http://psp.sagepub.com/ Bulletin

Personality and Social Psychology

http://psp.sagepub.com/content/34/10/1303 The online version of this article can be found at:

DOI: 10.1177/0146167208320061

2008 34: 1303 originally published online 3 July 2008Pers Soc Psychol Bull Laura E. Buffardi and W. Keith Campbell

Narcissism and Social Networking Web Sites

Published by:

http://www.sagepublications.com

On behalf of:

Society for Personality and Social Psychology

can be found at:Personality and Social Psychology BulletinAdditional services and information for

http://psp.sagepub.com/cgi/alertsEmail Alerts:

http://psp.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://psp.sagepub.com/content/34/10/1303.refs.htmlCitations:

What is This?

- Jul 3, 2008 OnlineFirst Version of Record

- Sep 3, 2008Version of Record >>

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

1303

Narcissism and Social Networking Web Sites

Laura E. Buffardi W. Keith Campbell University of Georgia

Recently, there has been a tremendous amount of attention in the media surrounding the issue of narcis- sism and social networking Web sites (e.g., Baldwin & Stroman, 2007; Orlet, 2007; Vaidhyanathan, 2006). The concern is that these Web sites offer a gateway for self-promotion via self-descriptions, vanity via photos, and large numbers of shallow relationships (friends are counted—sometimes reaching the thousands—and in some cases ranked), each of which is potentially linked to trait narcissism.

The vast popularity of these sites suggests that the general psychology of the members will be largely nor- mative. That is, the 21 million members of Facebook arguably (although it is, of course, still an empirical question) look similar to others in society with similar demographics. This was not necessarily the case in the past and still might not be the case with lower base rate Web presences, like freestanding personal Web pages. For example, a study published in 2006 showed that owners of personal Web sites did differ from controls on several dimensions of the Big Five, and women, but not men, did differ in narcissism with female Web page owners reporting the higher narcissism scores (Marcus, Machilek, & Schütz, 2006).1

Our focus will thus not be on the mean-level narcis- sism score of the average social networking Web site user. Instead, we will address the question of how narcis- sism is manifested in these sites: Does narcissism predict

Authors’ Note: We would like to thank Laura Aikens, Laura Aquilino, Joel Frost, Jesse Hauch, Abby Levin, and Ben Porter for their assis- tance with data collection and coding. Please address correspondence to Laura E. Buffardi or W. Keith Campbell, Department of Psychology, University of Georgia, Athens, GA 30602-3013; e-mail: [email protected] or [email protected].

PSPB, Vol. 34 No. 10, October 2008 1303-1314 DOI: 10.1177/0146167208320061 © 2008 by the Society for Personality and Social Psychology, Inc.

The present research examined how narcissism is manifested on a social networking Web site (i.e., Facebook.com). Narcissistic personality self-reports were collected from social networking Web page own- ers. Then their Web pages were coded for both objective and subjective content features. Finally, strangers viewed the Web pages and rated their impression of the owner on agentic traits, communal traits, and narcis- sism. Narcissism predicted (a) higher levels of social activity in the online community and (b) more self- promoting content in several aspects of the social networking Web pages. Strangers who viewed the Web pages judged more narcissistic Web page owners to be more narcissistic. Finally, mediational analyses revealed several Web page content features that were influential in raters’ narcissistic impressions of the owners, includ- ing quantity of social interaction, main photo self- promotion, and main photo attractiveness. Implications of the expression of narcissism in social networking communities are discussed.

Keywords: narcissism; social networking Web sites; Internet; self-presentation

The migration of individuals, especially teenagers andyoung adults, onto the Internet has occurred in staggering proportions. In particular, social networking Web sites—nonexistent just years ago—have drawn literally millions of users. Web sites such a MySpace .com (total users: 90 million per month; Stone, 2007) and Facebook.com (total users: 21 million members; Geist, 2007) have been at the forefront of this migration. These Web sites offer individuals the abilities, among others, to (a) create an individual Web page, (b) post self-relevant information (e.g., self-descriptions, photos), (c) link to other members (e.g., “friends lists”), and (d) interact with other members.

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

overall activity in a Web community? Is narcissism appar- ent in the content of the Web page, and if so, how? Finally, can the narcissism of a page owner be gleaned from the content of the Web page? Before discussing our design in detail, we briefly describe the construct of nar- cissism, review the literature on the social qualities of the Internet, and specify our predictions.

Narcissism and Social Behavior

Narcissism refers to a personality trait reflecting a grandiose and inflated self-concept. Specifically, narcis- sism is associated with positive and inflated self-views of agentic traits like intelligence, power, and physical attractiveness (Brown & Zeigler-Hill, 2004; Campbell, Rudich, & Sedikides, 2002; Gabriel, Critelli, & Ee, 1994; John & Robins, 1994) as well as a pervasive sense of uniqueness (Emmons, 1984) and entitlement (Campbell, Bonacci, & Shelton, 2004).2 From a basic trait perspective, narcissism is associated with a high degree of extraversion/agency and a low level of agree- ableness or communion (e.g., Miller & Campbell, 2008; Paulhus & Williams, 2002). A similar high agency pattern (and negative but typically small/nonsignificant correla- tions with communion) is also found in narcissists’ explicit self-conceptions (Campbell, Foster, & Finkel, 2002), implicit self-conceptions (Campbell, Bosson, Goheen, Lakey, & Kernis, 2007), and implicit motives (Carroll, 1987).

Central to most theoretical models of narcissism in social-personality psychology is the use of social relation- ships in part to regulate self-esteem, self-concept positiv- ity, or narcissistic esteem (Campbell, 1999; Campbell, Brunell, & Finkel, 2006; Morf & Rhodewalt, 2001; Raskin, Novacek, & Hogan, 1991). Narcissists do not focus on interpersonal intimacy, warmth, or other posi- tive long-term relational outcomes, but they are very skilled at both initiating relationships and using rela- tionships to look popular, successful, and high in status in the short term. Narcissists participate in this dynamic “self-construction” (Morf & Rhodewalt, 2001) via rela- tionships to constantly affirm their narcissistic esteem. It has been suggested that this process is due, at least par- tially, to narcissists’ dispositional impulsivity (Vazire & Funder, 2006). How does this narcissistic self-regulation or self-construction (we use these terms largely inter- changeably) operate in the context of interpersonal rela- tionships more specifically? First, narcissism is linked positively with relationship formation. For example, nar- cissism is associated with being (a) liked in initial inter- actions (Oltmanns, Friedman, Fiedler, & Turkheimer, 2004; Paulhus, 1998), (b) perceived as a leader (Brunell, Gentry, Campbell, & Kuhnert, 2006), (c) perceived as exciting (Foster, Shrira, & Campbell, 2003), (d) socially

confident (Brunell, Campbell, Smith, & Krusemark, 2004), (e) entertaining (Paulhus, 1998), and (f) able to obtain sexual partners (Foster, Shrira, & Campbell, 2006). Second, narcissism is associated negatively with seeking out or creating long-term relationships that have qualities of closeness, empathy, or emotional warmth (Brunell et al., 2004; Campbell, 1999; Campbell & Foster, 2002). Third, narcissism is associated with using relationships as an opportunity or forum for self- enhancement. For example, narcissists brag and show off (Buss & Chiodo, 1991), perform well when there is an opportunity for public glory (Wallace & Baumeister, 2002), and seek attractive, high-status, “trophy” roman- tic partners (Campbell, 1999). Others in relationships with narcissistic individuals, however, often suffer in the longer term as narcissism is linked to aggressiveness (Bushman & Baumeister, 1998), psychological control (Campbell, Foster, et al., 2002), game playing and infi- delity (Campbell, Foster, et al., 2002; Le, 2005; Schmitt & Buss, 2001), and lower levels of commitment (Campbell & Foster, 2002). Indeed, longitudinal research on rela- tionships has found that the initial likeability associated with narcissism fades and is even reversed in the longer term (Paulhus, 1998). Similarly, longitudinal research in clinical settings has found a significant long-term conse- quence of narcissism is the suffering of close others (Miller, Campbell, & Pilkonis, 2007).

How might narcissism operate in a social networking Web site? These online communities may be an especially fertile ground for narcissists to self-regulate via social connections for two reasons. First, narcissists function well in the context of shallow (as opposed to emotionally deep and committed) relationships. Social networking Web sites are built on the base of superficial “friendships” with many individuals and “sound-byte” driven commu- nication between friends (i.e., wallposts). Certainly, indi- viduals use social networking sites to maintain deeper relationships as well, but often the real draw is the abil- ity to maintain large numbers of relationships (e.g., many users have hundreds or even thousands of “friends”). Second, social networking Web pages are highly con- trolled environments (Vazire & Gosling, 2004). Owners have complete power over self-presentation on Web pages, unlike most other social contexts. In particular, one can use personal Web pages to select attractive photographs of oneself or write self-descriptions that are self-promoting. Past research shows that narcissists, for example, are boastful and eager to talk about themselves (Buss & Chiodo, 1991), gain esteem from public glory (Wallace & Baumeister, 2002), are prevalent on reality television (Young & Pinsky, 2006), and enjoy looking at themselves on videotape and in the mirror (Robins & John, 1997). Personal Web pages should present a similar opportunity for self-promotion.

1304 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

In sum, given the behaviors of narcissists in offline relationship contexts, we expect that they will take advantage of the new virtual arena for pursuing a simi- lar self-regulatory agenda. This will include relatively high levels of relationship formation, self-promoting images of oneself, and an overall agentic (rather than communal) self-presentation.

Personality Manifestation on Web Pages

To what extent will Web page observers be able to esti- mate Web page owners’ narcissism? In order to make pre- dictions about the expression of narcissism on Web pages, it is important to ask two broader questions: Is personal- ity, in general, apparent on personal Web pages? And if so, how is personality inferred from Web pages?

Past research has shown that personality can be accu- rately detected from Web pages and that this detection is linked to specific Web page content (e.g., Marcus et al., 2006; Vazire & Gosling, 2004). For example, research concerning self-presentation on the Internet has shown that Web site viewers tend to make appropriate infer- ences about owners’ Big Five personality based on per- sonal Web pages, particularly with regard to Openness to Experience (Marcus et al., 2006; Vazire & Gosling, 2004). Correlations between individuals’ actual person- ality scores and the perceptions of these personality dimensions drawn from their Web sites were similar in magnitude to those impressions drawn from other personality-reflective but nonelectronic contexts (e.g., their bedrooms; Vazire & Gosling, 2004).

In addition, evidence suggests that Web page viewers utilize page content to form impressions of owners’ per- sonalities. Marcus and colleagues (2006) found that Web site features relevant to each of the Big Five personality dimensions correlated with viewers’ impressions of the owners’ Big Five personalities. For example, a couple of the correlations observed in this study are as follows: (a) Impressions of high Extraversion correlated with the number of photographs owners posted picturing them- selves with other individuals and themselves at parties, and (b) Conscientious Web page owners tended to post their resumes and count the number of visitors to their sites.

In sum, past research suggests that personality fac- tors are (a) somewhat accurately expressed by Web page owners through personal Web page content and (b) appropriately inferred by Web site viewers. These find- ings give us reason to expect that narcissism will also be expressed by Web page owners and inferred by Web page viewers based on Web page content.

The Present Research

The goal of this study was to examine the role of nar- cissism in personal Web pages on social networking

sites. We were particularly interested in social network- ing Web sites rather than other types of personal Web pages for two reasons. First, they are a venue for both self-presentation and social interaction. Social network- ing sites are designed for not only conveying informa- tion about one’s self to others but also, to a greater extent than other forms of personal Web sites, for being social online. Second, social networking sites are extremely popular and are still growing in popularity.

There are a number of social network sites available to Internet users; among the most widely used are MySpace.com, Friendster.com, and Facebook.com. We chose Facebook for our investigation for two reasons. First, Facebook is the most commonly used site by individuals in our sample—college students. Facebook.com was originally designed as a way for college students to electronically communicate with other students at their own university and, later, with others across the country. When it was first created, a university e-mail address was required to register and to set up an account. Although now all e-mail addresses are granted accounts, Facebook.com remains popular with college students (about half of its members are in college; Kirkpatrick, 2007). Second, Facebook is the most structured social network Web site. That is to say, Facebook profile pages have a fixed format and this consistency allows for a more con- trolled comparison of Web pages. A Facebook page utilizes a fill-in-the-blank system of personalization. All pages share common social characteristics, such as links to friends’ pages, membership in virtual groups, and an electronic bulletin board, called the wall, where friends post messages to the owner. Facebook pages also have sections designated for the owner’s self-pre- sentation. In a section titled About Me, page owners typically describe themselves and their likes and dis- likes. Another section is designated for Quotes. This is where page owners list their favorite quotes, either from literature, popular culture, or funny anecdotes from real-life conversations. Uploaded photographs are the other main self-presentational element of a Facebook page. The owner’s main picture appears the most frequently. It is dominant on the page in size and appears on every wallpost the owner makes and each time the owner comes up in a member search. Albums are a subsidiary feature that act as an online photo gallery and typically include pictures of friends, par- ties, trips, and so on. Since the data for this study were collected, Facebook has evolved (Locke, 2007). Although the main features discussed here remain prominent, users now have more options in the struc- ture and content of their profiles. Researchers who plan to use Facebook in future research should be aware of these changes.

Buffardi, Campbell / NARCISSISM AND WEB SITES 1305

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

Our investigation consisted of four steps:

1. We examined the link between narcissism and social activity in a Web community.

2. We inspected the associations between narcissism and the content of social networking Web pages. Consistent with past research on narcissistic self-regulation, narcis- sism should be associated with (a) a greater amount of social activity (Hypothesis 1) and (b) self-promotion in text and pictures (Hypothesis 2).

