Scholarly article on topic 'The Third-Person Effect'

The Third-Person Effect Academic research paper on "Media and communications"

CC BY-NC-ND
0
0
Share paper
Academic journal
Nordicom Review
OECD Field of science
Keywords
{""}

Academic research paper on topic "The Third-Person Effect"

DE GRUYTER

10.1515/nor-2017-0248

The Third-Person Effect

Only a Media Perception?

Bengt Johansson

Abstract

This article focuses on people's beliefs about how other people's political attitudes are shaped and examines how the hypothesis of a third-person effect is related to non-mediated sources of information such as personal experience and interpersonal communication. Also presented are results on the perceived impact of different media such as television, newspapers and political advertising. A representative sample of the Swedish population answered a national survey during the period November - December 2001, and the results show general support for a third-person effect. Mediated information sources and interpersonal communication are believed to influence others more than oneself. Personal experience, on the other hand, is believed to be more important for oneself than for other people, and first-person effects were found among people with a high level of education or a strong political interest. Thus, one conclusion is that people tend to believe their own picture of politics is more dependent on personal experience and that others' political attitudes are more dependent on mass media or people in their social environment. Keywords: third-person effect, political communication

Introduction

According to Davison's (1983) third-person effect, people tend to believe that other people are more influenced by mass media messages than they themselves are. This perception may lead to attitudinal or behavioral outcomes, such as support for censorship of media content. A large number of empirical studies have supported the existence of this perceptual bias. The behavioral component has been analyzed less frequently and enjoys less support (Paul, Salwen, & Dupagne, 2000; Perloff, 1993, 1996, 2002). The literature shows a broad perspective in studying this perceptual bias. Third person-effects have been examined in relation to different types of messages (pro-social vs. antisocial), different types of media genres (advertising, news, rap lyrics) and different types of media sources (biased vs. credible) (Paul, Salwen, & Dupagne, 2000; Perloff 1993, 1996, 2002). One limitation, however, is that this perceptual tendency has only been related to the impact of mass media messages. But is the tendency to believe other people are more vulnerable just a media phenomenon or can it be found in relation to non-mediated forms of information as well? The focus of this article is to examine how the

Bengt Johansson, Senior Lecturer, Ph.D, Department of Journalism and Mass Communication, Göteborg University, SE-405 30 Göteborg, bengt.johansson@jmg.gu.se

hypothesis of third-person effects is related to non-mediated sources of information such as individuals' perception of the influence of personal experience and interpersonal communication. 1

Third-Person Effects and Social Desirability

The third-person effect research has largely focused on controversial or anti-social media messages, in which influence would likely be considered negative, such as libelous news stories (Cohen, Mutz, Price, & Gunther, 1988; Gunther, 1991), pornography/ media violence (Gunther, 1995; Rojas, Shah, & Faber, 1996), media violence (Hoffner et al., 1999; Innes & Zeits, 1988; Scharrer, 2002), violent rap lyrics (Eveland & McLeod, 1999; McLeod, Detenber, & Eveland. 2001; McLeod, Eveland, & Nathanson, 1997) and advertising (Salwen, 1998; Shah, Faber, & Youn, 1999).

This interest in third-person effects in relation to anti-social messages captures one of the basic theoretical explanations for third-person perception, which is self-enhancement. Motivated by a desire to maintain and enhance a positive self-conception, people seem to evaluate themselves in more favorable terms than they evaluate others (Brown, 1986; Duck & Mullin, 1995; Gunther & Thorson, 1992). Mass media messages have traditionally been credited with having powerful and largely undesirable effects on the attitudes and behaviors of a gullible and susceptible mass audience (Miller, 1986; Smith, 1986). This self-serving tendency to believe that other people are more vulnerable than oneself may be, therefore, particularly strong in connection with the mass media, in general, and anti-social messages, in particular.

Past research on third-person effects has focused, therefore, on the impact of media messages with potentially negative effects. But according to the self-enhancement view, if the third-person effect is driven by a motivation to preserve self-esteem, people should be willing to acknowledge that they themselves are influenced by pro-social messages such as public service announcements (PSAs), anti-drunk driving campaigns or seatbelt campaigns. In the literature, this kind of perceptual bias is called a reversed third-person effect or a first-person effect (Neuwirth & Frederick, 2002; Tiedge, Silverblatt, Havice, & Rosenfeld, 1991).

