took part in a study entitled Persuasive Message Evaluation Study. This sample is appropriate for two reasons. First, although parents with young children may be more concerned about the child abuse issue, college students are also a potential target audience because many of them will become parents in the near future. Second, professional child abuse prevention campaigns for example, the Ad Councils Trust your instincts target the public in general, which of course includes college students. For the sample characteristics,

169

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

females (68.5%) outnumbered males, and a majority of students (98.7%) were aged between 18 and 24. Procedures In the classrooms where surveys were administered, participants were told that they were participating in a message evaluation study. They filled out a paper-and-pencil questionnaire about their demographics (e.g. age, gender) and their issue involvement for child abuse. Then they watched the three PSAs accessed on YouTube and projected on the lecture hall screen. To control for any possible order effect on the study outcomes, the videos were played in different orders for each class. To check systematic differences across all classes, two of the classes watched the videos in the same order. After watching each PSA, participants were asked to evaluate it and to complete a series of questions regarding their attitudinal and behavioural intention responses. Participants across all classes were randomly assigned to one of the written messages indicating the producer type. Measures Issue involvement Participants rated their agreement about personal interest in the child abuse issue (e.g. I have thought about child abuse issue quite seriously, I really dont care about the issue of child abuse [reverse-coded], Child abuse is a serious social problem and I have paid close attention to child abuse issues) on a seven-point Likert scale ranging from strongly disagree (1) to strongly agree (7). Exploratory factor analysis (EFA) with principal axis factoring (PAF) extraction method showed one factor (55% of total variance). Cronbachs alpha reliability also indicated reasonable internal consistency (a = 0.73). After averaging all four items to construct the variable of issue involvement for child abuse, we classified it into low and high involvement, following past studies (Gangestad & Snyder 1985; Snyder & DeBono 1985). Seventy-four participants who scored at the midpoints (i.e. 5.0 to 5.4) of personal involvement (M = 5.13, SD = 0.98, median = 5.25) were removed for sub-analysis (for similar procedures, see Shavitt et al. 1992). Methodologically, using two extreme groups can also prevent the

170

PEER OR EXPERT?

problem of misclassifying a participant whose score is close to the median as low or high (Paek et al. 2010). Attitudes towards video (Av ) Attitudes towards PSA video were measured using the following four items drawn from previous studies measures of attitude towards advertisement (MacKenzie et al. 1986; Homer 1990): bad/good, dislike/ like, uninteresting/interesting and unfavourable/favourable (seven-point semantic differential scale ranging from 3 to 3). The EFA result indicated a one-factor solution for the three types of PSAs: for the fear PSA, 79% of total variance explained, = 0.91; for the emotional PSA, 81% of total variance explained, = 0.92; and for the informational PSA, 80% of total variance explained, = 0.91. The averaging index was used for subsequent analyses. Issue importance (Ii ) This variable was measured with four items rated on a seven-point semantic differential scale: trivial/serious, unimportant/important, not worth much concern/worth a lot of concern, and irrelevant/relevant (cf. Dillard & Peck 2000). After checking the EFA (85% for the fear PSA; 89% for the emotional PSA; 83% for the informational PSA) and Cronbachs alpha (0.94, 0.96 and 0.93, respectively), the four items were averaged to construct the variable. Behavioural intention (Bi ) Participants were asked to answer five questions with a seven-point Likert scale (1 = not at all likely to 7 = very likely). These questions were modified from past studies on prosocial campaigns (e.g. Beaudoin & Thorson 2007) to fit the child abuse context (e.g. Bagozzi & Moore 1994): After viewing this video, how likely are you ? (a) to give help to abused children? (b) to seek more information about the child abuse issue? (c) to share the video with your friends and family? (d) to call the number shown in the video? (e) to report someone whom you suspect of child abuse? EFA results showed that the five items clearly constructed one factor for the three types of PSA: for the fear PSA, 68% of total variance explained, = 0.88; for the emotional PSA, 68% of total variance explained, = 0.88; and for the informational PSA, 66% of total variance explained, = 0.87.

171

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Thus, the five items were averaged to create an index of behavioural intention. Confound check: class effects and order effects First, because a systematic difference between classes may arise, we checked whether any class effect occurred. Two out of four classes were shown the PSAs in the same order. An independent samples t-test indicated no class effects in terms of mean differences of the three dependent variables Av, Ii and Bi. Next, we checked whether any order effect was present for showing either of the three PSAs first. Order may affect responses to the dependent variables due to participants fatigue and/or familiarity with the topic and corresponding questions (see Mook 1982; Lucas 1992). A repeatedmeasures ANOVA indicated significant order effects for the fear PSA and the informational PSA on Av and Bi. For this reason, we controlled the order of PSA viewing in hypothesis testing. That is, the order was coded as two dummy variables: the order of fear emotional informational PSAs (FEI order) = 1 and the order of emotional informational fear PSAs (EIF order) = 1. Finally, we controlled gender as a possible covariate, based on previous findings that gender was related to the following: the behaviour of helping others (Shrum et al. 1995; Perse et al. 1996); the response to online persuasive messages (Guadagno & Cialdini 2002); and the attitude and intention to use new media (Choi et al. 2009). Gender was coded as female = 1 and male = 0.