3. We tested whether individuals viewing a social network- ing page could detect the owner’s expression of agentic/ communal traits and level of narcissism. Consistent with past research on personality and Web pages, we predicted that narcissism would be perceived by Web page observers, as would the general agentic qualities associated with narcissism (Hypothesis 3). As a corol- lary to this, we hypothesized that narcissistic Web page content would relate positively to observers’ narcissistic impressions (Hypothesis 4).

4. We examined which social networking page content was the most influential in forming impressions of narcissism (i.e., mediational analyses, Hypothesis 5). Although it is reasonable to assume that there are ele- ments of the Web page that lead viewers to judge the Web page owner’s personality, we did not predict that one type of content or another would be more central in this process.

METHOD

Participants

Owners. Owner participants were 156 undergradu- ate Facebook page owners (100 females) participating for partial course credit. They ranged in age from 18 to 23 years (M = 18.97, SD = 1.14). A total of 129 (83%) gave permission to save their Facebook pages to be used in the present research. Those who did not give permis- sion to save their Facebook pages were excluded from the main analyses.

Raters. Rater participants were 128 undergraduates (86 females) participating for partial course credit. They ranged in age from 18 to 26 years (M = 19.41, SD = 1.59).

Materials and Procedures

Personality data and Facebook page collection. Owner participants came to the lab individually. Upon consenting to participate, they first completed demo- graphic information and the Narcissistic Personality Inventory (NPI; Raskin & Terry, 1988). The NPI is a 40-item forced-choice format personality questionnaire designed for use on a normal population. Example items include “My body is nothing special” versus “I like to look at my body” and “I am more capable than other people” versus “There is a lot that I can learn

from other people.” Higher scores on the NPI indicate more narcissistic personality (range = 2 to 30, M = 17.11, SD = 6.00, α = .78).

Second, participants were presented with a waiver form to sign if they consented to having their Facebook pages saved for this research project (129 agreed). The mean narcissism score for those participants who granted permission to save their Facebook Web pages (M = 17.14, SD = 6.21) did not differ significantly from the mean narcissism score of those who did not grant permission (M = 16.67, SD = 4.95), t(154) = –.37, ns. Before signing the waiver form, participants were informed that if they gave permission to have their Facebook page saved, we would gather information from the page and other undergraduates would also view the page and report on the impressions they drew from it. Participants were also assured that before other undergraduates viewed their Facebook page, identifying information, such as last name, phone number, and e-mail address, would be deleted.

Third, if the participant declined, he or she was fully debriefed. If the participant granted permission to have his or her Facebook page saved, he or she was asked to log onto Facebook.com on a computer in the lab and dis- play his or her main profile page on the Internet browser. An important distinction to point out is that owner par- ticipants were asked to display their Facebook profile page as opposed to their Facebook home page. The pro- file page is the page that others in their networks have access to viewing. The home page can be thought of as a control panel that only the owner has access to. The experimenter then saved the profile page. Participants were also asked to display the Web page linked to the profile page text View Photos of Me. This link displays 20 photographs. The first 12 are the most recent that the owner posted, regardless of which album they belong to. The second 8 are the most recent pictures that the own- ers’ friends have uploaded and tagged the owner in. Tagging a photo on Facebook means that the photo will be linked to the tagged individual’s profile page. Owners have the ability to accept or decline tags and therefore have control of all the pictures on the View Photos of Me link. The experimenter saved this page as well, and finally, the participants were fully debriefed.

Objective Facebook page coding. Objective criteria were extracted from the Facebook pages. These were quantitative features that are common to standard Face- book profiles. The four objective criteria were (a) number of friends (M = 171.21, SD = 121.90), (b) number of wall- posts (M = 479.94, SD = 437.29), (c) number of groups (M = 35.17, SD = 32.27), and (d) number of lines of text in the About Me section (M = 4.95, SD = 8.75).

A quantity of social interaction measure was created by summing of the number of friends linked to the

1306 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

Facebook page (standardized) and the number of wall- post messages that were accumulated on the Facebook page (standardized) (range = –2.44 to 7.09, SD = 1.79, α = .75).3 A quantity of information about self measure was simply the number of lines of text that the owner had written in the section titled About Me (M = 4.95, SD = 8.75).

Subjective Facebook page coding by research assis- tant (RA) raters. Four features of the 129 saved Facebook pages were coded independently by five undergraduate RAs for the extent to which they were self-promoting. Each rater coded (a) the content of the About Me section, (b) the content of the Quotes section, (c) the main profile photograph (usually a headshot of the owner), and (d) the 20 pictures on the page linked to View Photos of Me. These four features were coded separately, that is, without the context of the other fea- tures present (e.g., the items from the About Me section and the Quotes section were pasted into separate docu- ments and separated from the photos).

The contents of the About Me section were rated on the following characteristics: self-absorbed, self-conscious (reversed), self-important, and self-promoting. The con- tents of the Quotes section were rated on arrogant, clever, entertaining, and self-promoting. The main profile photograph was rated on the amount of clothing worn, modest (reversed), physically attractive, self-promoting, sexy, and vain. The pictures from the View Photos of Me link were rated on exciting, fun, modest (reversed), provocative, self-centered, and self-promoting. Impor- tantly, prior to completing the coding, all of the RA coders met to agree on the definitions of the adjectives that each subjective feature was rated on. For example, self-promoting was distinguished from self-important in that self-promoting connoted persuading others about one’s own positive qualities, whereas self-important expressed a statement about viewing one’s self as impor- tant or as an individual with important opinions. All ratings were on a 7-point Likert scale ranging from 1 (not at all) to 7 (very much). The agreement of the five raters was sufficient (alphas ranging from .60 to .90, with the majority between .75 and .80). Thus, the means of the five RA raters’ ratings were taken.

From the RA rater means, the following specific nine measures were created:

1. A measure of self-promoting information posted about self was created by taking the mean of the adjectives (across coders) that the About Me sections were rated on: self-absorbed, self-conscious (reversed), self-important, and self-promoting (M = 3.67, SD = .89, α = .76).

2. A measure of self-promoting quotes was created by tak- ing the mean of the coders’ judgments of quote arro- gance and self-promotion (M = 1.83, SD = .83, α = .94).

3. A measure of entertaining quotes was created by taking the mean of the coders’ ratings of how clever and enter- taining the quotes seemed (M = 2.91, SD = .85, α = .88).

4. The mean of the coders’ judgments of the physical attrac- tiveness of the owner’s main photo served as the measure of main photo attractiveness (M = 3.88, SD = 1.19).

5. A measure of main photo self-promotion was created by taking the mean of the coders’ ratings of how self- promoting and vain the photo was (M = 2.29, SD = .92, α = .89).

6. A measure of main photo sexiness was created by tak- ing the mean of the coders’ judgments of how sexy and modest (reversed) and clothed (reversed) the individual in the main photo appeared to be (M = 2.31, SD = .80, α = .71).

7. A measure of self-promoting pictures was created by taking the mean of the coders’ judgments of how self- centered and self-promoting the pictures were (M = 2.46, SD = .97, α = .95).

8. A measure of provocative pictures was created by taking the mean of the coders’ judgments of how provocative and modest (reversed) the pictures were (M = 2.50, SD = .99, α = .85).

9. A measure of fun pictures was created by taking the mean of the coders’ judgments of how fun and exciting the pictures were (M = 3.70, SD = .73, α = .88).

Facebook page impression ratings. Rater participants came to the lab individually. Upon consenting to partic- ipate, each was randomly assigned five Facebook pages that were collected from owners earlier in the study. Facebook page owners’ identifying information (i.e., last names, e-mail addresses, and phone numbers) was removed from their Facebook pages. If the raters indi- cated that they had been assigned the Facebook page of someone they knew (a friend or acquaintance), they were told not to rate that page and it was replaced ran- domly with another.

Rater participants viewed the Facebook pages they were assigned one at a time and then rated the extent to which they believed the owner of the page possessed 37 different personality traits. Each rater rated only five Facebook pages to minimize fatigue and the potential for perceiver effects (Kenny, 1994). The personality traits that the raters used in their evaluations were drawn roughly from previous research (Campbell et al., 2007; Paulhus 1998; see the following). The pages were rated on a 7-point Likert scale ranging from 1 (not at all) to 7 (very much). From these ratings, we created three indices to measure raters’ impressions of the Facebook owners’ trait-level agency, communion, and narcissism. In computing these indices we examined the consistency of the five raters in their ratings of the adjectives com- prising each index. Because the interrater reliabilities were sufficient (agency α = .92, communion α = .92, and narcissism α = .88), we computed the means of the five raters’ ratings for each of the 37 traits.

Buffardi, Campbell / NARCISSISM AND WEB SITES 1307

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

An agentic impression composite was created by tak- ing the mean of the raters’ average impressions of Facebook page owners on the following traits: assertive, active, boring (reversed), confident, dominant, ener- getic, entertaining, enthusiastic, high in status, impor- tant, inhibited (reversed), intelligent, outspoken, quiet (reversed), reserved (reversed), silent (reversed), with- drawn (reversed), submissive (reversed) (M = 4.98, SD = .53, α = .93).

A communal impression composite was created by taking the mean of the raters’ average impressions of Facebook page owners on the following traits: affection- ate, cooperative, cruel (reversed), friendly, generous, grouchy (reversed), hostile (reversed), kind, likeable, mean (reversed), pleasant, quarrelsome (reversed), rude (reversed), stingy (reversed), warm (M = 5.41, SD = .53, α = .95).

A narcissistic impression composite was created by taking the mean of the raters’ average impressions of Facebook page owners on the following traits: arro- gant, narcissistic, self-centered, vain (M = 2.75, SD = .80, α = .92).

RESULTS

Simple Correlations

To test Hypothesis 1 that narcissism would predict social activity and the quantity of information posted about the self on Facebook pages, we correlated narcissism scores with the quantity of social interaction composite and the quantity of information about self (both measures of objective information drawn from Facebook pages that we collected from owner participants).4 As predicted, higher scores on the NPI were related to higher quanti- ties of interaction on Facebook. No relationship was found between owners’ narcissism scores and the quan- tity of information owners posted about themselves.5 (All correlations can be seen in Table 1.)

We next correlated owners’ narcissism scores with RA coders’ perceptions of Facebook profile features to test Hypothesis 2. First, owners’ narcissism was corre- lated with the self-promoting information about self measure. A marginally significant positive relationship was found between owners’ narcissism and the extent to which information posted about the owner in the About Me section was self-promoting.6 Second, own- ers’ narcissism scores were correlated with the self-pro- moting quotes and entertaining quotes measures. Correlations revealed that those owners with high nar- cissism scores posted marginally more self-promoting quotes than those with low narcissism scores and their quotes were less entertaining and clever than those with low narcissism.

We then examined the relationships between owners’ narcissism and the contents of the images posted on the Facebook pages. First, narcissism scores were correlated with the main photo attractiveness, the main photo self- promotion composite, and the main photo sexiness composite. Owners with higher NPI scores were seen as more physically attractive in their main photograph than those with low NPI scores. Likewise, in their main photograph, narcissistic owners also appeared to be sexier and more self-promoting than nonnarcissists as predicted by Hypothesis 2. Second, correlations were tested between owners’ narcissism scores and the self- promoting pictures, provocative pictures, and fun pic- tures measures. Narcissistic owners posted pictures that were more fun, but contrary to Hypothesis 2, they were not perceived to be significantly more provocative or self-promoting than nonnarcissistic owners’ pictures.

Next, to assess Hypothesis 3 that observers can detect a Web page owner’s self-reported narcissism by viewing his or her Facebook Web page, correlations were tested between owners’ NPI scores and participant raters’ nar- cissistic impression as well as agentic impression and communal impression. As predicted, narcissistic Facebook page owners were perceived to be more nar- cissistic. In addition, narcissism was associated with a pattern of high perceived agency and no link to per- ceived communion, very similar to the relationship found in narcissists’ explicit and implicit self-ratings (Campbell et al., 2007; Campbell, Rudich, et al., 2002).

1308 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

TABLE 1: Facebook Correlates With Owners’ Narcissism Scores and Raters’ Narcissistic Impression Ratings

Raters’ Owners’ Narcissistic

Narcissism Impression

Objective Facebook page criteria Quantity of social interaction (α = .75) .23*** .31*** Quantity of information listed about .07 .29***

self

Subjective Facebook page criteria coded by research assistant (RA) raters Self-promoting information about .18* .32***

self (α = .76) Self-promoting quotes (α = .94) .19* .21** Entertaining quotes (α = .88) –.28*** –.11 Main photo attractiveness .18** .33*** Main photo self-promotion (α = .89) .19** .37*** Main photo sexiness (α = .71) .20** .33*** Self-promoting pictures (α = .95) .10 .28*** Provocative pictures (α = .85) .07 .34*** Fun pictures (α = .88) .18** .17*

Impression ratings by participant raters Communal impression (α = .95) –.09 –.60*** Agentic impression (α = .93) .28*** .39*** Narcissistic impression (α = .92) .25***

*p ≤ .10. **p ≤ .05. ***p ≤ .01.

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

According to the magnitude of these correlations, the association between owners’ NPI scores and raters’ agentic impression scores was stronger than that between owners’ NPI scores and raters’ narcissistic impressions. To determine if raters’ perception of narcissism was fully cued by the perception of agency, owners’ NPI scores were regressed onto raters’ narcissistic impression and agentic impression simultaneously. A significant coeffi- cient for agentic impression, β = .21, t(125) = 2.31, p = .02, and a marginally significant coefficient for narcissistic impression, β = .16, t(125) = 1.75, p = .08, resulted, indicating that raters’ perceptions of both agency and narcissism accounted for unique variance (at least mar- ginally) in owners’ narcissism.

