Theoretical Background
* * *
Theoretical Origin: The Third-Person Effect (TPE)
Influence of presumed influence is an outgrowth of the third-person effect, an influential framework in the past three decades that has rerouted theoretical thinking about media effects. The idea of TPE is quite simple: Individuals tend to perceive a persuasive communication to have a greater impact on other people—the “third persons”—than on themselves (the perceptual component); and such beliefs may lead to real actions (the behavioral component; Davison, 1983). On the perceptual component, findings from a wide array of media contexts have attested to the robustness of the perceptual bias. A recent meta-analysis (Sun, Pan, & Shen, 2008) yields an average effect size of d = .646 (r = .307) based on 372 effect sizes from 106 studies, falling between the “medium” (d = .50) and “large” effect (d = .80; Cohen, 1988), and not subject to variations in methodological factors, such as study setting, population, and design.
Theoretical explanations of such perceptual disparity have been proposed and examined. There is some, but inconclusive, evidence for both motivational accounts, which theorize the perceptual difference as a “bias” resulting from individuals’ inherent drive to protect or enhance their ego (e.g., “self-serving bias,” Gunther & Mundy, 1993; or “self-enhancement bias,” Perloff, 2002), and cognitive accounts, which treat the perceptual difference as an “error” in cognitive processing and judgment-making due to differential cognitive schema or information structure related to self and other (e.g., fundamental attribution error, Gunther, 1991; the self-categorization explanation, Reid & Hogg, 2005). So far the existing evidence suggests that the self-other perceptual difference is probably determined by multiple factors (Perloff, 2009), and TPE scholars are working toward developing an integrated theoretical framework that can “differentiate, incorporate, and explain” both cognitive and motivational factors (Shen, Pan, & Sun, 2010, p. 51).
The empirical research surrounding the behavioral component has yielded a less clear picture. Whether and how the self-other perceptual disparity leads to behavioral consequences remains an open question. Though Xu and Gonzenbach’s (2008) meta-analysis of studies on the TPE behavioral hypothesis reports an overall effect size of r = .13, which the authors claim is “not that trivial” for mass communication research (p. 382), the pool is worryingly small (10 studies with 26 effect sizes) with almost all the studies with inconsistent findings excluded through screening procedures (e.g., Atwood, 1994; Salwen, 1998; Salwen & Driscoll, 1997; Tewksbury, Moy, & Wei, 2004). Understanding of the behavioral component is also limited by a strong bias in empirical research in favor of investigating media regulation or censorship behaviors in negative contexts.
For example, 17 out of the 26 effect sizes in Xu and Gonzenbach’s (2008) meta-analysis are related to censorship behaviors. A rough count of the journal articles published recently (from 2008 till September 2011) on the TPE behavioral component (counted only when self-other perceptual gap is used as a predictor) reveals a sustaining trend: 13 out of the 21 articles have examined support for censorship or media regulation as the outcome variable. When other behavioral contexts are examined, the evidence is often null or counterhypothetical (e.g., Chia, 2007, Choi, Leshner, & Choi, 2008, on ideal body image and dieting behaviors or body dissatisfaction; Eisend, 2008, on scarcity appeal in advertising and purchasing intentions). Given the lack of consistent empirical evidence, the constraint to censorship behaviors, and theoretical under-explication, what we know about the behavioral component of TPE remains quite limited.
Influence of Presumed Influence
Rooted in TPE, IPI grows out of the interest in explaining behavioral reactions as enacted perceptions about media influences. Compared to TPE, IPI posits a similar process: “People perceive some influence of a message on others (italics added) and then react to that perception of influence” (p. 201, Gunther & Storey, 2003). The critical difference between IPI and TPE is that the posited causal antecedent of behaviors in IPI is perceived effects on others, instead of self-other perceptual difference as in TPE. In other words, the presumed influence on others alone, regardless of perceived effect on self, is postulated to be a basis for attitudinal or behavioral decisions.
This theoretical move, as Gunther and Storey (2003) claim, makes IPI a “more general” model “with broader application” (p. 201). According to them, as perceived message influence on others in either positive or negative directions may lead to behavioral consequences, IPI is freed from constraints to negative message contexts and regulation behaviors. As a more general model, they argue, IPI allows for a wider range of attitudinal or behavior consequences to be examined. Indeed, as the empirical review shows in the next section, a much more diverse catalogue of behavioral contexts have been examined under the umbrella of IPI (though it should be noted that this difference is more an outcome of researchers’ choices and does not necessarily reflect the different theoretical scopes of the two frameworks).
