deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and...

12
Deterring illegal downloading: The effects of threat appeals, past behavior, subjective norms, and attributions of harm y Aron M. Levin 1 * , Mary Conway Dato-on 1 and Chris Manolis 2 1 Northern Kentucky University, Nunn Drive, Highland Heights, KY 41099, USA 2 Xavier University, USA The study employs two experiments to examine the effectiveness of various strategies used to dissuade consumers from downloading music illegally. The research investigates two specific strategies that the recording industry has used: (1) fear or threat appeals (e.g., the threat of punishment, such as fines and/or jail time), and (2) attribution of harm (informing consumers of the harm caused by the illegal downloading of music, such as financial loss to either the artist or the recording company). The study also considers whether past illegal downloading behavior reduces the effectiveness of these disincentive strategies. Finally, the impact of subjective norms (i.e., whether subjects think their friends would approve of downloading music) was also investigated. A 3 (level of threat: low, moderate, or high) X 2 (who is harmed by illegal downloading: artist or recording company) experimental design was employed for study one. Under- graduate students (n ¼ 388) participated in the study. Study two expanded on the design of the first study by adding a variable of subjective norms and by including previous downloading behavior in the model. Undergraduate students (n ¼ 211) also participated in the second experiment. Findings indicate a significant effect of threat appeal such that stronger threat appeals were found to be more effective than weaker threat appeals in reducing illegal down- loading. The first study also showed that prior illegal downloading behavior does not curtail the effects of threat appeals aimed at reducing illegal downloading. In addition, results reveal no differences in downloading behavior in terms of attribution of harm deterrent strategy (harm to either the recording artist or company). The most interesting finding from the second study is that subjective norms appear to equalize low versus high past downloaders, but only under conditions of weak fear. The current manuscript is the first to examine the impact of four different variables (threat appeals, attribution of harm, subjective norms, and previous downloading Journal of Consumer Behaviour J. Consumer Behav. 6: 111–122 (2007) Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/cb.211 *Correspondence to: Aron M. Levin, Northern Kentucky University, Nunn Drive, Highland Heights, KY 41099 E-mail: [email protected] y Aron M. Levin is Associate Professor of Marketing, Mary Conway Dato-on is Associate Professor of Marketing, and Chris Manolis is Associate Professor of Marketing. Copyright # 2007 John Wiley & Sons, Ltd. Journal of Consumer Behaviour, Mar.–Jun. 2007 DOI: 10.1002/cb

Upload: aron-m-levin

Post on 06-Jun-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Journal of Consumer BehaviourJ. Consumer Behav. 6: 111–122 (2007)Published online in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/cb.211

Deterring illegal downloading:The effects of threat appeals,past behavior, subjective norms,and attributions of harmy

Aron M. Levin1*, Mary Conway Dato-on1 and Chris Manolis2

1Northern Kentucky University, Nunn Drive, Highland Heights, KY 41099, USA

2Xavier University, USA

� T

*CorE-mayAroChri

Cop

he study employs two experiments to examine the effectiveness of various strategies used

to dissuade consumers from downloading music illegally. The research investigates two

specific strategies that the recording industry has used: (1) fear or threat appeals (e.g., the

threat of punishment, such as fines and/or jail time), and (2) attribution of harm

(informing consumers of the harm caused by the illegal downloading of music, such as

financial loss to either the artist or the recording company). The study also considers

whether past illegal downloading behavior reduces the effectiveness of these disincentive

strategies. Finally, the impact of subjective norms (i.e., whether subjects think their friends

would approve of downloading music) was also investigated.

� A

3 (level of threat: low, moderate, or high) X 2 (who is harmed by illegal downloading:

artist or recording company) experimental design was employed for study one. Under-

graduate students (n¼388) participated in the study. Study two expanded on the design

of the first study by adding a variable of subjective norms and by including previous

downloading behavior in the model. Undergraduate students (n¼ 211) also participated

in the second experiment.

� F

indings indicate a significant effect of threat appeal such that stronger threat appeals

were found to be more effective than weaker threat appeals in reducing illegal down-

loading. The first study also showed that prior illegal downloading behavior does not

curtail the effects of threat appeals aimed at reducing illegal downloading. In addition,

results reveal no differences in downloading behavior in terms of attribution of harm

deterrent strategy (harm to either the recording artist or company). The most interesting

finding from the second study is that subjective norms appear to equalize low versus high

past downloaders, but only under conditions of weak fear.