Finally, we tested the prediction set forth in Hypothesis 4 that Web page content indicative of nar- cissism would relate positively to raters’ narcissistic impressions. First, we correlated raters’ narcissistic impressions with the objective information drawn from the owners’ Facebook pages. As predicted, higher nar- cissistic impression ratings were related to higher quan- tities of social interaction on Facebook and higher quantities of information posted about the self.

Second, correlations between raters’ narcissistic impressions and RA coders’ subjective ratings of the written information on the Facebook pages were tested. In accord with Hypothesis 4, a significant positive rela- tionship was found between raters’ narcissistic impression and the extent to which self-promoting information about self was posted. Raters’ narcissistic impressions were also positively associated with self-promoting quotes but not related to the entertaining quotes measure.

Third, we examined the relationships between raters’ narcissistic impressions and the contents of the images posted on the owners’ Facebook pages. Owners who were judged to be more narcissistic were seen as more physically attractive in their main photograph, sexier, and more self-promoting than those viewed as less nar- cissistic. Raters’ narcissistic impressions were also posi- tively related to posting pictures that were perceived to be more self-promoting, more provocative, and more fun, as predicted by Hypothesis 4.

Mediation Analyses

We have established that owners’ narcissistic person- ality relates to objective and subjective Facebook page content. We have also shown that impressions of nar- cissism based on solely viewing Facebook pages are somewhat accurate and that the content features that are linked with owners’ narcissism also generally corre- late with raters’ narcissistic impression. We now test Hypothesis 5 to determine whether Facebook page con- tent features mediate the link between owners’ narcissistic

personality and raters’ narcissistic impression drawn from Facebook page observations. We also ask: If mediation occurs, what content of Web pages is influential in observers’ detection of narcissism?

All of the Facebook page content features that corre- lated significantly (p < .05) with both narcissistic personality scores and raters’ narcissistic impression scores were considered potential mediators, including (a) quantity of social interaction, (b) main photo attrac- tiveness, (c) main photo sexiness, and (d) main photo self-promotion. To determine whether together these four Facebook page content mediators fully mediated the relationship between owners’ narcissism and raters’ narcissistic impression, bootstrapping analyses were conducted using methods described by Preacher and Hayes (2007) for estimating direct and indirect effects with multiple mediators. There are three advantages to using this statistical method: (a) Multiple mediators can be tested simultaneously, (b) it does not rely on the assumption of a normal sampling distribution (see MacKinnon, Lockwood, & Williams, 2004; Preacher & Hayes, 2004; Shrout & Bolger, 2002), and (c) the number of inferential tests is minimized, thus reducing the likelihood of Type 1 error. Raters’ narcissistic impres- sion was entered as the dependent variable, owners’ narcissism was entered as the predictor variable, and quantity of social interaction, main photo attractiveness, main photo sexiness, and main photo self-promotion were entered as proposed mediators in the SPSS macro created by Preacher and Hayes for bootstrap analyses with multiple proposed mediators.

The bootstrap results indicated that the total effect of owners’ narcissism scores on raters’ narcissistic impression (total effect = .03, p = .01) became non- significant when the Facebook page content mediators were included in the model (direct effect of owners’ narcissism = .02, ns). Furthermore, the analyses revealed, with 95% confidence, that the total indirect effect (i.e., the difference between the total and direct effects) of owners’ narcissism on the outcome variable through the four mediators was significant, with a point estimate of .02 and a 95% BCa (bias-corrected and accelerated; see Efron, 1987) bootstrap confidence interval of .0070 to .0305. Thus, the four Facebook page content characteristics fully mediated the associa- tion between owners’ narcissism and raters’ narcissistic impression. The specific indirect effects of each pro- posed mediator showed that quantity of social inter- action, with a point estimate of .0052 and 95% BCa CI of .0010, .0154; main photo attractiveness, with a point estimate of .0088 and 95% BCa CI of .0002, .0138; and main photo self-promotion, with a point estimate of .0044 and 95% BCa CI of .0019, .0235 were all unique mediators; whereas main photo sexiness, with a

Buffardi, Campbell / NARCISSISM AND WEB SITES 1309

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

point estimate of –.0023 and 95% BCa CI of –.0149, .0048, did not add to the overall model.7

In sum, the bootstrap analyses indicate that (a) quan- tity of social interaction, (b) main photo attractiveness, and (c) main photo self-promotion together mediate the link between Facebook page owners’ narcissistic person- ality and Facebook page viewers’ ratings of their narcis- sistic impression of the owners (see Figure 1 for full mediational model).

DISCUSSION

Is the narcissism of a Web page owner in a social net- working community related to Web site activity, content, and perception by others? It appears that this is the case. In terms of objective criteria on the Web page, narcissism is related to a measure of Web site activity derived from the number of friends and the number of wallposts (i.e., messages) posted between friends. Narcissism was not found to be related to length of self-description on the Web page, however. In terms of the RA coder ratings of Web page written content, narcissism is positively (but only marginally) related to self-promoting information about the self and quotes and negatively related to enter- taining quotes (this latter finding is interesting in that it differs from related findings in direct social interaction; e.g., Paulhus, 1998). Similarly, in terms of coder ratings of Web page image content, narcissism is positively associated with main photograph attractiveness, self-promotion,

and sexiness. There were fewer relationships with the other photo album pictures, which were judged to be more fun but not more self-promoting or provocative. Although the sizes of these effects are small, given the complexity and the nuances of Facebook profiles, they are notable.

In terms of overall impression of the social network- ing Web page, strangers rated narcissistic individuals as more agentic (but not more communal) and more nar- cissistic, and furthermore, their impressions of narcis- sism and agency uniquely predicted Web page owners’ narcissism. Narcissistic impressions were related to all of the Facebook page content features that narcissistic personality scores were related to as well as three addi- tional features, namely, the quantity of information listed about self, self-promoting pictures, and provoca- tive pictures. Finally, mediational analyses revealed that the impression of narcissism is based primarily on the number of social interactions along with the extent to which the Web page owner appears to be self-promoting and attractive in his or her main photo.

Implications

This research has several implications both for nar- cissism and for Web page–mediated social interaction. First, the expression of narcissism on social networking Web sites is very similar to its expression in other social domains. Narcissism is related to a higher number of social relationships, self-promoting self-presentation, and the perception of having a large number of agentic characteristics. Only two differences were found between narcissism in the “real world” and in online communi- ties. The narcissists’ quotes were judged to be less enter- taining than those of nonnarcissists. This contrasts with the general finding that narcissists are entertaining at first meeting (e.g., Paulhus, 1998). We would like to point out that our incongruent finding, however, is not sufficient to suggest that narcissists’ ability to entertain others in social contexts is subdued online. Rather, it is possible that narcissists’ quotes are actually quite enter- taining to peers who know their context but not to our RA coders who were not acquainted with the profile owners personally. Based on our findings, it is also plau- sible that the extent to which the narcissists’ quotes and other aspects of their profiles were viewed as self- promoting directly interfered with the quotes’ entertain- ment and cleverness. Therefore, additional research in which the entertainment value of narcissists’ full profile, rather than just the quotes section, is necessary to deter- mine whether narcissists are less entertaining in online contexts than in face-to-face ones. In addition, the nar- cissists were judged to be more attractive based on their photos than the nonnarcissists. This differs from past research that has found no differences in attractiveness

1310 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Figure 1 Facebook page content mediators of narcissism-narcissistic impression link.

NOTE: Path values represent unstandardized regression coefficients. The value outside of the parentheses represents the total effect of nar- cissism on narcissistic impression prior to the inclusion of the mediat- ing variables. Value in parentheses represents the direct effect, from bootstrapping analyses, of narcissism on narcissistic impression after the mediators are included. *p < .05. **p < .01.

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

between narcissists and nonnarcissists (Gabriel et al., 1994). Why this disparity? One possible explanation is that narcissists’ self-enhancing biases might lead them to consider more attractive pictures of themselves to be more accurate representations of their true likenesses. We would also speculate, however, that the narcissists appear to be attractive on Facebook because they are strategically posting pictures that make them appear sexy and attractive; this was not an option in Gabriel and colleagues’ (1994) study.

Second, unacquainted raters are able to judge Web page owners’ narcissism with some level of accuracy. This finding is consistent with those concerning the accuracy of Big Five personality perception based on viewing Web pages (Marcus et al., 2006; Vazire & Gosling, 2004) in that Web page viewers can make rea- sonable estimates of personality from Web pages. More broadly, the correlation between self-reported narcissism and strangers’ impressions of narcissism also suggests that at least on Facebook, less narcissistic participants do not seem to be using the Internet as a channel for self- promotion. That is, more modest, less self-centered indi- viduals in real life do not appear to be self-promoting to the degree that narcissists are on the Web.

Third, our results indicate that, as in past Big Five research (Marcus et al., 2006), viewers use Web page content to form impressions of the Web page owner’s level of narcissism. We found that the quantity of social interaction and main photo in particular played a signif- icant role in this process. Conversely, the raters seemed to have omitted other Facebook features that related to NPI scores, including self-promoting information about the self, self-promoting quotes, less entertaining quotes, main photo sexiness, and fun pictures, in their impres- sion formation. Finally, narcissistic impression ratings related (falsely) to three profile characteristics that did not correlate with owners’ narcissistic personality, namely, quantity of information listed about the self, self-promoting pictures, and provocative pictures. This pattern in raters’ errors of omission and commission pre- liminarily suggests that individuals who have experience with social networking Web sites, as presumably our raters did, may have developed a system or script (Schank & Abelson, 1977) for viewing such Web pages and efficiently gleaning important information for the context of a social networking community. It is under- standable, if this is the case, that an individual’s social capital (how popular and socially active they are quan- tifiably) and attractiveness, which both can be deter- mined quickly, play an important role in impression formation, as our results suggest. In contrast, the some- what more difficult to decipher written information, such as that in the About Me and Quotes sections, may

be secondary. This remains speculative, however, and might be an interesting avenue for future research.

Fourth, there are some interesting applied implica- tions that potentially stem from the aforementioned findings. For example, the findings imply that the net- work of individuals on social networking Web sites will contain a relatively high percentage of narcissistic indi- viduals. Specifically, because narcissists have more social contacts on Facebook than the nonnarcissists, the average user will experience a social network that over- represents narcissists. This would be important infor- mation for social networking site users to be aware of. It also raises the possibility that—because of elevated exposure to narcissistic individuals and self-promotion— norms of expression on social networking sites will be pulled in the direction of greater self-promotion. Another possible application of this research pertains to individuals who are considering entering into a romantic relationship with someone they meet on a social network- ing site might be able to detect trait narcissism. Given what is known from past research about the negative tendencies of narcissists in relationships (e.g., Campbell, 1999; Campbell, Foster, et al., 2002), this would be valuable information. Though these are stimulating possi- bilities, the applied applications of this research should be read with considerable caution due to the small effect sizes that resulted in our findings.

Fifth, if narcissists use social networking sites as an outlet for maintaining their inflated, positive self-views, and if those who use social networking Web sites are exposed to a heightened level of narcissistic personality, do these Internet communities contribute to rising lev- els of narcissism over time? Based on the evidence adduced in this research, we cannot answer this question. Participating in social networking online is arguably attractive to narcissists in that it allows for controlled self-presentation, satiates the craving for attention, and promotes shallow relationships, all of which are associ- ated with narcissism. Thus, one might argue that using social networking Web sites can aid in regulating narcis- sists’ inflated positive self-views. Yet, it remains unclear whether or not social networking on the Internet results in an increase in dispositional narcissism for individuals with low narcissism scores. To appropriately address this question, we believe we need a program of converging research methods (e.g., user vs. nonuser comparisons as in Marcus and colleagues’ [2006] work; longitudinal research, i.e., examinations of the change in narcissism scores of Internet users over time; and controlled experi- mental research, e.g., manipulating Web site usage and assessing short- and long-term personality change). There is clearly a good deal of research to be done on Internet usage and personality.

Buffardi, Campbell / NARCISSISM AND WEB SITES 1311

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

Sixth and finally, we believe that this study makes an important contribution to psychological methodology. In the present investigation, our main variable of interest was narcissism, but we believe that with some modifica- tions, these methods could be employed in studying an array of individual difference variables. Key to our approach was supplementing impressions of the overall Web page made by participant raters with two types of information drawn from social networking Web pages: (a) objective profile measures (e.g., number of friends) and (b) coders’ ratings of subjective aspects made in iso- lation from the other information on the page. The objec- tive measures are currently limited in number, though arguably constantly increasing due to the innovations being made in social networking communities. However, there is an almost unlimited amount of information that can be obtained from coders’ ratings of more subjective aspects of social networking profiles. In combination, these two information extraction strategies represent a good deal of flexibility for future personality research. For example, our methods could easily be adapted to study individual differences like attachment styles (Hazan & Shaver, 1987), need for belongingness (Baumeister & Leary, 1995), rejection sensitivity (Downey & Feldman, 1996), self-monitoring (Snyder, 1987), or social anxiety (Leary, 1983). Furthermore, our method could also be viewed as a starting point for developing new ways to study broader, natural social systems as well as more general theoretical issues, such as perceiver accuracy and consensus (Kenny, 1994), self-observer agreement (Funder, 1980), and cue utilization in impression forma- tion (Brunswik, 1956). In particular, researchers inter- ested in how accurately individuals perceive information presented on Web pages would benefit from using the Brunswik (1956) approach.

Caveats

Two limitations of the present research need to be addressed. The first concerns the characteristics of our sample. Our sample was not diverse in age. The partic- ipants were all college students. Moreover, we only assessed activity and presentation on Facebook.com and have not examined the variety of other social networking sites available. Personality may potentially be presented and perceived somewhat differently on Web sites like MySpace.com as they are more customizable both in the aesthetics of the page and the available features.