A number of studies have tested the hypothesis using pro-social messages. Even if the concept of social desirability is central in these studies, the designs differ and the results are mixed. Innes and Zeits (1988) compared third-person effects in relation to media violence (negative), political advertising (neutral) and a reduce-drunk driving campaign (positive). It provided general support for the third-person effect, as this perceptual bias was strongest for media violence. However, although the third-person effect was reduced for positive or neutral versus negative messages, there was no first-person effect for these messages. A more common design is to study third-person perception in the context of PSAs. Gunther and Thorson's (1992) study of product advertisements and PSAs found that the third-person effect for PSAs diminished, but that it did not reverse, as had been predicted. Gunther and Mundy (1993) compared people's perception of the influence of "harmful" commercials, such as advertisements for schemes for winning the lottery and slimming pills, with the impact of "beneficial" commercials (advertising for seatbelts and safe sun tanning). The results showed that harmful advertising messages produced third-person effects, but respondents perceived no significant self-other difference in relation to beneficial advertisements. Eveland and McLeod (1999) were also unable to find firstperson effects in connection with pro-social rap lyrics, even if the perceptual bias was weaker as compared with anti-social rap lyrics. The above studies all show that anti-so-

cial messages produce strong third-person effects and that pro-social messages diminish or at least weaken the third-person effect.

There are, however, studies that provide evidence for first-person effects in relation to pro-social messages. Thorson and Coyle (1994) examined the perceived impact of product ads and PSAs. They found traditional third-person effects in relation to product ads, as expected. The results also showed first-person effects for PSAs. Duck and Mullin (1995) found that both positive and negative media content produced third-person effects and that PSAs generally produced first-person effects. However, in a second experiment in the same article, PSAs showed no significant differences. Duck, Terry and Hogg (1985) investigated the perceived impact of AIDS advertisements promoting safe sex. The author's found both third-person effects and first-person effects, depending on the respondent's perception of the advertisement.

Hoorens and Ruiter (1996) also found first-person effects when studying desirable messages. This study showed the strongest support for the notion that motivational factors such as self-enhancement underlie the third-person effect. However, the validity of these results has been called into question (Eveland & McLeod, 1999). Another study finding first-person effects is that of Henrikson and Flora (1999), who looked at how children perceive the impact of pro- and anti-smoking advertisements. The children in the study believed that cigarette ads had greater influence on others than on themselves, but perceived that anti-smoking PSAs had greater impact on themselves than on others.

Other studies have shown that third-person effects related to social desirability, is not limited to message content. By shifting the framing of the message, Brosius and Engel (1996) showed that statements such as "advertising influences me when I go shopping" (pro-social) produced larger third-person effects than did statements such as "advertising stimulates me when I go shopping" (pro-social). Results from the same article also showed that third-person effects become stronger depending on the information channel. The authors found the smallest third-person effect for television news and the greatest for TV commercials. They argued that the ostensibly truthful nature of news makes it more desirable (or less desirable) to believe than other genres. Whoever pays attention to the news and learns from it is considered well informed, and thus well educated. That is why the third-person effect is generally small there. Advertising causes one to buy things one does not need, and propaganda may seduce people to do or think things they never would have done otherwise. In line with these results, Gunther (1991) found the gap between perception of media effects on the self vs. others to be larger when the story was attributed to a biased source (The National Enquirer) than when it was attributed to a more objective and credible one (The New York Times). Studies have also found person effects by framing news stories in different ways. Neuwirth and Fredrick (2002) found both first- and third-person effects when studying the perceived influence of news stories containing racial cues.

Even if first-person effects can be found in the literature, the third-person effect seems to be the norm. In their meta-analysis of the perceptual aspect of the third-person effect, Paul, Salwen and Dupagne (2000) concluded that it seems reasonable to assume that, lacking any message desirability, the general undesirability associated with believing a media message will result in third-person perception.

In line with the self-enhancement explanation, there should be first-person effects in connection with pro-social messages. But as the literature review shows, there is no clear pattern of findings for pro-social media messages. One explanation put forward by Eveland and McLeod (1999) is that self-enhancement, as a motivational factor, should be replaced by the ego-defensive version of the motivational explanation. The third-person effect, as an ego-defensive response rather than an ego-enhancement response,

implies that the perception is less pronounced when the valence of the message makes the admission of personal influence less ego-threatening (Duck, Terry, & Hogg, 1995).

One weakness of many studies, however, is that perception of social desirability is not measured, but instead assumed (Eveland & McLeod, 1999; Paul, Salwen, & Dupagne, 2000; Peiser & Peter, 2001). Although most people would agree that it is socially undesirable to be influenced by media violence or pornography, it is less clear how advertising and news stories should generally be judged. And even if PSAs and other messages are seen as pro-social in the literature, it is not evident that respondents' actually view these messages as pro-social. This may explain why first-person effects have less frequently been found in previous third-person effect research. Another explanation may be that the pro-social messages tested are not pro-social enough. Eveland and McLeod (1999) discussed why pro-social rap lyrics do not produce first-person effects. One explanation may be that the pro-social rap lyrics studied were not desirable enough to produce first-person effects. And this perspective can be seen more broadly. Perhaps the very admission that mass media messages have general effects on oneself is seen as a negative experience. Even if mass media messages advocate beneficial and intelligent outcomes, it seems that people are often reluctant to admit to personal influence, perhaps preferring to think that change is generated internally, rather than through outside influence (Duck, Terry, & Hogg, 1995; Gunther & Mundy, 1993).