� T

he current manuscript is the first to examine the impact of four different variables

(threat appeals, attribution of harm, subjective norms, and previous downloading

respondence to: Aron M. Levin, Northern Kentucky University, Nunn Drive, Highland Heights, KY 41099il: [email protected] M. Levin is Associate Professor of Marketing, Mary Conway Dato-on is Associate Professor of Marketing, ands Manolis is Associate Professor of Marketing.

yright # 2007 John Wiley & Sons, Ltd. Journal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 2: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Cop

112 Aron M. Levin et al.

behavior) on subjects’ likelihood to illegally download music in the future. In particular, this

research illuminates the potential importance of social norms in discouraging a type of

undesirable consumer behavior but shows that this occurs only under a restricted set of

conditions: when threat is lowand the consumer is not a habitually high downloader. It should

be of interest to those in fields where intellectual property can be pirated on the Internet.

Copyright # 2007 John Wiley & Sons, Ltd.

Introduction

According to the Recording Industry ofAmerica (RIAA), the recording industryloses approximately $4.2 billion per year onpiracy, which can take the form of illegalrecording of concerts, producing andselling counterfeit compact disks, and enga-ging in online piracy (riaa.org). Online piracyrefers to the free exchange of MP3 files onwebsites such as Kazaa or Grokster in whichno royalties are paid to recording artists orcompanies. The RIAA has taken many steps tostop the illegal downloading of music, such ascreating public relations campaigns thatattempt to educate and inform consumers thatthis activity harms recording artists and/ormusic companies, and filing well-publicizedlawsuits wherein violators of online piracylaws are subject to fines and potential jail time.Information on the RIAA’s website indicatesthat since September 2003, more than 3,500illegal online music sharers (downloaders)have been sued.

Legal actions initiated by the music industryhave proven successful in shutting down someof the most well-known file sharing websitessuch as Napster and AudioGalaxy. The overalleffectiveness of the industry’s efforts to curtailonline piracy is questionable, however. A studyconducted in 2004 indicated that 8.3 millionpeople were illegally downloading music fromsites such as Kazaa; this represents a 19 percent increase from the preceding year (Gra-ham, USA Today 2004). Illegal downloading isnot limited to music. Movie piracy is alsogrowing with pirated versions normally widelyavailable (McBride, 2006). A recent studyconducted by online researcher OTX indicatedthat 25 per cent of people online have

yright # 2007 John Wiley & Sons, Ltd. Jo

downloaded a full-length feature film (cnet.-com 2004).

The current study includes two experimentswhich examine the effectiveness of variousdisincentives used to dissuade consumers fromdownloading music illegally. The first experi-ment examines the effectiveness of threatappeals and attributions of harm on dissuadingconsumers from illegally downloading. Thesecond experiment augments study one bymeasuring the effect of subjective norms, asidentified in the Bentler and Speckart’s (1979)extension of the Fishbein-Ajzen (1975) model ofreasoned action, as a potential motivating factorfor engaging in illegal music downloading. Thisresearch contributes to the literature bycombining two well researched theoreticalconstructs (threat appeals and subjectivenorms) to understand why consumers engagein illegal behavior and what marketers may doto redirect the behavior. The current manu-script is also the first to examine the impact offour different variables (threat appeals, attribu-tion of harm, previous downloading behavior,and subjective norms) on subjects’ likelihood toillegally download music in the future. Whilethe manuscript focuses on the context of musicdownloading, the theoretical and practicalcontributions should be of interest to those inany field where intellectual property can bepirated via the Internet.

The paper begins with a brief literaturereview of previous work on illegal musicdownloading. This is followed by theoreticaldiscussions regarding threat appeals, attribu-tions of harm, and prior behavior as influencersof future behavioral intentions. Hypotheses,methodology, and results for the first experi-ment follow. Next, the rationale for study twoas it relates to subjective norms and methods of

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 3: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Deterring illegal downloading 113

accounting for prior behavior are discussedalong with results. The manuscript concludeswith a general discussion, managerial implica-tions, and areas for future research.

Toward an understanding of illegal

downloading

Due to increasing bandwidth and a prolifer-ation of illegal file-sharing products andservices (e.g., Napster), online piracy hasflourished. Since its inception, illegal musicfile-sharing has been viewed as an easy way formusic fans to obtain free products, and,conversely, as an act of thievery that hasdramatically stunted sales and profits forrecord companies and artists.