Second, we examined perceptions of narcissism made by strangers but not other individuals (e.g., friends, poten- tial employers). Unlike other Web-based interpersonal contact (e.g., discussion boards, newsgroups, chatrooms), communication on social networking Web sites is for the most part not anonymous. The primary goal of interaction

on social networking Web sites is to strengthen bonds and keep in touch with friends made in the real, offline world. Past research has shown that communicating on the Internet with individuals whom one has never met before in person may be a venue for freely expressing the true self (Bargh, McKenna, & Fitzsimons, 2002; Spears & Lea, 1994). In contrast, because they are not anonymous, individuals may be motivated to maintain offline images and fulfill social roles and the obligations that come with them on social networking sites (Burke, 1980; Burke & Tully, 1977). This raises two further questions about presenting and perceiving narcissistic personality on the Internet. First, do individuals view their narcissistic friends as more narcissistic when they view their personal Web pages as compared to when they interact face to face? Second, as compared to an anonymous online environment, is it possible that nar- cissists tone down the extent to which they self-enhance and self-promote in a social networking community? Further research is required to provide answers to these remaining questions.

Conclusion

Today, interaction and self-presentation on the Internet are the norm. Maintaining a Web presence and keeping in contact with a large number of individuals via social networking Web sites has become part of many individuals’ daily lives. In the present study, our goal was to examine how trait narcissism is expressed on Web- based social networks. The results demonstrate that nar- cissists act, portray themselves, and are perceived on social networking sites in a manner similar to how they behave in real, offline life. We hope that these initial find- ings spur future inquiry into the manifestations of narcis- sism and other personality variables on social networking sites and other locations on the Internet.

NOTES

1. Women with personal Web sites reported higher narcissism scores than those without personal Web sites; men with and without Web sites had similar narcissism scores (and the men’s narcissism scores were similar to the women Web site owners’ narcissism scores).

2. Narcissism, as we are examining the construct, refers to a con- tinuous, nontaxonic normal personality variable (Foster & Campbell, 2007). This is distinct from narcissistic personality disorder (NPD) described in the Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1994). Nevertheless, trait narcissism, as we operationalize it with the Narcissistic Personality Inventory (NPI), does correlate significantly with clini- cians’ and clinical researchers’ prototypes of NPD (Miller & Campbell, 2008). We use the term narcissists in some instances as a shorthand for “high narcissism scores.” This is done in the interest of improved readability.

3. Originally, we included the number of groups (standardized) in the quantity of social interaction composite. Including this mea- sure, however, created a low level of internal consistency (α = .54).

1312 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

Therefore, we elected to drop the number of groups from the com- posite. We should report, however, that when number of groups was included, the results of the analyses performed on it remained the same. Quantity of social interaction correlated with narcissism, r(127) = .19, p = .03, and with narcissistic impression, r(126) = .25, p < .01.

4. These objective measures were not limited in range and therefore susceptible to positive skew. We addressed the possibility of nonnor- mality by log transforming (Winer, 1971) the following measures: number of friends, number of wallposts, and quantity of information posted about self. The quantity of social interaction composite measure that was used in the main analyses could not be log transformed because it is standardized. Importantly, when the transformed scores were correlated with NPI scores, a similar pattern of results was obtained such that the number of friends and narcissism were positively correlated, r(128) = .25, p = .01, and greater quantities of wallposts were marginally related to greater NPI scores, r(127) = .15, p = .09. Furthermore, neither the magnitude of the correlation between trans- formed number of friends and NPI nor between transformed number of wallposts and NPI significantly differ from the correlation between quantity of social interaction and NPI scores, ts < 1.23, ns. Also, the log- transformed quantity of information posted about self scores were not correlated with NPI scores, r(78) = .03, ns, as in the main analyses.

5. When controlling for gender, the correlations between narcis- sism and each objective, subjective, and impression measure remained the same with one exception. That is, all significant correlations pre- sented in this section remain significant and all nonsignificant corre- lations remain nonsignificant except for the association between narcissism and self-promoting information about self. This correla- tion became nonsignificant, β = .17, t(75) = 1.03, ns. In addition to this, gender was a significant predictor in three models—quantity of social interaction, β = .28, t(124) = 3.37, p = .001; main photo attrac- tiveness, β = .21, t(125) = 2.42, p = .02; and fun pictures, β = .21, t(124) = 2.48, p = .01—such that females were higher on all three of these measures.

6. The N in this correlation is lower than in other tests because only 79 of 129 Web page owners included About Me sections on their Facebook profiles. The mean narcissism score for those participants who included About Me sections (M = 17.81, SD = 6.04) did not dif- fer significantly from the mean narcissism score of those who did not (M = 16.12, SD = 6.38), t(127) = –1.52, ns.

7. We also tested each of the four combined mediators in separate mediational models using Baron and Kenny’s (1986) recommended procedure. Although none fully mediated the link between owners’ narcissism and raters’ narcissistic impression, all four of the content features—quantity of social interaction, main photo attractiveness, main photo sexiness, and main photo self-promotion—each partially mediated this association. The Sobel (1982) tests for each of these models was also significant, with the exception of the model includ- ing main photo attractiveness, for which the Sobel test was marginally significant.

REFERENCES

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Baldwin, T., & Stroman, A. (2007, March 1). How self-esteem classes are “breeding selfish generation.” The Times London. Retrieved May 14, 2007, from http://www.timesonline.co.uk/tol/news/uk/ education/article1455519.ece

Bargh, J. A., McKenna, K. Y. A., & Fitzsimons, G. M. (2002). Can you see the real me? Activation and expression of the “true self” on the Internet. Journal of Social Issues, 58, 33-48.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator vari- able distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529.

Brown, R. P., & Zeigler-Hill, V. (2004). Narcissism and the non- equivalence of self-esteem measures: A matter of dominance? Journal of Research in Personality, 38, 585-592.

Brunell, A. B., Campbell, W. K., Smith, L., & Krusemark, E. A. (2004, February). Why do people date narcissists? A narrative study. Poster session presented at the annual meeting of the Society for Personality and Social Psychology, Austin, TX.

Brunell, A. B., Gentry, W., Campbell, W. K., & Kuhnert, K. (2006, January). Narcissism and emergent leadership. Poster session pre- sented at the annual meeting of the Society for Personality and Social Psychology, Palm Springs, CA.

Brunswik, E. (1956). Perception and the representative design of psychological experiments. Berkeley: University of California Press.

Burke, P. J. (1980). The self: Measurement requirements from an interactionist perspective. Sociometry, 43, 18-29.

Burke, P. J., & Tully, J. (1977). Reconstructing social identity. Personality and Social Psychology Bulletin, 19, 4-5.

Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, nar- cissism, self-esteem, and direct and displaced aggression: Does self- love or self-hate lead to violence? Journal of Personality and Social Psychology, 75, 219-229.

Buss, D. M., & Chiodo, L. M. (1991). Narcissistic acts in everyday life. Journal of Personality, 59, 179-215.

Campbell, W. K. (1999). Narcissism and romantic attraction. Journal of Personality and Social Psychology, 77, 1254-1270.

Campbell, W. K., Bonacci, A. M., & Shelton, J. (2004). Psychological entitlement: Interpersonal consequences and validation of a self- report measure. Journal of Personality Assessment, 83, 29-45.

Campbell, W. K., Bosson, J. K., Goheen, T. W., Lakey, C. E., & Kernis, M. H. (2007). Do narcissists dislike themselves “deep down inside”? Psychological Science, 18, 227-229.

Campbell, W. K., Brunell, A. B., & Finkel, E. J. (2006). Narcissism, interpersonal self-regulation, and romantic relationships: An agency model approach. In E. J. Finkel & K. D. Vohs (Eds.), Self and relationships: Connecting intrapersonal and interpersonal processes (pp. 57-83). New York: Guilford.

Campbell, W. K., & Foster, C. A. (2002). Narcissism and commitment in romantic relationships: An investment model analysis. Personality and Social Psychology Bulletin, 28, 484-495.

Campbell, W. K., Foster, C. A., & Finkel, E. J. (2002). Does self-love lead to love for others? A story of narcissistic game playing. Journal of Personality and Social Psychology, 83, 340-354.

Campbell, W. K., Rudich, E., & Sedikides, C. (2002). Narcissism, self- esteem, and the positivity of self-views: Two portraits of self-love. Personality and Social Psychology Bulletin, 28, 358-368.

Carroll, L. (1987). A study of narcissism, affiliation, intimacy, and power motives among students in business administration. Psychological Reports, 61, 355-358.

Downey, G., & Feldman, S. I. (1996). Implications of rejection sensi- tivity for intimate relationships. Journal of Personality and Social Psychology, 70, 1327-1343.

Efron, B. (1987). Better bootstrap confidence intervals. Journal of the American Statistical Association, 82, 171-185.

Emmons, R. A. (1984). Factor analysis and construct validity of the Narcissistic Personality Inventory. Journal of Personality Assessment, 48, 291-300.

Foster, J. D., & Campbell, W. K. (2007). Are there such things as “narcissists” in social psychology? A taxometric analysis of the Narcissistic Personality Inventory. Personality and Individual Differences, 43, 1321-1332.

Foster, J. D., Shrira, I., & Campbell, W. K. (2003, June). The trajec- tory of relationships involving narcissists and non-narcissists. Poster session presented at the annual meeting of the American Psychological Society, Atlanta, GA.

Foster, J. D., Shrira, I., & Campbell, W. K. (2006). Theoretical models of narcissism, sexuality, and relationship commitment. Journal of Social and Personal Relationships, 23, 367-386.

Funder, D. C. (1980). On seeing ourselves as others see us: Self-other agreement and discrepancy in personality ratings. Journal of Personality, 48, 474-493.

Buffardi, Campbell / NARCISSISM AND WEB SITES 1313

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

Gabriel, M. T., Critelli, J. W., & Ee, J. S. (1994). Narcissistic illusions in self-evaluations of intelligence and attractiveness. Journal of Personality, 62, 143-155.

Geist, M. (2007, May 9). Facing up to Facebook fears. BBC News. Retrieved May 10, 2007, from http://news.bbc.co.uk/2/hi/ technology/6639417.stm

Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511-524.

John, O. P., & Robins, R. W. (1994). Accuracy and bias in self-per- ception: Individual differences in self-enhancement and the role of narcissism. Journal of Personality and Social Psychology, 66, 206-219.

Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford.

Kirkpatrick, D. (2007, May 24). Facebook’s plan to hook up the world. Fortune. Retrieved June 5, 2007, from http://money.cnn .com/2007/05/24/technology/facebook.fortune/index.htm

Le, T. N. (2005). Narcissism and immature love as mediators of vertical individualism and ludic love style. Journal of Social and Personal Relationships, 22, 542-560.

Leary, M. R. (1983). Social anxiousness: The construct and its measure- ment. Journal of Personality Assessment, 47, 66-75.

Locke, L. (2007, July 17). The future of Facebook. Time. Retrieved December 13, 2007, from http://www.time.com/time/business/ article/0,8599,1644040,00.html

MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the prod- uct and resampling methods. Multivariate Behavioral Research, 39, 99-128.

Marcus, B., Machilek, F., & Schütz, A. (2006). Personality in cyberspace: Personal websites as media for personality expressions and impres- sions. Journal of Personality and Social Psychology, 90, 1014-1031.

Miller, J. D., & Campbell, W. K. (2008). Comparing clinical and social-personality conceptualizations of narcissism. Journal of Personality, 76, 449-476.

Miller, J. D., Campbell, W. K., & Pilkonis, P. A. (2007). Narcissistic personality disorder: Relations with distress and functional impairment. Comprehensive Psychiatry, 48, 170-177.

Morf, C. C., & Rhodewalt, F. (2001). Unraveling the paradoxes of nar- cissism: A dynamic self-regulatory processing model. Psychological Inquiry, 12, 177-196.

Orlet, C. (2007, March 2). The look-at-me generation. The American Spectator. Retrieved May 14, 2007, from http://www.spectator .org/dsp_article.asp?art_id=11093

Oltmanns, T. F., Friedman, J. N., Fiedler, E. R., & Turkheimer, E. (2004). Perceptions of people with personality disorders based on thin slices of behavior. Journal of Research in Personality, 38, 216-229.

Paulhus, D. L. (1998). Interpersonal and intrapsychic adaptiveness of trait self-enhancement: A mixed blessing? Journal of Personality and Social Psychology, 74, 1197-1208.

Paulhus, D. L., & Williams, K. (2002). The dark triad of personality: Narcissism, Machiavellianism, and psychopathy. Journal of Research in Personality, 36, 556-568.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 717-731.

Preacher, K. J., & Hayes, A. F. (2007). Asymptotic and resampling strategies for assessing and comparing indirect effects in simple and multiple mediation models. Manuscript submitted for publication.

Raskin, R., Novacek, J., & Hogan, R. (1991). Narcissism, self- esteem, and defensive self-enhancement. Journal of Personality, 59, 19-38.

Raskin, R., & Terry, H. (1988). A principal-components analysis of the Narcissistic Personality Inventory and further evidence of its con- struct validity. Journal of Personality and Social Psychology, 54, 890-902.

Robins, R. W., & John, O. P. (1997). Effects of visual perspective and narcissism on self-perception: Is seeing believing? Psychological Science, 8, 37-42.

Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, goals and understanding: an inquiry into human knowledge structures. Hillsdale, NJ: Lawrence Erlbaum.

Schmitt, D. P., & Buss, D. M. (2001). Human mate poaching: Tactics and temptations for infiltrating existing relationships. Journal of Personality and Social Psychology, 80, 894-917.

Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422-445.

Snyder, M. (1987). Public appearances as private realities: The psy- chology of self-monitoring. New York: Freeman.

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology 1982 (pp. 290-312). Washington, DC: American Sociological Association.

Spears, R., & Lea, M. (1994). Panacea or panopticon? The hidden power in computer-mediated communication. Communication Research, 21, 427-459.