Third-person Effects and the Shaping of Political Attitudes

There are bodies of literature focused on opinion formation and the role of mass media and how these relate to other factors such as interpersonal communication and personal experience (Lazarsfeld & Katz, 1956; Lenart, 1994; Mutz, 1998). Although media messages are important, results clearly show that direct media effects do not tell the whole story.

Even if there is no such thing as "all-powerful media", it is perhaps the case that people generally credit the mass media with having great power to form attitudes. According to the third-person effect hypothesis, people may believe that others are more influenced by media messages about politics than they themselves are. But do people's perceptions of other forms of political communication also produce third-person effects?2 The purpose of this study was to test the third-person effect in connection with mediated and non-mediated sources of information to examine whether third-person effects are limited to media messages or whether they can be observed in relation to other sources of information as well.

When studying perceptions such as the third-person effect, however, one should consider cultural differences. Although it does seem to be a cross-cultural phenomenon (Paul, Salwen, & Dupagne, 2000), sociopolitical factors may moderate the effect. People may perceive media influence differently depending on factors such as press freedom, advertising regulations, availability of media and media use. This study was carried out Sweden, and therefore the media system, media use and the role of political advertising as a channel for political communication in Sweden need to be considered. Although Sweden is known to be among the world's top five in terms of Internet use, newspaper reading is also strong. More than 80 percent of the adult population read a newspaper on an average day, and almost all newspapers are local or regional. At the beginning of the new century, the television market was dominated by the two public service channels (SVT1, SVT2) and the commercial channel (TV4) (Weibull, 2001).

Political advertising is not permitted on public service radio, public service television or on TV4. The main channels for political advertising are, therefore, newspapers, bill-

boards, cinemas and to some extent commercial radio. Thus, the political parties do not view political advertising as a particularly important strategic campaign channel. Media coverage and interpersonal communication are thought to be more effective (Esaiasson, 1990; Nord & Strömbäck, 2003).

Hypotheses

Reviewing the literature on the third-person effect and social desirability enables formulation of hypotheses concerning mediated and non-mediated sources of information in connection with political communication. Because previous research indicates that media influence per se is probably seen as negative, the first hypothesis of this study is:

H1: The third-person effect will be stronger for mediated sources of information than for non-mediated sources of information.

But as suggested by previous research, the key factor underlying third-person perception may be the attempt to persuade, which may explain why the traditional third-person effect is not always reversed, but only weakened or diminished for pro-social messages or credible media genres/sources. Logically, then, personal experience should be the most self-enhancing way to achieve attitude formation or behavior change as compared with media messages as well as non-mediated sources of information such as interpersonal communication. Believing that you have a greater ability to learn from personal experiences than do others is a self-enhancing tendency. This leads to the second hypothesis:

H2: The influence of personal experience on others will be perceived to be less than such an influence on the self (first-person effect).

Research has shown that credibility perceptions also affect the third-person effect. Advertising is seen as less credible than other genres such as news (Brosius & Engel, 1996). In the literature, we also find that newspapers are generally seen as less credible than television news (Carter & Greenberg, 1965; Gaziano & McGrath, 1986, Holmberg & Weibull, 2004). In line with the theory of social desirability presented above, advertising should be the least desirable genre in terms of effects on the self, newspapers more desirable and television the most desirable. This leads to the third hypothesis:

H3: The third-person effects gap will be greater for lower credibility media than for higher credibility media.

Method

The survey National SOM (Society, Opinion, Mass media) is carried out every autumn in the form of a questionnaire mailed to 3,000 randomly selected individuals (between 15 and 85 years of age) in Sweden, using the Swedish National Population Register as the sampling frame. The central questions addressed in the survey are attitudes about mass media, politics and public services. The survey is the result of an interdisciplinary collaboration between the Institute for Journalism and Mass Communication, the Department of Political Science and the School of Public Administration at Göteborg University.

The National SOM 2001 was carried out during November - December 2001, and 69 percent of the respondents answered the questionnaire. To test the quality of the

sample, demographics were compared between the Swedish population (using official statistics) and the group of respondents. These comparisons show very small differences between the sample and the population: women tend to answer more than men, and young people (15-29) and the very old (80-85) are not as likely to answer. These differences are on the whole rather small (3-5 percent), and the conclusion must be that the sample is a good reflection of the Swedish population (Nilsson, 2002).

Measures

Effects of Mediated and Non-mediated Factors

Respondents were asked two sets of questions about their perceptions of the influence of personal experience, interpersonal communication, television, newspapers and political advertising on the formation of political attitudes. In the first set, respondents were asked how these information sources influenced their own political attitudes. The exact wording was:

Discussions sometimes concern what influences our picture of politics. How important do you think these information channels are for your own political attitudes?

In the second set of questions, respondents judged the perceived influence of the information sources on other people in general. The respondents answered the questions using a scale from 4 (very important) to 1 (very unimportant). 3

To measure third-person perception, a "perceptual bias" transformation variable was created by computing the difference between each individual's other-scale and self-scale. Perceptual bias ranged from 3 (4 for others minus 1 for the self) to - 3 (1 for the self minus 4 for others). Positive values indicated more effect on others than on oneself, or third-person perceptual bias. Negative values indicated more effect on oneself than on others, or a first-person effect. A zero value indicated no perceptual bias.