Results
Manipulation check To secure the producer type manipulation of the PSA videos, participants completed several manipulation checks. First, in a screening question on PSA producers, participants were asked if they knew who produced the PSA video. Out of 332 students, 235 (71%) supplied correct answers (n = 119 for the peer condition; n = 116 for the expert condition); thus, those who did not answer correctly were removed from the final analyses.

172

PEER OR EXPERT?

Second, to verify the manipulation of PSA video producer characteristics, participants rated their perceived similarity and perceived expertise towards the PSA producer respectively. For perceived similarity, participants were asked directly whether the video producers were not similar to you (3)/similar to you (3) and someone not likely to be your peer (3)/someone likely to be your peer (3) (Feick & Higie 1992). For perceived expertise, participants were asked whether the video producers were not credible (3)/credible (3); not an expert (3)/expert on the issue (3); not trustworthy (3)/trustworthy (3); amateurish (3)/professional (3) (Dholakia & Sternthal 1977; Hass 1981; Ohanian 1990). EFA indicated a one-factor solution for both the peer producer (78% of total variance explained; inter-item correlation = 0.52) and the expert producer (76% of total variance explained; = 0.88). For subsequent analyses, these items were averaged. Two sets of independent samples t-tests indicated that participants perceived the peer producer to be more similar [Mean diff. = 1.08, t (227) = 6.71, p < 0.001] and to have less expertise than the expert producer [Mean diff. = 0.68, t (227) = 4.70, p < 0.001]. Thus, the manipulation check was successful. Hypothesis testing For testing H1 and H2, a multivariate analysis of covariance (MANCOVA) is the most appropriate procedure because there were multiple dependent variables (Av, Ii and Bi), two factors (producer type and issue involvement) and three covariates (gender and two dummy variables for order). However, our examination of MANCOVA assumptions reveals that homogeneity of variance was found not to exist for the fear and emotional PSAs (Boxs M = 70.53 and 100.87, F = 3.94 and 5.96, both at p < 0.001). The result indicates a failure to meet assumptions needed to perform a MANCOVA (for details regarding the assumptions and testing, see Hair et al. 1998). Thus, three sets of ANCOVA for the three dependent variables were performed across the three PSA types (nine sets in total). Tables 13 report ANCOVA results for each of the three dependent variables across the three PSA types.

173

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Table 1: Summary of ANCOVAs for the three dependent variables (fear PSA)
Factor Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) Order 2 (covariate) Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4 3

M1

SE2

F 1.045

df 1, 171

p 0.308

Dependent variable 1: Attitude towards video (Av.) 0.781 1.028 0.862 0.947 0.172 0.169 0.107 0.185 0.168 1.668 0.934 9.186 0.123 9.033 2.211 2.582 2.110 2.682 0.088 0.086 18.756 0.094 0.086 2.516 9.899 0.123 3.055 0.083 4.217 4.279 3.842 4.654 0.154 0.151 12.378 0.165 0.150 0.020 5.061 9.048 1.627 1, 171 1, 171 1, 171 1, 171 0.889 0.026 0.003 0.204 1, 171 0.001 1, 171 1, 171 1, 171 1, 171 1, 171 0.115 0.002 0.727 0.082 0.774 1, 171 <0.001 1, 171 1, 171 1, 171 1, 171 1, 171 0.198 0.335 0.003 0.726 0.003 1, 171 0.744

Order 1 (covariate) 4
5

Dependent variable 2: Issue importance (Ii)

Order 2 (covariate) 5 Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4

Dependent variable 3: Behaviour intention (Bi)

Order 2 (covariate) 5
Notes: 1 M = marginal mean (after controlling for all the covariates) 2 SE = standard error 3 Sex: female = 1 and male = 0 4 Order 1: the order of fear emotional informational PSAs (FEI order) = 1 5 Order 2: the order of emotional informational fear PSAs (EIF order) = 1.

174

PEER OR EXPERT?