Stone, B. (2007, March 20). MySpace restrictions upset some users. The New York Times. Retrieved May 10, 2007, from http://www .nytimes.com/2007/03/20/technology/20myspace.html?ei=5124& en=8e52c7873eb73159&ex=1332043200&pagewanted=all

Vaidhyanathan, S. (2006, December 28). Me, “Person of the Year”? No thanks. MSNBC.com. Retrieved May 14, 2007, from http://www .msnbc.msn.com/id/16371425/

Vazire, S., & Funder, D. C. (2006). Impulsivity and the self-defeating behavior of narcissists. Personality and Social Psychology Review, 10, 154-165.

Vazire, S., & Gosling, S. D. (2004). e-perceptions: Personality impres- sions based on personal websites. Journal of Personality and Social Psychology, 87, 123-132.

Young, S. M., & Pinsky, D. (2006). Narcissism and celebrity. Journal of Research in Personality, 40, 463-471.

Wallace, H. M., & Baumeister, R. F. (2002). The performance of nar- cissists rises and falls with perceived opportunity for glory. Journal of Personality and Social Psychology, 82, 819-834.

Winer, B. (1971). Statistical principles in experimental design. New York: McGraw-Hill.

Received August 14, 2007 Revision accepted February 19, 2008

1314 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

at FLORIDA INTERNATIONAL UNIV on June 13, 2012psp.sagepub.comDownloaded from

242 JOURNAL OF ADVERTISING RESEARCH June 2019 doi: 10.2501/Jar-2018-018

INTRODuCTION

E­cigarette use is an increasingly serious health issue. The Centers for Disease Control and Preven- tion found that 3.5 percent of U.S. adults were cur- rent e­cigarette users in 2015, and among e­cigarette users overall, 58.8 percent were also cigarette smokers (Centers for Disease Control and Preven- tion, 2016). The health implications of e­cigarettes have ignited debate between harm-reduction and abstinence-only public-health professionals (Berry, Burton, and Howlett, 2017). Harm­reduction advo- cates cite research evidence finding e­cigarettes to be efficacious cessation aids in a step­down approach from conventional tobacco (Benowitz,

Donny, and Hatsukami, 2017), but abstinence-only advocates argue that e­cigarettes, like conventional tobacco, contain cancer-causing toxins and pollut- ants and therefore are not safer tobacco substitutes (Huerta, Walker, Mullen, Johnson, and Ford, 2017). Researchers also found that e­cigarette use signifi- cantly predicted future conventional tobacco uptake (McCabe, Veliz, McCabe, and Boyd, 2017). Effective August 8, 2016, the U.S. Food and Drug Adminis- tration began regulating e­cigarettes, requiring health warnings on packages, banning free sam- ples and vending-machine sales, and restricting sales to those 18 years and older (U.S. Food and Drug Administration, 2016). No regulations were

E-Cigarette Marketing

On Social Networking Sites Effects on Attitudes, Behavioral Control,

Intention to Quit, and Self-Efficacy

JOE PHuA

university of Georgia

[email protected]

this study examined exposure to three types of e-cigarette marketing—sponsored

advertisements, brand pages, and user-created groups—on social networking sites and their

influence on health-related outcomes. Results (N = 1,016) indicated that e-cigarette users

who joined user-created groups had significantly more negative attitudes toward quitting and

lower behavioral control, intention to quit, and self-efficacy than those who were exposed to

sponsored advertisements or who followed brand pages. Exposure to two or more types of

marketing had an additive effect on health-related outcomes. Social identification, attention

to social comparison, and subjective norms also moderated between exposure to e-cigarette

marketing and key dependent measures.

• E-cigarette users who were in user-created groups had significantly more negative health-related outcomes than those who saw advertisements or followed brand pages.

• Exposure to three types of e-cigarette marketing had a significant additive effect on health-related outcomes, compared with exposure to two or fewer types.

• social identification, attention to social comparison, and e-cigarette subjective norms moderated between exposure to e-cigarette marketing and health-related outcomes.

• the results of this study also might apply to advertising of other health-related products on social networking sites, including over-the-counter pharmaceuticals and prescription drugs (e.g. opioids).

submitted November 1, 2016;

revised July 7, 2017;

accepted august 3, 2017;

published online may 14, 2018.

Text variables

Enter text variables here to define ruun- ing heads and running feet

Main running head is defined by title (not here) NB use forced line break not return to create line break E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

Enter publication month here June 2019

Enter DOI here doi: 10.2501/Jar-2018-018

Enter special feature theme here WHAT WE KNOW ABOuT Segmentation AND TARGETING

G

June 2019 JOURNAL OF ADVERTISING RESEARCH 243

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

announced for e-cigarette marketing, however, and, as such, e­cigarette brands continue to use social networking sites to advertise to consumers.

According to a 2016 U.S. Surgeon Gen- eral’s report, use of social networking sites by e-cigarette brands for market- ing is increasingly prevalent, because of the ability to reach teenagers and young adults, who are most susceptible to peer and media influence (U.S. Department of Health and Human Services, 2016). In a 2016 special report on social media success metrics, the American Marketing Asso- ciation (AMA) outlined three prominent types of social network­based marketing:

• sponsored advertisements (paid media posts advertising a brand’s products);

• brand pages (owned media posts, which let a company craft a consistent brand message, allowing users to like, follow, and comment);

• user-created groups (earned media in which users engage in electronic word of mouth and pass along user-generated content; AMA, 2016).

The current research therefore is impor- tant, because e­cigarette marketers mainly use these three types of social networking site-based marketing to engage their target audiences (i.e., sponsored advertisements, brand pages, and user-created groups). To date, no prior studies have examined whether exposure to these three types of marketing, on their own or in combina- tion with one another, can have an impact on e-cigarette attitudes and behavioral intentions.

To address this research gap, in the current investigation the author applied the elaboration likelihood model (Petty, Cacioppo, and Schumann, 1983), the the- ory of planned behavior (Ajzen, 1991), and online information-seeking strate- gies (Ramirez, Walther, Burgoon, and

Sunnafrank, 2002) to assess whether con- sumers’ exposure to these three types of social networking site­based e­cigarette marketing would exert a significant and additive effect on attitudes and behavioral intentions toward e­cigarette use. The elab- oration likelihood model and the theory of planned behavior were chosen as relevant theoretical frames for this study because of their applications to advertising-message processing and health-behavioral change, respectively.

The popularity of social networking sites in the United States increasingly has led marketers to use these sites to engage their target customers. According to a 2017 Pew Research Center report, 69 percent of all American adults 18 and older use social networking sites, with the highest usage among those 18–29 years old (86 percent) and 30–49 years old (80 percent; Pew Research Center, 2017). The most popular social networking sites include Facebook (68 percent), Instagram (28 percent), Pin- terest (26 percent), LinkedIn (25 percent), and Twitter (21 percent; Pew Research Center, 2017).

Social networking sites allow marketers to purchase sponsored advertisements— paid media that users see on their news feeds. Consumer interaction with social networking site-based advertisements has been found to influence brand preferences (Gensler, Völckner, Liu-Thompkins, and Wiertz, 2013). Marketers also use social networking site brand pages (owned media) to engage consumers through brand-related posts. These posts facili- tate liking, sharing, and commenting on messages, which is earned media (Tay- lor, Lewin, and Strutton, 2011). Members of user-created brand communities also share information and experiences based on consumption of particular brands, often in the form of user-generated con- tent and electronic word of mouth. This is also earned media, and it results in

strong brand loyalty (Kim, Sung, and Kang, 2014).

The efficacy of social networking site­ based brand promotion has led e­cigarette brands to expand such marketing efforts (Richardson, Ganz, and Vallone, 2015). As of March 2017, because of a lack of U.S. Food and Drug Administration guide- lines, social networking sites self-regulate e­cigarette­related content:

• Instagram maintains lists of banned hashtags, but none of these refer to tobacco-related content.

• Facebook prohibits promotion of tobacco-related products, except for blogs and groups affiliated with tobacco products, which are permitted if their activity does not lead directly to tobacco sales.

• Twitter prohibits tobacco and e­cigarette promotion but allows news and infor- mation about tobacco products.

Because marketers use three major types of social networking site-based market- ing to engage their target audiences— sponsored advertisements, brand pages, and user-created groups (AMA, 2016)—it therefore is important to assess empirically each type’s effect on attitudes and behav- ioral intentions toward e­cigarettes and e­cigarette brands.

LITERATuRE REVIEW

The elaboration likelihood model (Petty et al., 1983), a frequently used theoreti- cal framework in traditional advertising research, proposes two major routes by which attitude change occurs. The cen- tral route requires individuals to process messages cognitively, which leads to high message elaboration. The peripheral route involves individuals processing messages through inferential cues in advertisements, which results in less enduring attitude change.

244 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

In research applying the elaboration like- lihood model to social networking site- based marketing, advertising cues such as brand-post popularity (Chang, Yu, and Lu, 2015), celebrity-spokesperson charac- teristics (Jin and Phua, 2014), and message personalization (De Keyzer, Dens, and De Pelsmacker, 2015) were found to influ- ence brand-related outcomes. Studies also found online-advertisement avoidance to be a major factor inhibiting user engage- ment with digital advertisements (Baek and Morimoto, 2012; Edwards, Li, and Lee, 2002; Fransen, Verlegh, Kirmani, and Smit, 2015; Yeu, Yoon, Taylor, and Lee, 2013). In research on Internet-based antitobacco advertising, persuasive cues in online advertisements also strongly influenced attitudes and behavioral intentions toward smoking (Pechmann, Delucchi, Lakon, and Prochaska, 2015; Vallone et al., 2016).

An additional study tested the elabora- tion likelihood model’s relevance in the online environment by conducting a rep- lication of the original (Petty et al., 1983) study (Kerr, Schultz, Kitchen, Mulhern, and Beede, 2015). Despite some results diverging from the premise of the model, the second group concluded that their study did offer support for learning and persuasion through subconscious pro- cessing of advertising exposure in online environments. They called for advertising researchers to further explicate traditional advertising theories, such as the elabora- tion likelihood model, in online contexts such as social networking sites to better reflect online consumers.

On the basis of those recommendations (Kerr et al., 2015), the current study opera- tionalized advertising-message processing on the basis of three levels of elaboration and agency: low, medium, and high. These three levels corresponded to the three types of social networking site­based e­cigarette brand marketing examined in this article: sponsored advertisements, brand pages,

and user-created brand groups, respec- tively. When consumers are exposed only to social networking site-based sponsored advertisements, they passively process advertising messages, with a low level of elaboration and agency involved.

The next two types of e­cigarette mar- keting messages (e­cigarette brand pages and user-created groups) require greater message elaboration and agency. People following e-cigarette brand pages like, comment on, and share branded posts but might not participate actively in creat- ing posts. They therefore have a medium level of elaboration and agency. Those who join user­created e­cigarette brand groups, in contrast, more likely will par- ticipate actively in community-based activities, including information sharing, user-generated content, and electronic word of mouth. These activities require a high level of message elaboration and agency.

This study proposed that these three types of e­cigarette marketing correspond to increasing elaboration of persuasive messages: low elaboration–low agency (sponsored advertisements), medium elaboration–medium agency (brand pages), and high elaboration–high agency (user-created groups). The study proposed that user-created groups (high elabora- tion–high agency) would have the most significant impact on attitude and behav- ioral intentions toward e­cigarettes, com- pared with sponsored advertisements (low elaboration–low agency) and brand pages (medium elaboration–medium agency).

Another relevant theoretical framework for the current investigation is online

information-seeking strategies (Ram- irez et al., 2002). On the basis of previ- ous research on online communication in computer­mediated settings, the author proposed three strategies brand consum- ers use to extract social and brand informa- tion online: passive, active, and interactive. Consumers using passive strategies to acquire information from social network- ing sites merely observe persuasive mes- sages, such as advertisements, from the sidelines without personal participation in creating or passing along the mes- sages. In the current context, exposure to e­cigarette­related sponsored advertise- ments on social networking sites entails a passive information-seeking strategy, because consumers see advertisements on their news feeds without actually creating or passing along the advertisements.

Active information-seeking strategies involve indirect information gathering (e.g., following conversations of others familiar with the topic at hand) and may include a low degree of interpersonal com- munication. Following e­cigarette brand pages and liking, commenting on, and sharing brand posts therefore constitutes an active information-seeking strategy, because these activities require greater consumer involvement in propagating brand-related messages (Araujo, Nei- jens, and Vliegenthart, 2015; Van Noort, Antheunis, and Velergh, 2014). Interactive information-seeking strategies involve direct communication between consum- ers and brands, whereby greater depth of discussion, self-disclosure, reciprocity, and alteration of behavior on the basis of feedback are possible. Joining user-created

This study proposed that these three types of

e-cigarette marketing correspond to increasing

elaboration of persuasive messages.

June 2019 JOURNAL OF ADVERTISING RESEARCH 245

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

groups require the highest level of message elaboration and agency and are the most interactive online information-seeking strategy.

H1: E­cigarette users who are mem- bers of user­created e­cigarette brand groups on social network- ing sites will have significantly (a) more negative attitudes toward quitting e-cigarettes, (b) lower perceived behavio- ral control, (c) lower intention to quit, and (d) lower smoking cessation self­efficacy compared with those who saw sponsored advertisements or who follow e­cigarette brand pages on social networking sites.

In addition to exposure, online adver- tising frequency is another major factor influencing consumer brand engagement (Cheong, De Gregorio, and Kim, 2010; Schmidt and Eisend, 2015). Previous research found that when consumers are exposed multiple times to advertising, they have higher brand recall and pur- chase intention (Brettel, Reich, Gavilanes, and Flatten, 2015; Lee, Ahn, and Parks, 2015). Too many exposures, however, also leads to less effectiveness because of the advertising wear­out effect (Campbell and Keller, 2003). For familiar and well-liked brands that consumers already follow, the advertising wear­out effect can be delayed, because consumers continue to become more engaged even at higher frequencies of exposure (Schmidt and Eisend, 2015).