Third-person Controls

Several variables previously identified (Paul, Salwen, & Dupagne, 2000; Perloff 1993, 1996, 2002) as important predictors of the third-person effect were employed in an effort to minimize alternative explanations.

Education was measured by one question, which included a number of different types of education levels. These were computed into three categories: Low education (elementary school), middle-high education (high school) and high education (college/university).

The Age variable was based on the respondent's own statement of year of birth. In cases of missing age-information from the respondents, such information was taken from the register. In the regression models the age variable ranged from 15 to 85 years.

The Gender variable was based on the respondent's own replies, which were completed with register data if necessary. Male=1, Female=2.

Political interest can be seen as an indicator of involvement. It was measured by one question on interest in politics, ranging from 1 (very limited interest) to 4 (very strong interest).

Second-person effect. In recent studies of third-person effects, it has been suggested that a second-person effect should be considered. This is a component of the "diamond

model" developed by Whitt (1983) and recently adopted in third-person effect research (Eveland, Nathanson, Detenber, & McLeod, 1999; McLeod, Eveland, & Nathanson, 1997; Neuwirth & Frederick, 2002; Neuwirth, Frederick, & Mayo, 2002). The second-person effect represents the perceived joint influence of the media on the self and others. A parallel, additive "self plus other" variable was therefore created, which indicates this perceived joint influence of the information channels. The difference term in a regression represents a first- or a third-person effect, depending upon direction, controlling for the level of both (self and other) variables.

Results

The first hypothesis (H1) proposed stronger third-person effects for mediated sources of information than for non-mediated sources of information. The results indicated strong support for the third-person effect for mass mediated sources (Table 1). A series of paired t-tests show significant self-other differences for television (t=19,72, p < .000, N = 1649), newspapers (t = 16,53,p < .000, N = 1667) and political advertising (t = 26,789,p < .000, N = 1670). But the results also showed that the third-person effect could be related to interpersonal communication (t = 22,49,p < .000, N = 1659). People's perception of the role of interpersonal communication in shaping political attitudes seems to be equal to how mediated information channels are viewed. There was, however, no significant difference between belief in influence on self and others in relation to personal experience, and H1 is therefore both rejected and supported. The third-person effect was stronger for mediated information channels than for personal experience, but equal to interpersonal communication. This indicates that the third-person effect is not just a media effect perception, as it also seems to have implications for perception of the influence of other sources of information. The results also strengthen the argument that the key factor underlying third-person perception may be the attempt to persuade, and not necessarily mass media messages per se. Perception of persuasion - through mass media or other sources - produces third-person effects.

Table 1. Perceived Effects of Mediated and Non-mediated Information Sources on the Self's and Others' Formation of Political Attitudes (mean)

M Self SD Others M SD Person Effect Others-Self

Personal experience N=1 680 3,02 0,85 3,03 0,73 0,01

Interpersonal communication N=1 659 2,50 0,80 2,94 0,68 0,44

Television N=1 649 2,59 0,77 2,96 0,71 0,37

Newspapers N=1 667 2,52 0,77 2,84 0,71 0,32

Political advertising N=1 670 1,70 0,71 2,21 0,80 0,51

In line with this argument, H2 proposed a first-person effect for personal experience, but this was not supported by the results. Although the third-person effect for personal experience did diminish, there was no first-person effect as suggested by the hypothesis.

However, the relationship between personal experience and third-person perception is more complex. Table 2 presents both first- and third-person perception bias for the influence of personal experience, interpersonal communication, television, newspapers and political advertising on political attitudes.

Table 2. Distribution of Respondents ' Types of Perceptual Bias

Direction of Perceptual Bias First-Person No difference Third-Person N

Personal experience 18 % 63% 19% 1 680

Interpersonal communication 43% 49% 8% 1 659

Television 36% 57% 7% 1 649

Newspapers 34% 56% 10% 1 667

Political advertising 44% 52% 4% 1 670

The results showed two rather strong opposite tendencies in people's views on the role of personal experience in attitude formation. Although most people believed others and themselves to be equally influenced by personal experience, 37 percent showed a perception bias, but in different directions. Nineteen percent assumed others were more influenced by personal experience (third-person effect) and 18 percent believed themselves to be more influenced (first-person effect). The table also shows that the third-person effect bias is weaker for personal experience than for other sources of information and that the first-person effect is stronger. Despite the fact that there is no firstperson effect on an aggregate level, the results show that first-person perceptions must be taken into consideration when the role of personal experience is discussed in relation to third-person effects.