Table 2: Summary of ANCOVAs for the three dependent variables (emotional PSA)
Factor Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) Order 2 (covariate) Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4 3

M1

SE2

F 9.817

df 1, 172

p 0.002

Dependent variable 1: Attitude towards video (Av.) 0.616 1.303 0.772 1.198 0.154 0.159 4.122 0.168 0.152 1.959 2.001 0.040 0.445 10.446 2.177 2.547 1.985 2.739 0.081 0.081 37.863 0.088 0.079 8.968 9.155 0.044 0.108 0.855 4.171 4.357 3.761 4.767 0.143 0.142 21.761 0.155 0.139 1.797 3.406 3.368 0.564 1, 173 1, 173 1, 173 1, 173 0.182 0.067 0.068 0.454 1, 173 <0.001 1, 174 1, 174 1, 174 1, 174 1, 173 0.003 0.003 0.834 0.743 0.356 1, 174 <0.001 1, 172 1, 172 1, 172 1, 172 1, 174 0.163 0.159 0.842 0.506 0.001 1, 172 0.044

Order 1 (covariate) 4
5

Dependent variable 2: Issue importance (Ii)

Order 2 (covariate) 5 Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4

Dependent variable 3: Behaviour intention (Bi)

Order 2 (covariate) 5
Notes: 1 M = marginal mean (after controlling for all the covariates) 2 SE = standard error 3 Sex: female = 1 and male = 0 4 Order 1: the order of fear emotional informational PSAs (FEI order) = 1 5 Order 2: the order of emotional informational fear PSAs (EIF order) = 1

175

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Table 3: Summary of ANCOVAs for the three dependent variables (informational PSA)
Factor Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) Order 2 (covariate) Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4 3

M1

SE2

F 3.247

df 1, 170

p 0.073

Dependent variable 1: Attitude towards video (Av.) 0.870 1.230 0.793 1.307 0.140 0.141 5.741 0.153 0.139 0.032 0.473 7.372 6.375 4.968 2.128 2.422 1.914 2.636 0.093 0.093 26.200 0.101 0.091 3.352 4.440 4.423 3.777 0.012 3.979 3.957 3.413 4.523 0.137 0.137 28.580 0.149 0.134 0.232 1.618 0.689 4.260 1, 173 1, 173 1, 173 1, 173 0.631 0.205 0.408 0.041 1, 173 <0.001 1, 173 1, 173 1, 173 1, 173 1, 173 0.069 0.037 0.037 0.054 0.911 1, 173 <0.001 1, 170 1, 170 1, 170 1, 170 1, 173 0.859 0.493 0.007 0.012 0.027 1, 170 0.018

Order 1 (covariate) 4
5

Dependent variable 2: Issue importance (Ii)

Order 2 (covariate) 5 Producer type Expert Peer Involvement Low High Producer type involvement Gender (covariate) 3 Order 1 (covariate)
4

Dependent variable 3: Behaviour intention (Bi)

Order 2 (covariate) 5
Notes: 1 M = marginal mean (after controlling for all the covariates) 2 SE = standard error 3 Sex: female = 1 and male = 0 4 Order 1: the order of fear emotional informational PSAs (FEI order) = 1 5 Order 2: the order of emotional informational fear PSAs (EIF order) = 1

176

PEER OR EXPERT?

H1: Main effects of PSA producer peer vs expert The first set of hypotheses predicted that a child abuse prevention PSA created by a peer producer would be more effective than one created by an expert producer in terms of attitudes towards video (H1a), issue importance (H1b) and behavioural intention (H1c). For H1a, a significant main effect of the producer type was found on attitudes towards video for the emotional PSA, after controlling for order and gender. A pairwise comparison test with the Bonferroni method indicated that participants had more favourable attitudes towards the PSA created by peer than that by expert [ = 0.69, p < 0.01]. But, for the fear PSA, there was no significant main effect of producer type on attitudes towards video. For the informational PSA, the main effect of producer type on attitudes towards video was marginally significant at p = 0.07 in the expected direction. Thus, H1a was partially supported. For H1b, the main effect of the producer type appeared significant on issue importance for all three PSAs. A pairwise comparison test with the Bonferroni method indicated that participants reported a higher level of issue importance when viewing the peer-produced PSA than when viewing the expert-produced PSA [ = 0.37 for the fear PSA, p < 0.01; = 0.37 for the emotional PSA, p < 0.01; = 0.29 for the informational PSA, p < 0.05]. Thus, H1b was supported. For H1c, there were no significant main effects of the producer type on behavioural intention across all three PSAs. Thus, H1c was not supported. H2: Moderation effects of involvement The second set of hypotheses predicted that the peer producer would be more effective than the expert producer on attitudes towards video (H2a), issue importance (H2b) and behavioural intention (H2c) for low-involved participants than for high-involved participants. Among these sub-sets of hypotheses, a significant interaction effect of producer type and involvement was found on issue importance only for the emotional PSA. As shown in Figure 1, when the participants were low-involved, their level of issue importance was significantly higher in the peer producer condition than in the expert producer condition, after controlling for the covariates. By contrast, when the participants were high-involved, their mean level of issue importance was similar between peer and expert condition. Thus, H2b was partially supported, while H2a and H2c were not supported.