Though Gunther and Storey (2003) go as far as to declare that “the third-person effect is just a special case of this broader general model” (p. 201), this claim is somewhat premature. As IPI and TPE propose different causal antecedents (i.e., presumed influence on others vs. self-other perceptual differences), the two are virtually competing hypotheses for the message perception-behavior process. The viability and/or contingency of these two frameworks require further theoretical and empirical investigations. This chapter’s focus on IPI, rather than suggesting that IPI supersedes or transplants TPE, stems from the very recognition that the two literatures need separate scrutiny and synthesis. IPI is chosen as the focus of this chapter for two reasons. First, a few extensive reviews focusing on TPE are already available for interested readers, such as Perloff (2009), Gunther et al. (2008), and Tal-Or, Tsfati, and Gunther (2009), to name a few recent ones. Second, IPI tackles behavioral consequences in a more direct and focused way, and thus is more pertinent to the study of persuasion processes. The wider range of behaviors examined in extant IPI literature also renders the review and discussion more interesting to persuasion scholars.
Current Findings on IPI: A Process View
* * *
IPI as a Multistep Process
Drawing on the previous work on “persuasive press inference” (Gunther, 1998), Gunther and Storey (2003) delineate a few consecutive steps to describe the indirect route from message to behavioral outcomes. First, individuals form basic impressions about the media content upon exposure (self-exposure). Second, through “presumed reach” (Gunther, Bolt, Borzekowski, Liebhart, & Dillard, 2006), individuals assume that the media content they are exposed to reaches a broader audience. Based on their own exposure to the media content, they infer other audience members’ level of exposure (self-exposure à other-exposure). Third, individuals further assume that others, who presumably have viewed or will view the media content, have been or will be influenced by it accordingly (other-exposure à presumed influence). Finally, the presumed reactions by others to the media content serve as guidance for individuals’ own attitudinal or behavioral decisions (presumed influence on others à influence on self). Figure 22.1. depicts this entire process.
This process has been tested in an array of behavioral contexts, with variations at times in terms of what elements are included or how they are measured. The overarching goal guiding these studies is to verify the mediating role of presumed influence on others between message exposure and attitudinal/behavioral manifestations. Empirical findings from different contexts are reviewed next.
Health-Related Attitudes and Behaviors
Smoking
Using data collected from sixth through eighth graders from the spring and fall terms of 2003, Gunther and colleagues examined how smoking-related messages may indirectly influence teenagers’ smoking attitudes and intentions. Testing the IPI process with regard to both pro-smoking and anti-smoking messages with the data from the spring, Gunther et al. (2006) a
scertained that the indirect pathways were significant and in the expected directions. More specifically, teenagers’ exposure to both pro-smoking and anti-smoking messages, via increased estimates of peer exposure to such messages, influenced their perceptions of smoking prevalence among their peers, which then affected their smoking attitudes and susceptibility.
Figure 22.1 Influence of Presumed Influence Process
Source: “The Influence of Presumed Influence,” by A.C. Gunther and J.D. Storey, 2003, Journal of Communication, 53, p. 199-215. Copyright 2003 by Wiley.
Paek and Gunther (2007), focusing on anti-smoking messages using the data from the fall, analyzed non-smokers and smokers separately, and for each group, investigated the potentially differential roles of proximal peers (“your close friends”) versus distal others (“other students your age in your school”). The expected IPI process, for smokers and nonsmokers alike, was confirmed when close friends were the referent others. When distal others were considered, presumed influence had no significant relationship with attitude toward smoking. Consistent with these findings, Paek’s (2009) study on nonsmoking college students found that presumed influence of cigarette advertising on close peers, but not on distant others, mediated the relationship between self-reported exposure to such advertisements and smoking intention. Taking advantage of the two-wave panel data, Paek, Gunther, McLeod, and Hove (2011) examined how the IPI process might unfold over time. Presumed influence of anti-smoking messages on close peers at Time 1 had no direct impact on smoking attitudes or susceptibility at Time 2, though its indirect effect was significant, mediated by presumed influence at Time 2.
Ideal Body Images on the Media and Body Dissatisfaction
Teenagers who consume media representations of ideal body images can be impressionable to the portrayed media norm and perceived peer judgment. Through an in-depth interview, Milkie (1999) uncovered prevalent perceptions held by high-school girls that they were judged by their peers in accordance with the norms portrayed on the media. Gentles and Harrison (2006) showed that African American adolescent girls were not immune to this process either: Increased consumption of media body images led to heightened perceptions of peer expectations using the media images as the standard. More specifically, girls with larger body size tended to think that their peers expected them to be smaller, and those with smaller body size felt the opposite.
Park’s (2005) study, formally testing the IPI process using structural equation modeling (SEM hereafter), showed that exposure to beauty and fashion magazines was associated with perceived prevalence of the thin-ideal images, which then led to greater presumed influence of such images on others. The presumed influence had an indirect impact on one’s desire to be thin, mediated by the presumed influence on self. Direct association between presumed influence and behavioral intentions, however, was either negative, opposing the hypothesis (when other women were considered), or nonsignificant (when other men were considered). In their study of male college students in Singapore, Chia and Wen (2010) found that perceived effects of media portrayals of ideal body images on male friends and female friends were unrelated to body dissatisfaction, intention of going on a diet or going to a gym regularly, but negatively related to intention of going through cosmetic surgery (for “male friends”) and taking diet pills (for “female friends”) respectively.