The third hypothesis (H3) predicted that third-person effects would be stronger for less credible mass media - a notion that is supported by the results. Political advertising, which is probably seen as more biased than newspapers and television, produced the strongest third-person effect. Paired t-tests comparing third-person perception showed significant differences between the third-person effect of political advertising as compared with newspapers (t= 9,14, p <.000, N= 1654) and television (t = 6,82, p < .000, N = 1636). In line with H3, the third-person effect should also be stronger for newspapers than for television. This notion, however, was not supported. Results indicated a significant difference in judgment of the impact of television and newspapers, but in opposite directions (t = 3,45, p < .001, N=1634). In relation to the perceived influence on oneself, people tend to believe that others' political attitudes are more influenced by television than by newspapers.

In Table 3, important predictors of third-person effects were employed in an effort to minimize alternative explanations. In the literature, education has been shown to be a strong predictor of third-person effects. This was also the case here.

However, the relationship goes in different directions depending on information source. The third-person effect becomes stronger with higher education for interpersonal communication (beta = .12 p < .001), television (beta = .11 p < .001), and political advertising (beta = .11 p < .001). With regard to personal experience, the relationship was the opposite. Here the results showed a first-person effect (beta = -.07 p < .01). People with a higher level of education seemed to believe they are more influenced by their own experience than are others.

Table 3. Multiple Regressions Predicting First- and Third-Person Effects (standardized regression coefficients)

Personal experience Interpersonal communication Television Newspapers Political advertising

Education -0,07 ** 0,12 *** 0,11 *** 0,01 0,11 ***

Political interest -0,21 *** 0,07 *** 0,01 -0,04 0,03

Age -0,04 -0,08 *** -0,09 *** -0,13 *** -0,06 **

Gender 0,10 *** 0,00 -0,03 0,02 -0,04

2 PE joint effect1 -0,11 *** -0,21 *** -0,12 *** -0,11 *** 0,13 ***

R2 (%) .10 .06 .03 .03 .04

N= 1 588 1 568 1 560 1 577 1 578

1 Controls for level of information channels joint effect on self and other group when predicting self-other differences

Note: * p< .05; ** p< .01; *** p< .001.

Previous research has shown that involvement may also cause third-person effects. But we only found third-person effects for perceived influence of interpersonal communication (beta = .07 p < .001), which shows that individuals with greater interest in politics assume that others' political attitudes are formed by interpersonal communication to a greater extent than are their own. A first-person effect was also found in relation to the role of personal experience (beta = -.21 p < .001). Respondents with strong interests in politics seemed to consider personal experience as more important for shaping their own political attitudes than for shaping others' political attitudes.

In previous research, age has been seen as a predictor of the third-person effect. The theory proposes that, with increasing age, people may judge themselves to be less influenced by the media. The present data did show a person effect in relation to age, but it went in the opposite direction. First-person effects were found for television (beta = -.09p < .001), newspapers (beta = -.13 p < .001) and political advertising (beta = -.06 p < .001). Age also predicted a first-person perception for interpersonal communication (beta = -.08 p < .001).

The results also showed a rather weak relationship between gender and third-person perception. There are, however, some significant results worth mentioning. Men believe they are more influenced by personal experience (beta = .10 p < .001). This indicates that men, as compared with women, have a view on attitude formation wherein their own picture of politics is more dependent on personal experience.

The perception ofjoint influence (second-person effects) showed that second-person effects produced a first-person effect in relation to all information channels. An increased perception of joint influence on oneself and others is associated with the perception that the self is more influenced than others are.

To explore the second-person effect, an analysis was performed on the level of media (and personal experience and interpersonal communication) influence on the self and others, controlling for self-other differences (first- and third-person effects); see Table 4.

The results showed that increased education and political interest produced significant second-person effects. People with greater interest in politics and greater educational attainment are more likely to perceive that both themselves and others are jointly influenced by politics. The only exception was predicting estimated influence on political advertising, where education was not a significant predictor. Gender and age were also significant predictors of second-person effects. In general, women, as compared

with men, perceive that they and others are jointly influenced by the mass media and that personal experience and interpersonal communication are more influential. The results also show that second-person effects decrease with age. The older you are, the less likely you are to believe that all sources jointly influence yourself and others.

Table 4. Multiple Regressions Predicting Second-Person Effects (standardized regression coefficients)

Personal Interpersonal Political

experience communication Television Newspapers advertising

Education 0,29 *** 0,22 *** 0,23 *** 0,23 *** -0,01

Political interest 0,09 *** 0,06 *** 0,06 *** 0,06 *** -0,11 ***

Age -0,08 *** -0,17 *** -0,17 *** -0,18 *** -0,22 ***

Gender 0,11 *** 0,12 *** 0,13 *** 0,13 *** 0,13 ***

Person effect -0,10 *** -0,20 *** -0,12 *** -0,10 *** 0,12 ***

R2 (%) .14 .12 .10 .10 .08

N= 1 588 1 568 1 560 1 577 1 578

Note: * p< .05; ** p< .01; *** p< .001.