177

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Figure 1: Interaction effect of producer type and involvement on issue importance (emotional PSA)
4 High involvement Low involvement 3 Issue importance 2.726 2.752 2.341 2 1.628 1

Expert

Peer

Note: Means indicate marginal means (after controlling for all the covariates)

Effects of covariates Although not hypothesised in the current study, some significant effects of the covariates, video order and gender, are worth reporting. To begin with, viewing a video first may affect the findings in an unfavourable way. For example, when participants viewed the fear video first, they tended to report their attitudes towards video (B = 0.87, se = 0.29, t = 3.03, p < 0.01) and behavioural intention (B = 0.77, se = 0.26, t = 3.01, p < 0.01) in less favourable ways compared to when they viewed it later. By contrast, when they viewed the informational PSA second or third, they reported more favourable attitudes towards video (FEI order: B = 0.64, se = 0.23, t = 2.72, p < 0.01; EIF order: B = 0.63, se = 0.25, t = 2.53, p < 0.05), issue importance (FEI order: B= 0.33, se= 0.16, t = 2.10, p < 0.05; EIF order: B= 0.32, se = 0.16, t = 1.94, p = 0.05) and behavioural intention (EIF order: B= 0.50, se = 0.24, t = 2.06, p < 0.05), compared to when they viewed the PSA first. On the other hand, compared to males, females reported lower issue importance for the fear PSA (B = 0.46, se = 0.15, t = 3.15, p < 0.01), the informational PSA (B = 0.33, se = 0.16, t = 2.11, p < 0.05) and the emo-

178

PEER OR EXPERT?

tional PSA (B = 0.41, se = 0.14, t = 3.03, p < 0.01). Females also showed lower behavioural intention than males for the fear PSA (B = 0.57, se = 0.25, t = 2.25, p < 0.05).

Discussion
Public service and cause-related advertising campaigns are increasingly inviting members of the general public to produce their own PSAs on prosocial and health issues, and to share them on uGC websites such as YouTube. But, despite this trend, little is known about how the type of video producer a similar peer or an issue expert affects viewers attitudes towards video, issue importance and behavioural intention. This study was designed to examine these producer effects on those three outcome variables, along with the moderating effects of issue involvement. Our main findings are twofold. First, compared to child abuse prevention PSA videos produced by an issue expert, those produced by a perceivably similar peer were more effective in the enhancement of attitudes towards the PSA and issue importance. Second, the effect of a perceivably similar peer producer on issue importance was particularly more pronounced among low-involved than among high-involved viewers. The first finding coincides with existing research indicating that perceived similarity seems to be a more influential attribute than perceived expertise, at least among young people in the context of prosocial messages (e.g. Andsager et al. 2006; Ahn et al. 2009). Source effects literature has documented that message receivers can identify more closely with sources who are similar to themselves (Kelman 1961). According to social influence literature, young individuals may want to respond more favourably to messages produced by someone like themselves (Hyman & Singer 1968; Paek & Gunther 2007). On uGC websites such as YouTube, audiences probably expect to encounter videos produced by laypeople similar to themselves, and not by professionals and organisations who promote specific social causes and issues. As for the second finding, we found that the effect of a perceivably similar producer is more pronounced when viewers have low rather than high issue involvement. This finding is consistent with existing literature on attitude change, which indicates that low-involved audiences are more influenced by source characteristics (e.g. attractiveness), while high-

179

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

involved audiences are more influenced by message quality and argument strength (Chaiken 1980; Petty & Cacioppo 1986a, 1986b). In addition, the YouTube platform itself often provides noticeable cues about whether the video is created by someone who might be a peer or by an organisation with issue expertise. These cues can enable individual users to quickly determine how much they like the message and how much they regard its featured issue as salient (Zillmann & Brosius 2000; Andsager et al. 2006). Additional evidence supporting this argument is that, when high-involved viewers evaluated the three PSA videos, they were not affected by the type of producer. Finally, our findings showed that there were neither main effects of producer types on behavioural intention nor interaction effects between producer types and issue involvement on behavioural intention. Such insignificant findings may indicate the relative difficulty for ads to affect behavioural intention compared to other advertising effects such as attitudes towards ad or brand. According to the traditional hierarchy of advertising effects, ads that aim to achieve a behavioural goal should produce awareness, then interest, then evaluation, and finally conviction. Furthermore, a well-established advertising effects model also posits the mediation relationship among attitude towards ad, attitude towards brand and behavioural intention (Mackenzie et al. 1986). Although the mediating mechanism among these outcome variables was not the main focus of this study, these existing advertising effects models provide plausible explanations for why we found null findings on behavioural intention. For public service campaigns, these findings have several practical implications. Most obviously, federal governments and non-profit organisations would be wise to encourage individual participation in causerelated campaigns through viral video contests and other techniques. Such efforts may be more effective at reaching young audiences and motivating them about prosocial concerns. Viral marketing has become widely used in the commercial sector to engage audiences, particularly youths. However, it has only begun to be used in the non-profit sector. A recent Ad Council report addresses the significance of social media and uGC websites for delivering PSAs to American youths aged 18 to 24, especially to motivate their involvement in social issues (Ellazar et al. 2009). The report highlights the potential for using such websites as a channel for PSAs, based on the descriptive