Sexual Attitudes and Behaviors
Studies that applied the IPI framework to adolescents’ sexual attitudes and behaviors have revealed a rather complicated picture. On one hand, there was supportive evidence suggesting that greater exposure to sex-related media content, positively predicting perceived peers’ exposure to these contents (Chia, 2006; Chia & Lee, 2008), led to adolescents’ perceptions of increasingly permissive peer norms (Chia, 2006; Chia & Gunther, 2006), which then fed into their own sexual attitude and their intentions to engage in sexual activities (Chia, 2006). On the other hand, across these three studies, there was equally strong evidence for the “projection effect” as an alternative explanation (which posits that perceptions of peer norms are a result of one’s projecting their own attitudes onto others). Chia and Gunther (2006) concluded that college students’ misperceptions of peer sexual norms could be a function of both presumed influence of sexual media content and projection of their own attitudes.
Advertising
Advertising and Materialism
Chia and her colleagues adopted the IPI framework to examine how the expanding advertising landscape in Asian countries may contribute to increased materialistic values. Studies of Chinese college students (Jiang & Chia, 2009) and adolescents in Singapore (Chia, 2010) supported the indirect effect of self-exposure to advertising on materialistic attitudes, mediated by perceived peer exposure, and presumed influence on peers. Perceived parents’ viewing of advertisements, however, did not predict perceived level of materialism of the parents (Chia, 2010).
Direct-to-Consumer Advertising
Huh and Langteau’s (2007) study examined how physicians’ perceptions of the influence of DTC advertising on patients may affect their support for government regulations of DTC ads and their own prescription decisions. Support for regulation was explained mainly by physicians’ attitude toward DTC advertising, not by presumed influence on patients. Among a host of prescription-related decisions that were examined in the study, only the refusal to prescribe requested drugs was predicted by the perceived detrimental influence of DTC advertising.
Political Communication
In political realms, IPI has been used as a tool to understand individuals’ decisions on political issues and politicians. Tsfati and Cohen, in their studies of minority groups in Israel (such as Arabs in Israel, Tsfati, 2007; peripheral developmental towns, Tsfati & Cohen, 2003; and Gaza settlers, Tsfati & Cohen, 2005), showed that members of a minority group, perceiving the media coverage of their group to have influenced the general audience, believed that such coverage created or reinforced the negative, stigmatized image of their group in the mind of the public. Such perceptions culminated in a stronger sense of political and social alienation (Tsfati, 2007), greater political inefficacy (Tsfati & Cohen, 2005), and a stronger inclination toward relocation (Tsfati & Cohen, 2003; Tsfati & Cohen, 2005).
Cohen and Tsfati (2009) applied IPI to the study of strategic voting—shifting one’s vote away from a personally preferred party in order not to waste the vote—in Israel, where the multiparty system makes strategic voting more important. They measured strategic voting both using self-report survey data from the years 2003 and 2006, and by identifying actual shifts in votes after the election. Their findings showed that above and beyond the perceived effect of media coverage on self and a host of other relevant factors, the presumed influence of media coverage of the elections on others consistently predicted strategic voting. Sophisticated voters, as shown by their studies, seem to indeed assess the trend of public opinion by gauging the possible media effect on other voters and then make voting decisions on that basis.
Cohen, Tsfati, and Sheafer (2008) turned their attention to political elites and examined how political elites’ belief in media power may play a role in their use of media for achieving political goals. Surveying 56 members from the Israeli Knesset, the authors showed that political elites’ perceptions of media influence on the public, via increasing media motivation and effort, were associated with the increased media coverage they received as well as their parliamentary activities. In other words, the belief in media power on the public was shown to lead politicians to more actively use media to promote their political agenda.
Conceptual Underpinnings: A Component View of IPI
* * *
Explicating “Influence”
The influence of presumed influence, the endpoint of the process that anchors the importance of the model, refers to an individual’s attitudinal and/or behavioral responses resulting from considerations about how relevant social others may react to certain media messages. Taken as a “process concept” (
McLeod & Pan, 2005, p. 17), it implies changes as a result of preceding causes. Taken as a “variable concept” or a “mega-concept” (p. 17), it denotes a collection of responses—attitudinal or behavioral—that come about in various message contexts of interest.
In IPI studies, the construct of “influence” has been operationalized in diverse ways. Based on existing studies and a few previous categorizations (i.e., Gunther et al., 2008; Tal-Or et al., 2009), we can delineate two conceptual dimensions undergirding the behavioral responses. One dimension concerns the direction of response, that is, where an individual shifts his or her actions in relation to the presumed trend of others’ responses. One can decide to act either largely with (converge with) or against (diverge from) the perceived trend. The second dimension concerns the mechanism of response (i.e., the nature of considerations and motives that primarily drive the behavioral decision). Behaviors can be driven primarily by considerations of others’ expectations, interpersonal or social pressures (i.e., normative influence), or by considerations of potential consequences or implications of others’ behaviors (i.e., ecological influence).
The SAGE Handbook of Persuasion Page 76