On social networking sites, consumers are exposed to sponsored advertisements on the basis of profile information they entered themselves. When they like brand pages or join user-created groups, they voluntarily allow themselves to be exposed to brand-related posts on their news feeds. This study therefore proposed that when

e­cigarette groups on social networking sites, sharing user-generated content, and propagating electronic word of mouth entail an interactive information-seeking strategy, because users become both con- sumers and creators of brand-related con- tent (Levy and Gvili, 2015). Applying these online information-seeking strategies (Ramirez et al., 2002), the current author proposed that interactive (user-created group) information-seeking strategies would have the most significant impact on e­cigarette attitudes and behavioral intentions, compared with active strate- gies (brand pages) and passive strategies (sponsored advertisements).

A key variable in the theory of planned behavior is behavioral control, which refers to individuals’ perceived ease or dif- ficulty with performing particular health­ related activities (Ajzen, 1991). Previous research found behavioral control to exert a strong influence on attitudes and inten- tions toward smoking (Namkoong, Nah, Record, and Vanstee, 2016). Self­efficacy is a tenet of social-cognitive theory that refers to the extent to which individuals believe they have the ability to accomplish partic- ular tasks (Bandura, 2001). Prior research found participation in social networking site-based support groups to predict smok- ing cessation self-efficacy significantly (Phua, 2013).

Applying the elaboration likelihood model and online information-seeking strategies to social networking site-based brand communications, the current author proposed that consumers who joined user- created social networking site e­cigarette groups would have significantly more neg- ative attitudes toward quitting e­cigarettes, lower behavioral control, lower intention to quit, and lower smoking cessation self- efficacy, compared with consumers who were exposed to sponsored advertisements or who followed brand pages. This predic- tion was based on the fact that user-created

e­cigarette users are exposed to two or more types of social networking site-based e­cigarette marketing, they more likely will continue to have positive brand evalu- ations, because of the high interactivity and high message elaboration and agency associated with following brand pages and joining user-created groups.

The author thus proposed that exposure to all three types of e­cigarette market- ing—sponsored advertisements, brand pages, and user-created groups—would result in significantly more negative atti- tudes toward quitting e­cigarettes, lower behavioral control, lower intention to quit, and lower smoking cessation self­efficacy, compared with exposure to two or fewer types of e­cigarette marketing. Thus:

H2: E­cigarette users who are exposed to all three types of e­cigarette marketing messages on social networking sites (i.e., saw sponsored ads, joined e-cigarette brand pages, and joined user-created e-cigarette groups) will have significantly (a) more negative attitudes toward quitting e­cigarettes, (b) lower perceived behavioral con- trol, (c) lower intention to quit, and (d) lower smoking-cessation self-efficacy compared with those who are exposed to two or fewer types of e­cigarette mar- keting messages.

“Social identity” refers to the part of a person’s self-concept deriving from membership in social groups (Tajfel and Turner, 1986). When individuals’ social identity is salient, they see themselves as interchangeable exemplars of the larger social group. This deindividuation effect results in socially identified individuals incorporating behavioral expectations of the group into the self, which guides

246 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

behavioral control, intention to quit, and self­efficacy (See Figure 1).

H3: Social identification, atten- tion to social comparison, and e-cigarette subjective norms will moderate the relationship between exposure to social net- working site-based e-cigarette marketing messages and (a) atti- tude toward quitting e­cigarettes, (b) perceived behavioral control, (c) intention to quit, and (d) self­efficacy.

METHODOLOGY

Participants

A total of 1,016 participants took part in this study. Mean age was 41.6 years (SD = 13.43). (See Table 1 for participant demographics.)

Procedure

Data for the study were collected with Qualtrics Panel, an online research- participant recruitment service. The ser- vice posted an online questionnaire to its participants to recruit a nationally

identity-relevant behaviors. Strong smoker identification has been found to lead to greater resistance toward cessation mes- sages and smoking escalation (Hertel and Mermelstein, 2012).

“Attention to social comparison” refers to the extent to which individuals are dis- positionally susceptible to reference-group influence (Lennox and Wolfe, 1984). Social comparison has been found to moderate significantly between perceived norms and health behavior (Novak and Craw- ford, 2001). “Subjective norms” (Ajzen, 1991) refers to individuals’ perceptions of normative behavior in social groups, influ- enced by the judgment of others. In previ- ous research, subjective norms have been found to affect tobacco uptake and mainte- nance significantly (Rise, Kovac, Kraft, and Moan, 2008).

On the basis of the findings of prior research, the author hypothesized that social identification as an e-cigarette user, attention to social comparison, and e­cigarette subjective norms would inter- act with exposure to social networking service-based e-cigarette marketing to affect attitude toward quitting e­cigarettes,

representative sample from across the United States. Only participants who sat- isfied the two screening criteria—active social networking site users and cur- rent e­cigarette users—were recruited by Qualtrics Panel for the study. A total of 1,016 participants completed the online questionnaire and received e-points from Qualtrics as an incentive.

Participants were asked to identify the one social networking site that they most frequently used and answer all subse- quent questions on the basis of their use of this particular site (See Table 2). The questionnaire also included an item ask- ing participants whether they had seen e-cigarette-brand sponsored advertise- ments, followed brand pages, or been members of user-created groups within the past month.

Measures

All measures in this study were drawn from previously used scales that have been validated empirically in published research. Attitude toward quitting e­cigarettes was assessed with four items (modified from Rise et al., 2008) on 7-point semantic­differential scales. Participants were asked whether quitting e­cigarettes in the next six months was “good,” “use- ful,” “pleasant,” or “comfortable” (Cron- bach’s α = .88).

Behavioral control toward quitting e­cigarettes was assessed with three items (modified from Rise et al., 2008) on 7-point Likert scales, ranging from “strongly disa- gree” to “strongly agree.” Items included “During the next six months, I can easily quit e­cigarettes if I want to” and “How much control do you have over quitting e­cigarettes during the next six months?” (Cronbach’s α = .91). Intention to quit e­cigarettes was assessed with four items (modified from Rise et al., 2008) on 7-point Likert scales. Items included “During the next six months, I intend to quit smoking

Figure 1 Conceptual model illustrating Hypotheses tested in the study (N = 1,016)

Types:

Exposure to E-Cigarette Marketing

a. Sponsored Ads

b. Brand Pages c. User-Created

Groups

H1a/H2a

H1b/H2b

H1c/H2c

H1d/H2d

H3a H3b

H3c H3d

H3a

H3b

H3c

H3d

H3a

H3b

H3c

H3d

Attitude toward Quitting E-Cigarettes

Perceived Behavioral Control

Intention to Quit E-Cigarettes

Smoking Cessation Self-Efficacy

Attention to Social Comparison

E-Cigarette Subjective Norms

Social Identification

June 2019 JOURNAL OF ADVERTISING RESEARCH 247

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

participants could refrain from smoking e­cigarettes in different situations, includ- ing “when nervous” and “when angry” (Cronbach’s α = .96). Social identifica- tion as an e­cigarette user was assessed with eight items (modified from Cam- eron, 2004) on 7-point Likert scales. Items

e­cigarettes” and “During the next six months, I will quit smoking e­cigarettes” (Cronbach’s α = .98).

Self­efficacy toward quitting e­cigarettes was assessed with a 19­item self­efficacy scale (Etter, Bergman, Humair, and Per- neger, 2000). Questions asked whether

included “I have a lot in common with e­cigarette users” and “I feel strong ties with e­cigarette users” (Cronbach’s α = .85).

Attention to social comparison was assessed with 13 items (modified from Lennox and Wolfe, 1984) on 7-point Lik- ert scales. Items included “I pay atten- tion to others’ reactions in order to avoid being out of place” and “It is important for me to fit into the group I’m with” (Cronbach’s α = .92). Subjective norms toward e­cigarettes were assessed with six items (modified from Rise et al., 2008) on 7-point Likert scales. Items included “Peo- ple who mean a lot to me think smoking e­cigarettes is unacceptable” and “Most of my close friends do not currently smoke e­cigarettes” (Cronbach’s α = .94).

The author assessed convergent and discriminant validity by calculating the average variance extracted (AVE) for each measure in the study. On the basis of previous recommendations (Fornell and Larcker, 1981), convergent validity is established when AVE for each meas- ure exceeds 0.50, whereas discriminant validity is established when AVE for each measure is greater than the squared correlation between each pair of meas- ures in the study. AVE values obtained ranged from 0.61 to 0.75, thereby estab- lishing convergent validity. Additionally, for each pair of measures in the study, the largest squared correlation was 0.36, whereas the lowest AVE previously obtained was 0.61, thereby establishing discriminant validity.

The author then assessed common method bias using Harman’s single- factor test. All items on the measures were loaded into an exploratory factor analysis, with the number of factors extracted set at 1, and the unrotated factor solution was examined (Harman, 1976). Results indi- cated that a single factor did not account for a majority of variance in the measures,

TABLE 1 demographic Characteristics of study Participants (N = 1,016) Demographic N (%)

Gender

male 461 (45.4%)

Female 555 (54.6%)

Ethnicity

Caucasian/White 833 (82.0%)

asian/asian american 41 (4.0%)

Hispanic/latino american 47 (4.6%)

Black/african american 57 (5.6%)

Native american 12 (1.2%)

mixed ethnicity 20 (2.0%)

other ethnicity 6 (0.6%)

Highest Educational Level

did not complete high school 14 (1.4%)

Completed high school 176 (17.3%)

technical/vocational school 56 (5.5%)

attended some college 242 (23.8%)

associate’s degree 128 (12.6%)

Bachelor’s degree 255 (25.1%)

master’s degree 104 (10.2%)

doctoral/professional degree 41 (4.0%)

Annual Household Income

$20,000 or less 126 (12.4%)

$20,001 to $40,000 232 (22.8%)

$40,001 to $60,000 202 (19.9%)

$60,001 to $80,000 181 (17.8%)

$80,001 to $100,000 144 (14.2%)

over $100,000 131 (12.9%)

248 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

• self­efficacy to quit, F(7, 1008) = 65.38, p < .001 (partial η2 = .081, power = 1.00).

Hypotheses 1 and 2 therefore were supported.

Levene’s tests of equality of error vari- ances were insignificant for all depend- ent measures, so the author used Scheffe post hoc tests to compare pairwise group means. Mean attitude toward quitting was highest for those not exposed to adver- tising (M = 5.18, SD = 0.77), followed by those who saw advertisements (M = 4.94, SD = 0.74), those who followed brand pages (M = 4.34, SD = 0.67), those who saw

and when all items were loaded onto a single factor, only 25.23 percent of vari- ance was accounted for. Common method bias therefore was not an issue.

RESuLTS

Attitude toward Quitting, Behavioral

Control, Intention to Quit,

And Self-Efficacy

The author conducted a one-way multi- variate analysis of variance to examine attitude toward quitting e-cigarettes, behavioral control, intention to quit, and self­efficacy, on the basis of exposure to social networking site­based e­cigarette marketing. Exposure categories were as follows (See Table 2):

• was not exposed; • saw sponsored advertisements; • followed brand pages; • followed user-created pages; • saw advertisements and followed

brand pages; • saw advertisements and followed user-

created pages; • followed brand pages and user-created

pages; • was exposed to all three.

Results revealed a significant multivari- ate main effect by exposure (Wilks’s λ= .560), F(18, 3625) = 157.91, p < .001 (par- tial η2 = .051, power = 1.00). Given the significance of the overall test, the author examined univariate analysis of variance results with p value set at <.0125 to con- trol for Type I error. Significant univari- ate main effects by exposure to e­cigarette marketing were obtained for:

• attitude toward quitting, F(7, 1008) = 64.41, p < .001 (partial η2 = .082, power = 1.00);

• behavioral control, F(7, 1008) = 38.78, p < .001 (partial η2 = .073, power = 1.00);

• intention to quit, F(7, 1008) = 48.41, p < .001 (partial η2 = .077, power = 1.00);

advertisements and followed brand pages (M = 4.02, SD = 1.07), those who joined user-created pages (M = 2.17, SD = 0.94), those who saw advertisements and joined user-created groups (M = 2.14, SD = 1.07), those who followed brand pages and joined user-created groups (M = 1.33, SD = 0.71), and those who were exposed to all three types of advertising (M = 1.31, SD = 0.46). Mean behavioral control was highest for those not exposed (M = 5.18, SD = 1.42), followed by those who saw advertisements (M = 4.63, SD = 1.36), those who followed brand pages (M = 3.82, SD = 0.87), those who saw advertisements

TABLE 2 study Participants’ (N = 1,016) social Networking site use and Exposure to E-Cigarette marketing Variable N (%)

Social Networking Site Most Frequently used

Facebook 854 (84.1%)

twitter 60 (5.9%)

instagram 36 (3.5%)

Pinterest 18 (1.8%)

Google+ 14 (1.4%)

snapchat 11 (1.1%)

linkedin 11 (1.1%)

tumblr 7 (0.7%)

other social networking site 5 (0.5%)

Exposure To E-Cigarette Marketing In Last Month

Not exposed 181 (17.8%)

advertisements only 129 (12.7%)

Brand pages only 47 (4.6%)

user-created pages only 7 (0.7%)

advertisements and brand pages 259 (25.4%)

advertisements and user-created groups 23 (2.3%)

Brand pages and user-created groups 9 (0.9%)

advertisements, brand pages, and user-created groups 361 (35.4%)

June 2019 JOURNAL OF ADVERTISING RESEARCH 249

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

brand pages (M = 3.34, SD = 0.84), those who saw advertisements and followed brand pages (M = 3.01, SD = 0.69), those who joined user-created groups (M = 2.57, SD = 1.27), those who saw advertisements and joined user-created groups (M = 2.04, SD = 0.71), those who followed brand pages and joined user-created groups (M = 1.44, SD = 0.53), and those who were exposed to all three types of advertising (M = 1.33, SD = 0.49). Mean self­efficacy was highest for those not exposed (M = 5.80, SD = 0.75), followed by those who

and followed brand pages (M = 3.51, SD = 0.72), those who joined user-created groups (M = 1.86, SD = 0.69), those who saw advertisements and joined user- created groups (M = 1.57, SD = 0.66), those who followed brand pages and joined user-created groups (M = 1.44, SD = 0.53), and those who were exposed to all three types (M = 1.37, SD = 0.48).