180

PEER OR EXPERT?

finding that young people usually obtain a majority of their information pertaining to social issues from online resources (51.7%) and from television (53.1%). It also indicates that young people are more likely to respond to PSAs that they encounter online. Furthermore, engaging audiences in content delivered through social media could prove to be an efficient and money-saving approach for non-profit organisations because it can save production and placement costs. At the same time, however, viral videos disseminated through social media might create ethical problems, especially when professionals use certain formats to simulate the look of user-generated web content and to piggy-back their public service announcements on the private channels of personal expression. For example, in April and May 2008, the Canadian company NB Liquor uploaded to YouTube a series of videos that comprise a public service campaign about the dangers of drunk driving. The videos simulate a series of non-fictional vlogs, or video diaries, of a teen named Paul underhill who killed his friends by driving under the influence. The campaign also includes several video responses to Pauls vlogs, purportedly from the peers at his high school who lost friends in the accident. Although these videos are part of a corporatesponsored and professionally produced public service campaign, they continue to appear on YouTube as if they had been uploaded by the teens who appear in the videos. As can be seen in the comment threads that accompany these videos, such ethically questionable practices may backfire and generate audience distrust in prosocial campaigns. In turn, this distrust may eventually lead to long-term problems such as reactance against prosocial messages or indifference towards the causes they promote. Practitioners should therefore be cautious about using such potentially deceptive practices. Or, at least, they should avoid disguising professional producers and sources as laypeople and fictionally contrived scenarios as actually occurring events. At the same time, to guarantee adequate protection from such practices, policy makers and consumer educators should keep abreast of the new media formats in which persuasive messages can be delivered. Other implications concern reaching audiences in the different temporal stages of what Downs (1972) called the issue attention cycle. To stimulate public awareness, practitioners should use user-created videos

181

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

in the early stages of publicising a social cause. During this stage, people know little about the cause and thus have not had a chance to think seriously about it. However, over the long term, attitude changes that occur only through peripheral rather than central processing may not become adequately entrenched. Because persuasion that occurs in lowinvolvement conditions is fragile and susceptible to attenuation, producer effects that are more salient among low-involved audiences may not have durable effects. This outcome raises another double-edged implication. While public service campaigners should try to involve and collaborate with the general public, they should also continue to pursue the long-term strategy of developing a consistent and effective message, and sticking to it. Other practical implications arise from our findings on order effects. A general pattern we found was that, if a PSA is shown first, its effects were lower than those of PSAs shown later. One possible reason is that, because the child abuse prevention message is prosocial, participants may want to seem more socially desirable by evaluating the PSAs higher the more they see them. Another possible reason is that, because most of the participants may not be familiar with the child abuse issue, they may become more emotionally engaged the more the PSAs are shown. Our finding indicating that the emotional PSA seems less vulnerable to order effects may also support this argument. If this is the case, one practical implication is that audiences should be exposed to child abuse PSAs for a relatively longer time and on multiple occasions rather than in one-shot instances. Such extended patterns of exposure may make audiences more emotionally engaged and involved in the issue. Given that most child abuse PSAs have been one-shot affairs, with different messages emphasised each time, a sustained campaign with consistent messages may be needed for more synergistic effects. Despite the novelty of this study and its findings, four limitations should be noted. First, although we tested three representative child abuse prevention PSA appeals to demonstrate robust findings, the representativeness of these findings could be strengthened by testing PSAs on multiple topics. But since many PSAs dealing with various prosocial topics are available on YouTube, our findings should be easily replicable in other study contexts. Second, even though we justified the appropriateness of using college students for this study, we recognise

182

PEER OR EXPERT?