A recent study by Levin, Conway Dato-on, andRhee (2004) found that ethical attitudes andopportunities to download were significantlyand independently predictive of whether or notcollege students engaged in the illegal down-loading of music. In other words, the illegaldownloading of music is not merely a matter of‘everybody is doing it’. While it is true thatstudents with high-speed Internet access weremore likely to download illegally, regardless ofconnection speed, illegal downloaders weresignificantly different in ethical profile (asmeasured by Muncy and Vitell’s 1994 ethicalprofile scale) from those who did not illegallydownload music. Just as consumers differ withrespect to ethical standards and attitudesregarding downloading music without properpayment, they likely react differently to tacticsused to deter them from engaging in suchbehavior. The first such tactic that we examineis the use of threat appeals.

Threat appeals and how theywork

Threat appeals are persuasive messages thatarouse fear. As noted above, record companieshave used these appeals to curb illegal down-loading of music. After nearly half a century ofresearch on what constitutes threat appeals, aswell as how and why these messages work, theliterature identifies three principal indepen-

Copyright # 2007 John Wiley & Sons, Ltd. J

dent variables: fear, perceived threat, andperceived efficacy (Witte and Allen, 2000).While fear is an emotion, perceived threat is acognition that can potentially lead a person toexperience fear. Perceived efficacy is a two-dimensional variable that guides an individual’sresponse to a threat and subsequent fear (Witteand Allen, 2000).

Recently, Witte (1992, 1998) integratedthree categories of threat appeals theory intoone: the extended parallel process model(EPPM). EPPM explains both successes andfailures of threat appeals and suggests thatperceived threats establish the scope ofresponse to a threat appeal and perceived

efficacy determines the character of theresponse. If, for instance, a threat is perceivedto be relevant or serious, one may decide tocontrol either the danger associated with thethreat (e.g., ‘I will stop smoking to avoidcancer’), or the fear of the danger when onedoes not perceive controlling the behavior tobe a viable alternative (e.g., ‘I cannot stopsmoking and will therefore tell myself that noteveryone who smokes gets cancer’) (Witte andAllen 2000). In the context of the currentstudy, we might expect downloaders’ mental-ity to be either a) ‘I will stop downloading toavoid prosecution’ or b) ‘I will continuedownloading and convince myself that mostpeople who download do not get caught’.

Another body of literature relevant to thecurrent setting is Fishbein and Ajzen’s (1975)theory of reasoned action. Fishbein and Ajzensuggest that attitudes, along with social norms,lead to behavioral intentions. Feelings of fearmay approximate an attitude directed towardthe act of downloading. Accordingly, the effectsof these feelings and the attached beliefs thatone will be punished for illegally downloadingmusic combine to form an attitude, which inturn predicts, to some degree, one’s futureintention to illegally download music.

In a subsequent update of Fishbein andAjzen’s theory, Bentler and Speckart (1979)suggest that prior behavior is an equally (equalto attitudes) important determinant of futureintentions. In the domain of threat appeals,numerous researchers suggest past behavior is

ournal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 4: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

114 Aron M. Levin et al.

particularly important to predicting futureintentions (Sutton and Eiser 1990; Umeh,2004; Witte, 1992, 1998). In accordance withthe two literature streams (i.e., attitudes andthreat appeals), we investigate the effects ofprior behavior—prior illegal downloading ofmusic specifically—in conjunction to threatappeals. By controlling for one’s prior down-loading behavior, we hope to more clearlyisolate the effects of fear appeals.

Consumer ethics literature (Fullerton, Kerchand Dodge 1996; Muncy and Vitell 1992)suggests the attribution of harm influenceswhether consumers engage in illicit behavior.Specifically, a consumer’s ambivalence towardunethical behavior becomes more acceptablewhen the seller suffers little or no economicharm from the consumer’s action (Fullertonet al., 1996). In the present context, con-sumers who believe illegally downloadingmusic harms companies and/or artists maybe less likely to engage in the behavior.

In the next section we develop hypothesesbased on the theories and literature discussedabove. We consider online piracy and test thecomparative effectiveness of two deterrentstrategies commonly implemented by themusic industry (i.e., threat appeals and attribu-tions of harm) while controlling for priorbehavior.

Hypotheses

LaTour and Rotfeld (1997) distinguishbetween threat and fear. According to theauthors, threat is ‘an appeal to fear, acommunication stimulus that attempts toevoke a fear response by showing some typeof outcome that the audience wants to avoid’(LaTour and Rotfeld, 1997, pg. 46). Fear, on theother hand, is an emotional response to aperceived threat. Fear can drive a change inattitude and/or behavioral intentions. In thisstudy we manipulate threat levels (low,moderate, and high) by way of a messageaimed at stopping the illegal downloading ofmusic. The message also indicates who isharmed by such unlawful activity (recordingcompany versus artist).