Mean intention to quit was highest for those not exposed (M = 4.54, SD = 0.70), fol- lowed by those who saw advertisements (M = 3.54, SD = 0.79), those who followed

saw advertisements (M = 5.27, SD = 0.79), those who followed brand pages (M = 4.38, SD = 0.80), those who saw advertise- ments and followed brand pages (M = 4.18, SD = 0.98), those who joined user- created groups (M = 2.71, SD = 1.49), those who saw advertisements and joined user- created groups (M = 2.22, SD = 0.67), those who followed brand pages and joined user-created groups (M = 1.89, SD = 0.60), and those who were exposed to all three types of advertising (M = 1.77, SD = 0.51; See Figure 2).

Figure 2 study Participants’ (N = 1,016) attitudes toward Quitting E-Cigarettes, Perceived Behavioral Control, intention to Quit, and self-Efficacy by Exposure to E-Cigarette marketing types

N ot

e xp

os ed

A ds

o nl

y

B ra

nd p

ag e

on ly

U se

r- cr

ea te

d gr

ou ps

o nl

y

A ds

a nd

br

an d

pa ge

s

A ds

a nd

u se

r- cr

ea te

d gr

ou ps

B ra

nd p

ag es

an

d us

er -c

re at

ed g

ro up

s

A ds

, br

an d

pa ge

s an

d us

er -c

re at

ed gr

ou ps

0

1

2

3

4

5

6

7

E st

im at

ed m

ar gi

na l m

ea ns

o f

at tit

ud es

t ow

ar d

qu itt

in g

ci ga

re tt

es

Attitude toward Quitting E-Cigarettes

Exposure to E-Cigarette Marketing Types

N ot

e xp

os ed

A ds

o nl

y

B ra

nd p

ag e

on ly

U se

r- cr

ea te

d gr

ou ps

o nl

y

A ds

a nd

br

an d

pa ge

s

A ds

a nd

u se

r- cr

ea te

d gr

ou ps

B ra

nd p

ag es

an

d us

er -c

re at

ed g

ro up

s

A ds

, br

an d

pa ge

s an

d us

er -c

re at

ed gr

ou ps

0

1

2

3

4

5

6

7

E st

im at

ed m

ar gi

na l m

ea ns

o f

be ha

vi or

al c

on tr

ol t

ow ar

d qu

itt in

g ci

ga re

tt es

Behavioral Control toward Quitting E-Cigarettes

Exposure to E-Cigarette Marketing Types

N ot

e xp

os ed

A ds

o nl

y

B ra

nd p

ag e

on ly

U se

r- cr

ea te

d gr

ou ps

o nl

y

A ds

a nd

br

an d

pa ge

s

A ds

a nd

u se

r- cr

ea te

d gr

ou ps

B ra

nd p

ag es

an

d us

er -c

re at

ed g

ro up

s

A ds

, br

an d

pa ge

s an

d us

er -c

re at

ed gr

ou ps

0

1

2

3

4

5

6

7

E st

im at

ed m

ar gi

na l m

ea ns

o f

in te

nt io

n to

q ui

t ci

ga re

tt es

Intention to Quit E-Cigarettes

Exposure to E-Cigarette Marketing Types

N ot

e xp

os ed

A ds

o nl

y

B ra

nd p

ag e

on ly

U se

r- cr

ea te

d gr

ou ps

o nl

y

A ds

a nd

br

an d

pa ge

s

A ds

a nd

u se

r- cr

ea te

d gr

ou ps

B ra

nd p

ag es

an

d us

er -c

re at

ed g

ro up

s

A ds

, br

an d

pa ge

s an

d us

er -c

re at

ed gr

ou ps

0

1

2

3

4

5

6

7

E st

im at

ed m

ar gi

na l m

ea ns

o f

se lf-

ef fic

ay t

o qu

it ci

ga re

tt es Self-Efficacy to Quitting E-Cigarettes

Exposure to E-Cigarette Marketing Types

250 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

networking site­based e­cigarette market- ing—sponsored advertisements, brand pages, and user-created groups—can exert a significant effect on health­related outcomes, on the basis of the elabora- tion likelihood model (Petty et al., 1983), the theory of planned behavior (Ajzen, 1991), and online information-seeking

Social Identification, Attention to Social

Comparison, and Subjective Norms

The author conducted hierarchical regres- sion analyses to test the moderation relationships proposed in Hypothesis 3. Each variable was centered, with interac- tion terms created between exposure to e­cigarette marketing and potential moder- ators, and entered into Model 2 of each set of regressions. For each potentially signifi- cant moderation effect, the author ran the PROCESS macro for SPSS software (Hayes, 2013) on the centered terms to examine the effect across 1,000 bootstrap samples.

Social identification significantly inter- acted with exposure to e­cigarette market- ing messages to influence

• behavioral control (ΔR2 = .009), ΔF(1, 1012) = 13.20, p < .001 (β= .490, 95 percent CI [.001, .207]), t(1012) = 3.40, p < .001;

• intention to quit (ΔR2 = .004), ΔF(1, 1012) = 6.62, p < .01 (β= .246, 95 percent CI [.111, .381]), t(1012) = 3.58, p < .001;

• self­efficacy (ΔR2 = .003), ΔF(1, 1012) = 4.60, p < .05 (β= .284, 95 percent CI [.130, .437]), t(1012) = 3.63, p < .001.

Attention to social comparison sig- nificantly interacted with exposure to e­cigarette marketing messages to influ- ence intention to quit (ΔR2 = .004), ΔF(1, 1012) = 6.51, p < .01 (β=.202, 95 percent CI [.093,.311]), t(1012) = 3.64, p < .001. Subjec- tive norms significantly interacted with exposure to e­cigarette marketing mes- sages to influence behavioral control (ΔR2 = .005), ΔF(1, 1012) = 8.02, p < .01 (β = −.200, 95 percent CI [−.369, −.030]), t(1012) = −2.32, p < .05 (See Figure 3).

DISCuSSION

The current study contributes to knowl- edge of effects of social networking site­based e­cigarette marketing in sev- eral ways. First, the results suggest that exposure to three different types of social

strategies (Ramirez et al., 2002). Current e­cigarette users who were members of social networking site user-created groups significantly more likely had more nega- tive attitudes toward quitting e­cigarettes, lower behavioral control, lower inten- tion to quit, and lower smoking cessation self­efficacy, compared with those who

Figure 3 Plots of significant interactions between Exposure to E-Cigarette marketing messages and moderators on key dependent measures (N = 1,016)

0

1

2

3

4

5

6

7

In te

nt io

n to

q ui

t e-

ci ga

re tt

es

Not exposed to e-cigarette marketing

Exposed to e-cigarette marketing

Low social indentification as e-cigarette user High social indentification as e-cigarette user

0

1

2

3

4

5

6

7

Pe rc

ei ve

d be

ha vi

or al

c on

tr ol

Not exposed to e-cigarette marketing

Exposed to e-cigarette marketing

Low social indentification as e-cigarette user High social indentification as e-cigarette user

0

1

2

3

4

5

6

7 S

el f-

ef fic

ac y

to q

ui t

e- ci

ga rr

et te

s

Not exposed to e-cigarette marketing

Exposed to e-cigarette marketing

Low social indentification as e-cigarette user High social indentification as e-cigarette user

0

1

2

3

4

5

6

7

In te

nt io

n to

q ui

t e-

ci ga

re tt

es

Not exposed to e-cigarette marketing

Exposed to e-cigarette marketing

Low attention to social comparison High attention to social comparison

0

1

2

3

4

5

6

7

Pe rc

ei ve

d be

ha vi

or al

c on

tr ol

Not exposed to e-cigarette marketing

Exposed to e-cigarette marketing

Strong anti-e-cigarette subjective norms Stong pro-e-cigarette subjective norms

June 2019 JOURNAL OF ADVERTISING RESEARCH 251

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

health­related outcomes. This finding sug- gests that e­cigarette users with the most points of contact with e­cigarette brands on social networking sites (i.e., exposed to all three types of e­cigarette marketing mes- sages in the past month) had the most nega- tive attitudes toward quitting e­cigarettes, lowest behavioral control, lowest intention to quit, and lowest smoking cessation self- efficacy. This result parallels studies that found that smokers who exhibited high brand loyalty toward particular cigarette brands had more negative attitudes toward quitting and lower intention to quit (Dawes, 2014). In the current study, multiple points of contact (sponsored advertisements, brand pages, user-created groups) with e-cigarette brands on social networking sites might have increased brand loyalty, in turn leading to more negative health- related outcomes.

This study also contributes to exist- ing literature by finding several mod- erators between exposure to social networking site­based e­cigarette market- ing and health-related outcomes. First, social identification as an e­cigarette user significantly moderated between expo- sure to e­cigarette marketing and behav- ioral control. When participants had been exposed to e­cigarette marketing, higher identification resulted in higher behav- ioral control, whereas if participants had not been exposed to e­cigarette market- ing, higher identification resulted in lower behavioral control.

Second, identification also signifi- cantly moderated between exposure and intention to quit. Higher identifica- tion resulted in greater intention to quit when participants had been exposed to

followed e­cigarette brand pages or saw sponsored advertisements. Because of the increased cognitive agency required to join and participate in user­created e­cigarette groups, these e­cigarette users engaged in the highest level of message elaboration, according to the elaboration likelihood model (Petty et al., 1983), which led to most enduring attitude and behavioral change toward e­cigarettes.

On the basis of online information- seeking strategies (Ramirez et al., 2002), members of user-created groups applied the most interactive strategies while par- ticipating in group activities, creating user-generated content, and spreading electronic word of mouth. Active partici- pation in user-created groups resulted in consumers developing stronger relation- ships with e­cigarette brands. Those par- ticipants therefore had the most negative attitudes toward quitting e­cigarettes, had the lowest behavioral control and intention to quit, and were the least likely to refrain from using e­cigarettes in social situations, compared with consumers who were exposed to sponsored advertisements or followed e­cigarette brand pages.

Another important finding is that expo- sure to all three types of social network- ing site­based e­cigarette marketing had a significant additive effect on dependent measures. E­cigarette users—who in the past month, had been exposed to e­cigarette brand sponsored advertisements, followed e­cigarette brand pages, and were mem- bers of user-created social networking site e­cigarette groups—were significantly more likely (than those who were exposed to two or fewer of the three types of e­cigarette marketing) to have more negative

e­cigarette marketing and lower intention to quit when they had not been exposed to e­cigarette marketing. Third, identifica- tion also significantly moderated between exposure and self­efficacy. Among those exposed to e­cigarette marketing, higher identification increased self-efficacy to quit, whereas among those not exposed, higher identification resulted in lower self­efficacy. Identification hence exerted a strong impact on health-related outcomes due to the deindividuation effect, whereby e-cigarette users use e-cigarette behav- ioral expectations of the group to guide their own e­cigarette behaviors (Tajfel and Turner, 1986).

Attention to social comparison also sig- nificantly moderated between exposure to e­cigarette marketing and intention to quit e­cigarettes. For those who were exposed to e­cigarette marketing, higher attention to social comparison resulted in greater intention to quit, whereas for those not exposed, higher attention to social com- parison resulted in lower intention to quit. This finding suggests that the degree to which one uses one’s reference groups as models for one’s own behavior (Lennox and Wolfe, 1984) can exert a strong impact on e­cigarette health­related outcomes.

This study also found e­cigarette sub- jective norms to moderate significantly between exposure and behavioral control. That is, for e­cigarette users who were exposed to e­cigarette marketing, strong pro­e­cigarette subjective norms did not change their behavioral control, but for those not exposed to e­cigarette marketing, strong pro-e-cigarette subjective norms resulted in higher behavioral control. This finding suggests that the degree to which individuals perceive e-cigarette use to be normative among their social groups exerts a strong influence on behavioral control (Rise et al., 2008).

There are some limitations to the cur- rent study that offer implications for future

Exposure to all three types of social networking

site-based e-cigarette marketing had a significant

additive effect on dependent measures.

252 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

user­created e­cigarette groups more likely had more negative health-related out- comes than those who were not exposed to e­cigarette marketing and those who were exposed to two or fewer types of social net- working site­based e­cigarette marketing. Regulators should therefore enact stricter guidelines for social networking site-based e­cigarette marketing, particularly brand pages and user-created groups, because the increased interactivity of such market- ing can result in greater consumer engage- ment with e­cigarette brands.

Because exposure to more social net- working site-based e-cigarette market- ing types can influence consumers’ attitudes and behavioral intention toward e­cigarettes, regulators also should restrict e-cigarette brands from using mislead- ing information, such as promoting e­cigarettes as cessation aids, and glamor- izing e­cigarette use in social networking site-based advertising. For advertising practitioners, it is important to work with regulators to establish guidelines for social networking site e­cigarette marketing, such as restricting access to those 18 years and older, including appropriate advertising disclosures, and self-regulating advertis- ing message contents.