Discussion

The purpose of this study was to examine how the hypothesis of the third-person effect is related to non-mediated sources of information such as personal experience and interpersonal communication. It also presented results on the perceived impact of different media such as television, newspapers and political advertising. The results showed general support for third-person effect perception. The first hypothesis (H1) proposed stronger third-person effects for mediated information sources than for non-mediated information sources. This was supported when comparing personal experience with television, newspapers and political advertising. But the hypothesis was rejected with regard to interpersonal communication, where respondents perceived equally strong third-person effects for mediated sources of information. This suggests that third-person effects may be driven by beliefs that communication processes, or more precisely attempts to persuade, have more powerful effects on others than on the self.

The second hypothesis (H2) proposed that personal experience would be seen as more beneficial and therefore be perceived to have stronger effects on oneself than on others. Based on the results, however, this notion was rejected. There was no significant self-other discrepancy in perceptions of attitude formation in relation to personal experience on an aggregate level. But as the analysis showed, people's view of direct personal experience and influence on oneself versus others is more complex. The results do provide support for the hypothesis when the analysis is controlled for predictors such as education and involvement (political interest). Why, then, is this perception limited to these factors? One reason may be that respondents have varying perceptions of the social desirability of personal experience in relation to attitude formation. The results showed that people with a high level of education and those with strong interests in politics perceived personal experience to have more influence on the self than on others. One explanation may be that these groups, as compared with other groups, consider personal experience to be more socially desirable. One limitation of this study is, how-

ever, that respondents were not asked about their perceptions of the social desirability of personal experience and other sources of information. Future research should address the question of whether perception of the social desirability of different information channels can explain perceptual self-other discrepancies in relation to political communication.

The results also supported the third hypothesis (H3), which predicted that perceived source credibility creates third-person effects when comparing political advertising to television and newspapers. Although political advertising, as compared with newspapers and television, is a marginal phenomenon in the Swedish political culture, people do tend to believe it can have an impact - especially on others.

But this hypothesis was rejected when comparing newspapers and television. Television and television news are generally seen as more credible than newspapers. But television produced significantly stronger third-person effects than did newspapers. There are at least two potential explanations. First, there is some recent research indicating that newspapers are perceived as more credible than television (Flanagin & Metzger, 2000, Kiousis, 2004). A shift in people's perceptions of media credibility may then explain why the hypothesis was rejected. Yet studies carried out in Sweden give no indication of this shift in perception of media credibility (Holmberg & Weibull, 2004).

A second - and more plausible - explanation may be that perceptions of media impact should be considered. Research shows that general beliefs in the power of media per se produce third-person effects (Brosius & Engel, 1996; Mutz, 1989; Price, Huang &, Tweksbury, 1997). Because television is generally considered to have a great impact on society as a whole (Smith 1987), the results indicate that not only the credibility of a message source should be taken into account, but also its perceived impact.4 Thus, the working principle here should not be the perceived credibility of the information source, but its perceived impact or persuasive power. This may explain why television produces stronger third-person effects than do newspapers.

The present tendency toward finding more significant second-person effects than third-/first-person effects supports the findings of other studies. And as Neuwirth, Frederick and Mayo (2002) suggested, Davison's original idea that people act on the basis of greater perceived media influence on others than on themselves may have to be modified. People's general belief in media influence should also be considered, and the results show that second-person effects seem to be more robust than third-person effects. Their results showed that 11 of 20 coefficients were significant when predicting third-person effects, whereas 19 of 20 coefficients were significant when predicting second-person effects (Neuwirth, Frederick, & Mayo, 2002).

The present study also provides general theoretical support for the idea that third-person perception is closely related to what is seen as self-enhancing tendencies and more important, that these tendencies seem to be cross-cultural. Most people view advertising as strongly threatening their own free will. Personal experience may be seen as the polar opposite of political communication. Believing that you yourself have a greater ability to learn from personal experiences than do others is also a self-enhancing tendency. In many people's view, learning from your own experiences instead of from media messages or from those around you may be an ideal. Interestingly, this picture of political communication is more prevalent among people with a higher level of education (and in some aspects also among people with strong interests in politics). These results strengthen the hypothesis that more highly educated people tend to see themselves as less vulnerable to influence from external information sources, and that they believe they are more in control of their own opinions than are other citizens.

1. This study was carried out within the research project Images of Media Power: Perceptions of Media Effects in Political Communication and was financed by The Bank of Sweden Tercentenary Foundation.

2. In the third-person effect literature, there are a number of studies of third-person effects and political communication: campaign effects (Cohen, & Davis 1991; Rucinski & Salmon, 1990), political identification (Duck, Terry, & Hogg, 1995), political outspokenness (Willnat, 1996), support for campaign message censorship (Salwen, 1998) and perception of political ads for candidates of different genders (Hitchon, Chang, & Harris, 2001)

3. The wording of this question was: "How important do you think these information channels are for other people's political attitudes? One possible weakness of this study is, however, the wording of the question. The scale used is anchored using important-unimportant, which may capture channel importance more than perceived channel influence. Although the question context provided to the respondents does refer to influence, this remains a weakness. Future studies examining third-person effects in relation to non-mediated information channels should use, therefore, question phrasing used in other third-person effect research.