Copyright # 2007 John Wiley & Sons, Ltd. Jo

A consensus of contemporary theorists andmeta-analytical results (e.g., Boster and Mon-geau, 1984; Mongeau, 1989; Witte and Allen,2000) conclude that a positive, monotonicrelationship exists between threat levels(strength of a threat) and attitude/behavioralchange intentions. That is, stronger versusweaker threats are more effective in changingfuture intentions provided that any recom-mended action is considered doable on behalfof those being targeted by the appeal. In otherwords, individuals must have the capacity orpossess the personal efficacy to performrecommended actions or behavior in the faceof a threat appeal. We offer the followinghypothesis:

H1: Increasing threat levels (from weaker

to stronger) in appeals aimed at curbing

the illegal downloading of music will

decrease intentions to (a) visit websites

where downloading music illegally is

possible and (b) download music illegally

from such sites.

Researchers acknowledge that response tothreats is situational, topic-relevant, andpersonal (Burnett and Oliver, 1979). Currently,we investigate online piracy in the form ofdownloading music and consider how individ-ual differences in terms of previous behavior(i.e., extent of prior illegal downloading)predict responses to various threat levels.

Threat appeals are often used without anyrecommended response when the attitude,behavior, or action that is to be avoided isinherently obvious. Furthermore, threatappeals often exclude any mention of theinherent ease by which one might avoid suchactions (i.e., the efficacy of a recommendedresponse). With respect to road safety, forexample, a public service announcementmight only show the disastrous effects ofdriving an automobile while under the influ-ence of drugs or alcohol and not make specificreference to the fact that people shouldconsider the issue as serious and take appro-priate action (cf. Rossiter and Thornton, 2004).When a recommended response, or the efficacy

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 5: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

1The testing of this hypothesis is necessarily dependentupon a significant finding for H3a.

Deterring illegal downloading 115

associated with such a response, is notspecifically provided in a threat appeal, researchindicates that individuals will rely on pastexperiences (i.e., prior behavior) and/or currentbeliefs or attitudes (i.e., the chances of beingcaught engaging in undesirable behavior) todetermine perceived efficacy of the desiredresponse (Witte, 1992, 1998). Sutton and Eiser(1990), for example, found that self-reportedfrequency of seat belt use was influenced moreby previous use frequency than by intentionsindicated following exposure to threat appeals.Additionally, Umeh’s (2004) study on condomuse to prevent sexually transmitted diseasesrevealed threat and previous behavior asindependent predictors of intended condomuse with previous behavior explaining variancebeyond self efficacy, response efficacy, andseverity variables.

Based on the above research and the ensuingreasoning, we predict that the more one hasdownloaded music illegally in the past, the less

one will be persuaded to reduce this type ofbehavior in the future due to threat appeals. Incontrolling for past illegal downloading beha-vior, we should see the effect of threat appealsdiminish. We offer the following hypothesis:

H2: When controlling for past download-

ing behavior, the effect of threat appeals on

one’s intentions to both visit websites

where illegally downloading music is

possible and their likelihood to download

music at these sites should diminish.

In addition to the use of threat appeals tocurb the illegal downloading of music, therecording industry also tried to attribute harmdue to this form of unlawful behavior.Specifically, record companies promoted thenotion that illegally downloading music finan-cially hurts the company, the artist on contractwith the company, or both.

People often rationalize the decision todownload music illegally by convincing them-selves that recording companies and artistshave over-charged them for music in the pastand that these companies and/or artists are notharmed financially by illegal downloading

Copyright # 2007 John Wiley & Sons, Ltd. J

(Levin et al., 2004). Just as feelings of fearcombine to form attitudes toward the act ofillegal downloading, attributing harm may alsobe based on attitudes toward potential victimsof the illegal act. Thus, if consumers holdpositive attitudes toward companies (orartists), they would be less likely to engagein illegal downloading.