The results of the current study also might apply to advertising of other health- related products on social networking sites, including over-the-counter phar- maceuticals and prescription drugs (e.g., opioids). Researchers should take this possibility into consideration when dis- cussing the broader implications of social networking site-based health-product marketing for consumer health. In par- ticular, researchers should examine fur- ther the interactivity of social networking site- based advertising, compared with other online and traditional media, to help practitioners and regulators come up with appropriate marketing plans for health- related products such as e­cigarettes.

research. First, the participants’ most fre- quently used social networking site and their exposure to e­cigarette marketing were self-reported. Future research should access actual statistical data of partici- pants’ exposure to each type of social net- working site­based e­cigarette marketing. Second, the study was cross-sectional. It is possible that the amount of time people are exposed to different e­cigarette mar- keting efforts can have a significant effect on health-related outcomes. Future stud- ies should examine longitudinal effects of exposure to social networking site-based e­cigarette marketing.

Third, the author asked participants to self­report e­cigarette brands they had been exposed to on social networking sites and did not control for prior attitudes toward these brands, which might have had a confounding effect on dependent measures. Each social networking site plat- form, in addition, might present e­cigarette brand marketing in a different way. Future studies should pretest for pre-existing brand attitudes and also account for dif- ferences in platform features.

Fourth, future research should use addi- tional data-analysis methods, such as mul- tigroup and structural equation modeling, and study designs that include experi- ments to further investigate the valid- ity of this study’s results. Future studies also should examine whether the current results apply to general social network- ing site brand messages and whether they have wider applications outside of the e­cigarette context.

This study also offers practical implica- tions for regulators and advertising prac- titioners. As the results show, exposure to all three types of social networking site- based e­cigarette marketing had a signifi- cant and additive effect on attitudes and behavioral intentions toward e­cigarettes. Individuals who saw sponsored advertise- ments, followed brand pages, and joined

LIMITATIONS AND FuTuRE RESEARCH

Overall, the current study contributes to ongoing investigations of how social net- working site­based e­cigarette marketing can influence consumers’ perceptions of e­cigarette brands. Results indicate that social networking site e­cigarette market- ing exerts a significant negative impact on health-related outcomes, depending on the type of marketing message (sponsored advertisements, brand pages, user-created groups) and its attendant influence on con- sumer agency, message elaboration (low, medium, high), and interactivity (passive, active, interactive). Future research should continue to explicate social networking site e­cigarette advertising with potential nega- tive effects on consumer health, to guide federal and state regulations regarding e­cigarette marketing. The long­term goal of these efforts is curbing and preventing e­cigarette use and uptake among vulner- able populations, such as teenagers and young adults, the main target audience of e­cigarette brands.

aBout tHE autHor

Joe phua is an associate professor in the department

of advertising and Public relations at the university

of Georgia’s Grady College of Journalism and mass

Communication. His research examines how emerging

communication technologies influence and change

consumer attitudes and behaviors with regard to

advertisements, brands, and health issues. Phua

has been published in journals such as Journal of

Advertising and Journal of Health Communication.

rEFErENCEs

Ajzen, I. “The Theory of Planned Behavior.”

Organizational Behavior and Human Decision Pro-

cesses 50, 2 (1991): 179–211.

American Marketing Association. (2016,

August). “Special Report: Social Media Success

June 2019 JOURNAL OF ADVERTISING RESEARCH 253

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs tHEarF.orG

g o v / m m w r / v o l u m e s / 6 5 / w r / m m 6 5 4 2 a 7 .

htm?s_cid=mm6542a7_e

Chang, Y. T., H. Yu, and H. P. Lu. “Persuasive

Messages, Popularity Cohesion, and Message

Diffusion in Social Media Marketing.” Journal of

Business Research 68, 4 (2015): 777–782.

Cheong, Y., F. de Gregorio, and K. Kim. “The

Power of Reach and Frequency in the Age of

Digital Advertising.” Journal of Advertising

Research 50, 4 (2010): 403–415.

Dawes, J. “Cigarette Brand Loyalty and Pur-

chase Patterns: An Examination Using US Con-

sumer Panel Data.” Journal of Business Research

67, 9 (2014): 1933–1943.

De Keyzer, F. N. Dens, and P. de Pelsmacker.

“Is This for Me? How Consumers Respond to

Personalized Advertising on Social Network

Sites.” Journal of Interactive Advertising 15, 2

(2015): 124–134.

Edwards, S. M., H. Li, and J. H. Lee. “Forced

Exposure and Psychological Reactance: Ante-

cedents and Consequences of the Perceived

Intrusiveness of Pop-Up Ads.” Journal of Adver-

tising 31, 3 (2002): 83–95.

Etter, J. F., M. M. Bergman, J. P. Humair, and

T. V. Perneger. “Development and Validation

of a Scale Measuring Self­Efficacy of Current

and Former Smokers.” Addiction 95, 6 (2000):

901–913.

Fornell, C., and D. F. Larcker. “Evaluating

Structural Equation Models with Unobservable

Variables and Measurement Error.” Journal of

Marketing Research 18, 1 (1981): 39–50.

Fransen, M. L., P. W. J. Verlegh, A. Kirmani,

and E. G. Smit. “A Typology of Consumer Strat-

egies for Resisting Advertising, and a Review of

Mechanisms for Countering Them.” Interna-

tional Journal of Advertising 34, 1 (2015): 6–16.

Metrics.” Retrieved July 1, 2017 from https://

www.ama.org/resources/Pages/social-media-

success-metrics.aspx

Araujo, T., P. Neijens, and R. Vliegenthart.

“What Motivates Consumers to Re-Tweet Brand

Content?” Journal of Advertising Research 55, 3

(2015): 284–295.

Baek, T. Y., and M. Morimoto. “Stay Away

from Me.” Journal of Advertising 41, 1 (2012):

59–76.

Bandura, A. “Social Cognitive Theory of Mass

Communication.” Media Psychology 3, 3 (2001):

265–299.

Benowitz, N. L., E. C. Donny, and D. K. Hat-

sukami. “Reduced Nicotine Content Cigarettes,

E-Cigarettes and the Cigarette End Game.”

Addiction 112, 1 (2017): 6–7.

Berry, C., S. Burton, and E. Howlett. “The

Impact of E­Cigarette Addiction Warnings and

Health-Related Claims on Consumers’ Risk

Beliefs and Use Intentions.” Journal of Public

Policy & Marketing 36, 1 (2017): 54–69.

Brettel, M., J. C. Reich, J. M. Gavilanes, and

T. C. Flatten. “What Drives Advertising Suc-

cess on Facebook? An Advertising­Effectiveness

Model.” Journal of Advertising Research 55, 2

(2015): 162–175.

Cameron, J. E. “A Three-Factor Model of Social

Identity.” Self and Identity 3, 3 (2004): 239–262.

Campbell, M. C., and K. L. Keller. “Brand

Familiarity and Advertising Repetition Effects.”

Journal of Consumer Research 30, 2 (2003): 292–304.

Centers for Disease Control and Preven-

tion. (2016, October 28). “Cigarette Smok-

ing Status among Current Adult E­Cigarette

Users by Age Group—National Health Inter-

view Survey, United States, 2015.” Retrieved

November 1, 2016, from https://www.cdc.

Gensler, S., F. Völckner, Y. Liu­Thompkins,

and C. Wiertz. “Managing Brands in the Social

Media Environment.” Journal of Interactive Mar-

keting 27, 4 (2013): 242–256.

Harman, H. H. Modern Factor Analysis. Chi-

cago: University of Chicago Press, 1976.

Hayes, A. F. Introduction to Mediation, Mod-

eration, and Conditional Process Analysis: A

Regression-Based Approach. New York: Guilford

Press, 2013.

Hertel, A. W., and R. J. Mermelstein. (2012).

“Smoker Identity and Smoking Escalation

among Adolescents.” Health Psychology 31, 4

(2012): 467–475.

Huerta, T. R., D. M. Walker, D. Mullen,

T. J. Johnson, and E. W. Ford. “Trends in

E­Cigarette Awareness and Perceived Harmful-

ness in the US.” American Journal of Preventive

Medicine 52, 3 (2017): 339–346.

Jin, S. A. A., and J. Phua. “Following Celeb-

rities’ Tweets about Brands: The Impact of

Twitter-Based Electronic Word-of-Mouth on

Consumers’ Source Credibility Perception,

Buying Intention, and Social Identification with

Celebrities.” Journal of Advertising 43, 2 (2014):

181–195.

Kerr, G., D. E. Schultz, P. J. Kitchen, F. J. Mul-

hern, and P. Beede. “Does Traditional Adver-

tising Theory Apply to the Digital World? A

Replication Analysis.” Journal of Advertising

Research 55, 4 (2015): 390–400.

Kim, E., Y. Sung, and H. Kang. “Brand Fol-

lowers’ ReTweeting Behavior on Twitter: How

Brand Relationships Influence Brand Electronic

Word-of-Mouth. Computers in Human Behavior

37 (2014): 18–25.

Lee, J. W., J. H. Ahn, and B. H. Park. “The Effect

of Repetition in Internet Banner Ads and the

254 JOURNAL OF ADVERTISING RESEARCH June 2019

E-CiGarEttE markEtiNG oN soCial NEtWorkiNG sitEs

Petty, R. E., J. T. Cacioppo, and D. Schumann.

“Central and Peripheral Routes to Advertising

Effectiveness: The Moderating Role of Involve-

ment.” Journal of Consumer Research 10, 2 (1983):

135–146.

Pew Research Center. (2017, January 12).

“Social Media Fact Sheet.” Retrieved July

1, 2017, from http://www.pewinternet.org/

fact-sheet/social-media/

Phua, J. “Participating in Health Issue­Specific

Social Networking Sites to Quit Smoking: How

Does Online Social Interconnectedness Influ-

ence Smoking Cessation Self­Efficacy?” Journal

of Communication 63, 5 (2013): 933–952.

Ramirez, A., Jr., J. B. Walther, J. K. Burgoon,

and M. Sunnafrank. “Information-Seeking

Strategies, Uncertainty, and Computer-

Mediated Communication: Toward a Concep-

tual Model.” Human Communication Research 28,

2 (2002): 213–228.

Richardson, A., O. Ganz, and D. Vallone.

“Tobacco on the Web: Surveillance and Char-

acterization of Online Tobacco and E­Cigarette

Advertising.” Tobacco Control 24, 4 (2015):

341–347.

Rise, J., V. Kovac, P. Kraft, and I. S. Moan.

“Predicting the Intention to Quit Smoking and

Quitting Behavior: Extending Theory of Planned

Behavior.” British Journal of Health Psychology 13,

2 (2008): 291–310.

Schmidt, S., and M. Eisend. “Advertising Rep-

etition: A Meta­Analysis on Effective Frequency

in Advertising.” Journal of Advertising 44, 4

(2015): 415–428.

Moderating Role of Animation.” Computers in

Human Behavior 46 (2015): 202–209.

Lennox, R. D., and R. N. Wolfe. “Revision of

the Self-Monitoring Scale.” Journal of Personality

and Social Psychology 46, 6 (1984): 1349–1364.

Levy, S., and Y. Gvili. “How Credible Is

E-Word of Mouth across Digital-Marketing

Channels?” Journal of Advertising Research 55, 1

(2015): 95–109.

McCabe, S. E., P. Veliz, V. V. McCabe, and C.

J. Boyd. “Smoking Behaviors and Intentions

among Current E-Cigarette Users, Cigarette

Smokers, and Dual Users: A National Survey of

US High School Seniors.” Preventive Medicine 99

(2017): 228–235.

Namkoong, K., S. Nah, R. A. Record, and S.

K. Vanstee. “Communication, Reasoning,

and Planned Behaviors: Unveiling the Effect of

Interactive Communication in an Anti-Smoking

Social Media Campaign.” Health Communication

32, 1 (2016): 41–50.

Novak, K. B., and L. A. Crawford. “Perceived

Drinking Norms, Attention to Social Com-

parison Information, and Alcohol Use among

College Students.” Journal of Alcohol and Drug

Education 46, 3 (2001): 18–33.

Pechmann, C., L. Pan, K. Delucchi, C. M.

Lakon, and J. J. Prochaska. “Development of

a Twitter­Based Intervention for Smoking Ces-

sation That Encourages High-Quality Social

Media Interactions via Automessages.” Journal

of Medical Internet Research 17, 2 (2015): e50.

Tajfel, H., and J. C. Turner. “The Social Iden- tity Theory of Intergroup Behavior.” In Social Comparison Theories: Key Readings in Social Psy- chology, D. A. Stapel and H. Blanton, eds. New York: Taylor and Francis, 1986.

Taylor, D. G., J. E. Lewin, and D. Strutton. “Friends, Fans, and Followers: Do Ads Work on Social Networks?” Journal of Advertising Research 51, 1 (2011): 258–275.

U.S. Department of Health and Human Ser- vices. (2016, August). “E­Cigarette Use among Youth and Young Adults: A Report of the Sur- geon General.” Retrieved March 1, 2018, from https://e­cigarettes.surgeongeneral.gov/docu- ments/2016_sgr_full_report_non-508.pdf

U.S. Food and Drug Administration. (2016, June). “The Facts on the FDA’s New Tobacco Rule.” Retrieved November 1, 2016, from http:// www.fda.gov/downloads/ForConsumers/Con- sumerUpdates/UCM507132.pdf

Vallone, D., A. Smith, T. Kenney, M. Green- berg, et al. “Agents of Social Change: A Model for Targeting and Engaging Generation Z across Platforms.” Journal of Advertising Research 56, 4 (2016): 414–425.

Van Noort, G., M. L. Antheunis, and P. W. J. Velergh. “Enhancing the Effects of Social Net- work Site Marketing Campaigns: If You Want Consumers to Like You, Ask Them about Them- selves.” International Journal of Advertising 33, 2 (2014): 235–252.

Yeu, M. S., H. S. Yoon, C. R. Taylor, and D. H. Lee. “Are Banner Advertisements in Online Games Effective?” Journal of Advertising 42, 2–3 (2013): 241–250.

Copyright of Journal of Advertising Research is the property of Warc LTD 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.