that issues still remain with respect to the potential generalisability of our findings to other demographic groups. Future research should replicate this study using different demographic groups who are likely to be exposed to PSAs on social media. Finally, although our manipulation check was successful, it should be noted that the way we manipulated producer type by indicating the type of producer in the survey instrument rather than in the video could be more effective. This weak highlighting of the video producer may explain why almost 30% of the participants failed to correctly identify producer type. Future research should replicate this finding with more rigorous manipulation and in various study contexts. Scholars who study public communication and advertising have noted problems that are likely to arise when cause-related campaigns do not involve collaboration between practitioners and their target audiences (e.g. Guttman & Salmon 2004). In such cases, public communication campaigns may be viewed as strategies of social control because one group has taken it on itself to affect the beliefs or behaviours of another group (Dozier et al. 2001, p. 232). Public service advertising campaigners therefore need to remain sensitive to the ethical pitfalls that can occur as side effects of using new media channels. Nevertheless, if properly used, uGC websites such as YouTube may provide an alternative channel for engaging target audiences, and enabling them to participate in the activity of generating and spreading concern about social issues.

References
Ahn, H., Paek, H.-J. & Tinkham, S. (2009) The persuasive effects of anti-binge drinking public service advertising on college students binge drinking: the effectiveness of message sources, message appeals and their interactions. Paper presented at the 2009 annual convention of American Academy of Advertising, Cincinnati, OH. Andsager, J.L., Bemker, V., Choi, H.-L. & Torwel, V. (2006) Perceived similarity of exemplar traits and behavior. Communication Research, 33(1), pp. 318. Austin, E.W. & Meili, H.K. (1994) Effects of interpretations of televised alcohol portrayals on children. Journal of Broadcasting and Electronic Media, 38(4), pp. 417435. Bagozzi, R.P. & Moore, D.J. (1994) Public service advertisements: emotions and empathy guide prosocial behavior. Journal of Marketing, 58(1), pp. 5670.

183

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Beaudoin, C.E. & Thorson, E. (2007) Evaluating the effects of a youth health media campaign. Journal of Health Communication, 12(5), pp. 439454. Bian, J., Liu, Y., Agichtein, E. & Zha, H. (2008) Finding the right facts in the crowd: factoid question answering over social media. Paper presented at the International World Wide Web conference, Beijing, China. Chaiken, S. (1980) Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social Psychology, 39(5), pp. 752766. Cheng, X., Dale, C. & Liu, J. (2007) understanding the characteristics of Internet short video sharing: YouTube as a case study. Available online at: http://arxiv.org/ PS_cache/arxiv/pdf/0707/0707.3670v1.pdf (accessed 15 September 2009). Choi, Y.K., Kim, J. & McMillan, S. (2009) Motivators for the intention to use mobile TV: a comparison of South Korean males and females. International Journal of Advertising, 28(1), pp. 147167. Cialdini, R.B. (2001) Harnessing the science of persuasion. Harvard Business Review, 79, pp. 7281. Croft, A.C. (2008) Emergence of new media moves PR agencies in new direction: competitive pressure threatens agencies livelihood. Public Relation Quarterly, 52, pp. 1620. Daugherty, T., Eastin, M.S. & Bright, L. (2008) Exploring consumer motivations for creating user-generated content. Journal of Interactive Advertising. Available online at: http://www.jiad.org/article101 (accessed 15 September 2009). Department of Health and Human Services (DHHS) (2009) uS Department of Health and Human Services and the Ad Council launch national campaign to help prevent the spread of the H1N1 flu virus. Available online at: http://www.hhs.gov/ news/press/2009pres/10/20091008d.html (accessed 15 September 2009). Dholakia, R.R. & Sternthal, B. (1977) Highly credible sources: persuasive facilitators or persuasive liabilities? Journal of Consumer Research, 3(4), pp. 223232. Dillard, J.P. & Peck, E. (2000) Affect and persuasion: emotional responses to public service announcements. Communication Research, 27(1), pp. 461495. Downs, A. (1972). up and down with ecology the issueattention cycle. Public Interest, 28, pp. 3850. Dozier, D.M., Grunig, L.A. & Grunig, J.E. (2001) Public relations as communication campaign, in Rice, R.E. & Atkin, C.K. (eds) Public Communication Campaigns. Thousand Oaks, CA: Sage, pp. 231248. Ellazar, A., Hawkes, J., Heim, J., Kennedy, J., Vandehey, J., Varg, C. & Watts, A. (2009) Youth public service advertising effectiveness research. Ad Council. Available online at: https://mysite.wsu.edu/personal/jason_hawkes/e-portfolio/ Work%20Samples%20Library/Final%20Plans%20Book.pdf (accessed 15 September 2009). Feick, L. & Higie, R.A. (1992)The effects of preference heterogeneity and source characteristics on ad processing and judgments about endorsers. Journal of Advertising, 21(2), pp. 924. Freeman, B. & Chapman, S. (2008) Gone viral? Heard the buzz? A guide for public health practitioners and researchers on how Web 2.0 can subvert advertising