4. An analysis of joint influence also showed that it is stronger for television than for newspapers.

References

Brosius, H-B., & Engel, D. (1996) The Causes of Third-person Effects: Unrealistic Optimism, Impersonal Impact, or Generalized Negative Attitudes Towards Media Influence? International Journal of Public Opinion Research 7, 142-162.

Brown, J.D. (1986) Evaluations of Self and Others: Self-enhancement Biases in Social Judgements'. Social Cognition, 4, 353-376.

Carter, R., & Greenberg, B. (1965) Newspapers and Television: Which do You Believe? Journalism Quarterly, 42. 29-34.

Cohen, J., & Davis, R.G. (1991) Third-person Effects and the Differential Impact in Negative Political Advertising. Journalism Quarterly 68, 680-688.

Cohen, J., Mutz, D., Price, V., & Gunther, A.C. (1988) Perceived Impact of Defamation. Public Opinion Quarterly 52, 161-173.

Davison, W.P. (1983) The Third-person Effect in Communication. Public Opinion Quarterly 47, 1-15.

Duck, J.M., & Mullin, B.A. (1995) The Perceived Impact of the Mass Media: Reconsidering the Third-Person Effect. European Journal of Social Psychology, 25, 77-93.

Duck, J.M., Hogg, M.A. & Terry, D.J. (1995) Me, Us and Them. Political Communication and the Third-person Effect in the Australian Federation Election. European Journal of Social Psychology, 25, 195215.

Duck, J.M., Terry, D.J., & Hogg, M.A. (1995) The Perceived Influence of AIDS Advertising: Third-person Effects in the Context of Positive Media Content. Basic and Applied Social Psychology, 17, 305-325.

Esaiasson, P. (1990) Svenska valkampanjer 1866-1988. Stockholm: Publica.

Eveland, W.P., Jr., & McLeod, D.M. (1999) The Effect of Social Desirability on Perceived Media Impact: Implications for Third-person Perceptions. International Journal of Public Opinion Research, 11, 315-333.

Eveland, W.P., Jr., Nathanson, A.I., Detenber, B.H., & McLeod, D. M. (1999) Rethinking the Social Distance Corollary. Perceived Likelihood of Exposure and the Third-person Perception. Communication Research, 26, 275-302.

Flanagin, A.J., & Metzger, M.J. (2000) Perceptions of Internet Information Credibility. Journalism & Mass Communication Quarterly, 77, 515-540.

Gaziano, C., & McGrath, K. (1986) Measuring the Concept of Credibility. Journalism Quarterly, 63,451462.

Gunther, A.C. (1991) What we Think Others Think. Cause and Consequence in the Third-person effect. Communication Research 18, 355-372.

Gunther, A.C. (1995) Overrating the X-rating: The Third-person Perception and the Support for Censorship of Pornography. Journal of Communication 45, 27-38.

Gunther, A.C., & Mundy, P. (1993) Biased Optimism and the Third-person Effect. Journalism Quarterly. 70, 58-67.

Gunther, A.C., & Thorson, E. (1992) Perceived Persuasive Effects of Product Commercials and Public Service Announcements. Third-person Effects in New Domains. Communication Research 19, 547-596.

Henriksen, L., & Flora, J.A. (1999) Third-person Perception and Children. Perceived Impact of Pro- and Anti-smoking Ads. Communication Research, 26, 643-665.

Hitchon, J.C., Chang, C., & Harris, R. (1997) Should Woman Emote? Perceptual Bias and Opinion Change in Response to Political Ads for Candidates of Different Genders. Political Communication, 14, 4969.

Hoffner, C., Buchanan, M., Anderson, J.D., Hubbs, L.A., Kamigaki, S.K., Kowalczyk, L., Pastorek, A., Plotkin, R.S., & Silberg, K. J. (1999) Support for Censorship of Television Violence. The Role of the Third-person Effect and News Exposure. Communication Research 26, 726-742.

Holmberg, S., & Weibull, L. (2004) Samlande institutionsförtroende. In S. Holmberg, & L. Weibull (Eds.), Ju mer vi är tillsammans. Göteborgs universitet: SOM-institutet.

Hoorens, V., & Ruiter, S. (1996) The Optimal Impact Phenomenon: Beyond the Third-person Effect'. European Journal of Social Psychology, 26, 599-610.

Innes, J.M., & Zeits, H. (1988) The Public's View on the Impact of the Mass Media: A Test of the "Third-person" Effect. European Journal of Social Psychology, 18, 457-463.

Katz, E., & Lazarsfeld, P.F. (1955) Personal Influence. Glencoe: Free Press.

Kiousis, S. (2001) Public Trust or Mistrust? Perceptions of Media Credibility in the Information Age. Mass Communication & Society 4, 381-404.

Lenart, S. (1994) Shaping Political Attitudes. The Impact of Interpersonal Communication and Mass Media. Thousands Oaks, CA: SAGE.