184

PEER OR EXPERT?

restrictions and spread health information. Journal of Epidemiology and Community Health, 62, pp. 778782. Gangestad, S. & Snyder, M. (1985) To carve nature at its joints? On the existence of discrete classes in personality. Psychological Review, 92(3), pp. 317349. Gilly, M.C., Graham, J.L., Wolfinbarger, M.F. & Yale, L.J. (1998) A dyadic study of interpersonal information search. Journal of the Academy of Marketing Science, 26(2), pp. 83100. Guadagno, R.E. & Cialdini, R.B. (2002) Online persuasion: an examination of gender differences in computer-mediated interpersonal influence. Group Dynamics, 6, pp. 3852. Guttman, N. & Salmon, C.T. (2004) Guilt, fear, stigma and knowledge: ethical issues in public health communication interventions. Bioethics, 18(6), pp. 531552. Hair, J.F., Tatham, R.L., Anderson, R.E. & Black, W. (1998) Multivariate Data Analysis (5th edn). Englewood Cliffs, NJ: Prentice Hall. Harris, G. (1987) The implications of low-involvement theory for advertising effectiveness. International Journal of Advertising, 6(3), pp. 207221. Hass, G. (1981) Effects of source characteristics on cognitive responses and persuasion, in Petty, R.E., Ostrom, T.M. & Brock, T.C. (eds) Cognitive Responses in Persuasion. Hillsdale, NJ: Lawrence Erlbaum, pp. 141172. Heath, R. (2000) Low involvement processing: a new model of brands and advertising. International Journal of Advertising, 19(3), pp. 287298. Homer, P.M. (1990) The mediating role of attitude toward the ad: some additional evidence. Journal of Marketing Research, 27(1), pp. 7886. Hovland, C.I., Janis, I.L. & Kelley, H.H. (eds) (1953) Communication and Persuasion. New Haven, CT: Yale university Press. Hyman, H.H. & Singer, E. (1968) Readings in Reference Group Theory and Research. New York, NY: Free Press. Jenkins, H. (2006) Convergence Culture: Where Old and New Media Collide. New York, NY: New York university Press. Kahle, L.R. & Homer, P.M. (1985) Physical attractiveness of the celebrity endorser: a social adaptation perspective. Journal of Consumer Research, 11(4), pp. 954961. Kelman, H.C. (1961) Processes of opinion change. Public Opinion Quarterly, 25, pp. 5778. Liu, S.S. & Stout, P.A. (1987) Effects of message modality and appeal on advertising acceptance. Psychology & Marketing, 4(3), pp. 167187. Loechner, J. (2008) Cornucopia of April online web and video viewers and brands. Centers for Media Research. Available online at: http://blogs.mediapost.com/research_ brief/?p=1749 (accessed 15 September 2009). Lucas, C.P. (1992) The order effect: reflections on the validity of multiple test presentations. Psychological Medicine, 22(1), pp. 197202. MacKenzie, S.B., Lutz, R.J. & Belch, G.E. (1986) The role of attitude toward the ad as a mediator of advertising effectiveness: a test of competing explanations. Journal of Marketing Research, 23(2), pp. 130143.

185

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

McGinnies, E. & Ward, C.D. (1980) Better liked than right: trustworthiness and expertise as factors in credibility. Personality and Social Psychology Bulletin, 6(3), pp. 467472. McGuire, W.J. (2001) Input and Output Variables Currently Promising for Constructing Persuasive Communications. Thousand Oaks, CA: Sage. Mediaco (2007) using YouTube. Available online at: http://www.media.co.uk/ newsletter/251006/get-your-brand-on-youtube.htm (accessed 15 September 2009). Mook, D.G. (1982) Psychological Research: Strategy and Tactics. New York, NY: Harper & Row. Moscovici, S. (1985) Social influence and conformity, in Lindzey, G. & Aronson, E. (eds) Handbook of Social Psychology: Special Fields and Applications (3rd edn). New York, NY: Random House, pp. 341412. Ohanian, R. (1990) Construction and validation of a scale to measure celebrity endorsers perceived expertise, trustworthiness, and attractiveness. Journal of Advertising, 19(3), pp. 3952. Ohanian, R. (1991) The impact of celebrity spokespersons perceived image on consumers intention to purchase. Journal of Advertising Research, 31(1), pp. 4654. Paek, H.-J. & Gunther, A.C. (2007) How peer proximity moderates indirect media influence on adolescent smoking. Communication Research, 34(4), pp. 407432. Paek, H.-J., Choi, H. & Nelson, M. (2010). Product, personality or prose? Testing functional matching effects in advertising persuasion. Journal of Current Issues in Advertising Research. Perloff, R.M. (2003) The Dynamics of Persuasion: Communication and Attitudes in the 21st Century. Mahwah, NJ: Routledge. Perse, E.M., Nathanson, A.I. & McLeod, D.M. (1996) Effects of spokesperson sex, public service announcement appeal, and involvement on evaluations of safe-sex PSAs. Health Communication, 8(2), pp. 171189. Petty, R.E. & Cacioppo, J.T. (1986a) Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York, NY: Springer. Petty, R.E. & Cacioppo, J.T. (1986b) The elaboration likelihood model of persuasion. Advances in Experimental Social Psychology, 19, pp. 123205. Petty, R.E. & Cacioppo, J.-T. (1990) Involvement and persuasion: tradition versus integration. Psychological Bulletin, 107(3), pp. 367374. Petty, R.E., Brinol, P. & Priester, J.R. (2008) Media Effects: Advances in Theory and Research. New York: NY, Routledge. Pew Internet & American Life Project (2009) The share of online adults who watch videos on video-sharing sites has nearly doubled since 2006. Available online at: http://www.pewInternet.org/Reports/2009/13--The-Audience-for-OnlineVideoSharing-Sites-Shoots-up.aspx (accessed 15 September 2009). Salmon, C.T. & Atkin, C. (2003) using media campaigns for health promotion, in Thompson, T.L., Dorsey, A.M., Miller, K.I. & Parrott, R. (eds) Handbook of Health Communication. Mahwah, NJ: Lawrence Erlbaum Associates, pp. 449472. Shavitt, S., Lowrey, T.M. & Han, S.-P. (1992) Attitude functions in advertising: the interactive role of products and self-monitoring. Journal of Consumer Psychology, 1(4), pp. 337364.

186

PEER OR EXPERT?

Sherif, M. & Sherif, C.W. (1964) Reference Groups; Exploration into Conformity and Deviation of Adolescents. New York, NY: Harper & Row. Shirky, C. (2008) Here Comes Everybody: The Power of Organizing without Organizations. New York, NY: Penguin. Shrum, L.J., McCarty, J.A. & Lowrey, T.M. (1995) Buyer characteristics of the green consumer and their implications for advertising strategy. Journal of Advertising, 24(2), pp. 7182. Snyder, M. & DeBono, K.G. (1985) Appeals to image and claims about quality: understanding the psychology of advertising. Journal of Personality and Social Psychology, 49(3), pp. 586597. Todi, M. (2008) Advertising on social media websites. Wharton Research Scholars School, May, pp. 134. Vance, K., Howe, W. & Dellavalle, R.P. (2009) Social Internet sites as a source of public health information. Dermatologic Clinics, 27(2), pp. 133136. Wilson, E.J. & Sherrell, D.L. (1993) Source effects in communication and persuasion research: a meta-analysis of effect size. Journal of the Academy of Marketing Science, 21(2), pp. 101112. YouTube (2007) About YouTube. Available online at: http://www.youtube.com/t/about (accessed 15 September 2009). Zillmann, D. & Brosius, H.-B. (2000) Exemplification in Communication: The Influence of Case Reports on the Perception of Issues. Mahwah, NJ: Lawrence Erlbaum Associates.

About the authors


Hye-Jin Paek (PhD, university of Wisconsin-Madison) is an associate professor in the Department of Advertising, Public Relations and Retailing at Michigan State university. Her teaching and research interests include health communication, social responsibility in advertising, social marketing and international advertising/health promotion. Thomas Hove (PhD, university of Wisconsin-Madison) is an assistant professor in the Department of Advertising, Public Relations and Retailing at Michigan State university. His recent research focuses on ethical issues in advertising and public relations in both traditional and new media. Hyun Ju Jeong is a doctoral candidate of Department of Advertising, Public Relations and Retailing, Michigan State university. In her research she specialises in the strategic use of new media for advertising effectiveness. Mikyoung Kim (PhD, Michigan State university) is a senior researcher in the Communication Media Research Center at Ewha Womans university,

187

INTERNATIONAL JOuRNAL OF ADVERTISING, 2011, 30(1)

Seoul, Korea. Her research interests focus on consumer information processing of non-traditional marketing communications. Address correspondence to: Hye-Jin Paek, Department of Advertising, Public Relations and Retailing, College of Communication Arts and Science, Michigan State university, East Lansing, MI 48824, uSA. E-mail: paekh@msu.edu

188

Copyright of International Journal of Advertising is the property of World Advertising Research Center Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

You might also like