McLeod, D.M., Detenber, B.H., & Eveland, W.P. Jr. (2001) Behind the Third-person Effect: Differentiating Perceptual Processes for Self and Other. Journal of Communication 51, 678-695.

McLeod, D.M., Eveland, W.P. Jr., & Nathanson, A.I. (1997) Support for Censorship of Violent and Misogynic Rap Lyrics. An Analysis of the Third-person Effect. Communication Research 24, 153174.

Miller, G.R. (1987) Persuasion. In C.R. Berger & S.H. Chaffe (Eds.), Handbook of Communication Science. Newbury Park, CA: Sage.

Mutz, D. (1989) The Influence of Perception of Media Influence. International Journal of Public Opinion Research 1, 3-24.

Mutz, D.C. (1998) Impersonal Influence. How Perceptions of Mass Collectives Affect Political Attitudes. Cambridge: Cambridge University Press.

Neuwirth, K., & Frederick, E. (2002) Extending Framework of Third-, First, and Second-Person Effects. Mass Communication & Society, 5, 113-140.

Neuwirth, K., Frederick, E. & Mayo, C. (2002) Person-effects and Heuristic-systematic Processing. Communication Research, Vol. 29:320-359.

Nilsson, Â. (2002) Den nationella SOM-undersökningen 2001. In Holmberg, S., & Weibull, L. (Eds.) Det vâras för politiken. Trettiotvâ artiklar om politik, medier och samhälle. SOM-institutet: Göteborgs University.

Nord, L. & Strömbäck, J. (2003) Valfeber och nyhetsfrossa - politisk kommunikation i valrörelsen 2002. Stockholm: Stiftelsen Institutet för Mediestudier.

Paul, B., Salwen, M.B., & Dupagne, M. (2000) The Third-person Effect: A Meta-analysis of the Perceptual Hypothesis. Mass Communication & Society, 3, 57-85.

Peiser, W., & Peter, J. (2001) Explaining Individual Differences in Third-person Perception. A Limits/ possibilities Perspective. Communication Research, 28, 156-180.

Perloff, R. M. (1993) Third-person Effect Research 1983-1992: A Review and Synthesis. International Journal of Public Opinion Research 7, 167-184.

Perloff, R.M. (1996) Perceptions and Conceptions of Political Media Impact: The Third-person Effect and Beyond. In Crigler, A.N. (Eds.) The Psychology of Political Communication. Ann Arbor: Michigan University Press.

Perloff, R.M. (2002) The Third-person Effect. In B. Jennings & D. Zillman (Eds.) Media Effects: Advances in Theory and Research. NJ: Mahwah.

Price, V., Huang, L.N., & Tewksbury, D. (1997) Third-person Effects of News Coverage: Orientations toward Media. Journalism & Mass Communication Quarterly 74, 525-540.

Rojas, H., Shah, D.V., & Faber, R.J. (1996) For the Good of Others: Censorship and the Third-person effect. International Journal of Public Opinion Research 7, 163-186.

Rucinski, D., & Salmon, C.T. (1990) The "Other" as the Vulnerable Voter. A Study of the Third-person Effect in the 1988 U.S. Presidential Campaign. International Journal of Public Opinion Research 2, 345-368.

Salwen, M.B. (1998) Perception of Media Influence and Support for Censorship. The Third-person Effect in the 1996 Presidential Election. Communication Research 25, 259-285.

Scharrer, Erica (2002) Third-person Perception and Television Violence. The Role of Out-group Stereotyping in Perception of Susceptibility to Effects. Communication Research 29, 681-704.

Shah, D.V., Faber, R.J., & Youn, S. (1999) Susceptibility and Severity: Perceptual Dimensions Underlying the Third-person Effect. Communication Research 26, 240-267.

Smith, R. (1986) Television Addiction. In J. Bryant & D. Zillman (Eds.) Perspectives on Media Effects. Hillsdale, NJ: Lawrence Erlbaum.

Thorson, E., & Coyle, J. (1994) The Third-person Effect in Three Genres of Commercials: Product and Greening Ads, and Public Service Announcements. In K. King (Ed.) Proceedings of American Academy of Advertising. Athens, GA: University of Georgia.

Tiedge, J.T., Silverblatt, A., Havice, M.J., & Rosenfeld, R. (1991) Discrepancy Between Perceived Firstperson and Third-person Mass Media Effects. Journalism Quarterly 68, 141-154.

Weibull, L. (2001) The Swedish Media Landscape. Structure, Economy and Consumption. In Carlsson, U. & Harrie, E. (Eds.) Media Trends 2001 in Denmark, Finland, Norway and Sweden. Statistics and Analyses. Nordicom: Göteborg University.

Whitt, H.P. (1983) Status Inconsistency: A Body of Negative Evidence or a Statistical Artifact? Social Forces, 62(1), 201-233.

Willnat, L. (1996) Mass Media and the Political Outspokenness in Hong Kong: Linking the Third-person Effect and the Spiral of Silence. International Journal of Public Opinion Research 7, 187-212.