Levin et al. (2004) found that subjects heldmore positive attitudes toward artists thanmusic companies and that these attitudesinfluenced ethical perceptions toward down-loading music. Perhaps this difference inattitudes stems from the fact that consumersmay see the companies as being only inter-ested in profits, whereas the artists are strivingto create an artistic product for their fans toenjoy. Thus in the current study we expectconsumers to have more positive attitudestoward recording artists than music companiesand anticipate that in cases where recordingartists are financially harmed, respondents willreduce behavioral intentions related to illegallydownloading music. More specifically, wepredict a main effect for attribution ofharm—such that the most (least) effectivedeterrent will take place when a recordingartist (company) is said to be financiallyharmed amid a strong (weak) threat appeal.Furthermore, this effect is likely tempered byone’s prior illegal downloading activity. Weoffer the following hypotheses:

H3a: The effect of threat appeals on

intentions to visit websites where down-

loading music illegally is possible and to

download music illegally at these sites will

be more pronounced when a recording

artist versus a company is said to be

financially harmed.

H3b: When controlling for past down-

loading behavior, the effect noted in H3a

will subside. That is, heavy downloaders

will be less susceptible to this appeal than

will light downloaders.1

ournal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 6: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

116 Aron M. Levin et al.

Study one

Methodology

Three-hundred and eighty-eight completedsurveys were collected in undergraduatebusiness classes at a medium-sized southeast-ern university. Utilizing a 3 (level of threat:weak, moderate, or strong) X 2 (who is harmedby illegal downloading: artist or recordingcompany) full-factorial design, subjects wererandomly assigned to one of six experimentalconditions. In their classes, the students readone of six scenarios representing the con-ditions above followed by a series of questions(see below).

Independent variables

In order to create a three-tier (i.e., weak,moderate, and strong) threat manipulation, apretest was administered to 137 undergradu-ate students (this sample was independent ofthe ‘main sample’ described above). A series ofnine punishments or outcomes, each corre-sponding to a different consequence ofillegally downloading music (and gettingcaught), was presented to the subjects (e.g.,having to pay a fine). Each outcome had a10-point scale (1¼ ‘lowest fear’; 10¼ ‘highestfear’). The results of the pretest and thesubsequent threat manipulations were asfollows: ‘You will be required to delete allmusic files from your computer’ received amean score of 2.45 (SD¼ 1.97) and was usedfor the weak threat condition (N¼ 114); ‘Youwill have to perform community service of upto 25 hours’ had a mean of 5.45 (SD¼ 2.65)and was used for the moderate threat

condition (N¼ 112); and, ‘For each song,you will have to pay a fine of $2,500’ had amean of 9.01 (SD¼ 2.07) and was used for thestrong threat condition (N¼ 111). Manipula-tion checks in the main study confirmed thatthese consequences were indeed perceived asweak, moderate, and strong levels of threat,respectively.

The attribution of who is harmed by illegallydownloading music was manipulated by the

Copyright # 2007 John Wiley & Sons, Ltd. Jo

following two phrases: ‘for each song that anindividual downloads without paying, onedollar is removed from the music company’s

(the recording artist’s) profit’. Finally, a scalemeasured whether or not subjects had down-loaded music—either legally or illegally—inthe past, and, if so, to what extent (i.e., howmany songs had they downloaded, etc.). Amajority of subjects in the study (77.3 per cent)had downloaded music without paying (illeg-ally) in the past. Comparatively, a much smallerper centage (11.9%) had downloaded musicfrom ‘legitimate’ or pay sites (legally). Of thosewho downloaded music illegally, the rangewas from one to 8,000 songs with a mean of373.31 (SD¼ 837.57) and a median of 100.

Dependent variables

The dependent variables consisted of, one,how likely subjects are to visit a ‘free’ musicwebsite where downloading music illegally ispossible, and, two, how likely subjects are todownload music illegally from a ‘free’ websitegiven the particular manipulation they encoun-tered. Both the ‘visit’ and ‘download’ measureswere comprised of two items each, and all ofthe items followed a four-point responseformat (1¼not at all likely and 4¼ very likely;lower scores are indicative of a greaterdeterring effect or a decreased likelihood ofeither visiting or downloading). Cronbach’salpha for the visit and download compositevariables were 0.71 and 0.73, respectively.

Results

To test the hypotheses noted above, a seriesof analyses was conducted. First, a 3� 2 (threatX harm attribution) analysis of variance(ANOVA) revealed a significant main effectof threat appeal for both the visit, F(2,330)¼ 5.02, p< 0.01, and download, F(2,330)¼ 4.40, p< 0.01, dependent measures.Using tests of simple effects (t’s), we deter-mined that the strong threat condition com-pared with the weak threat condition wassignificantly different across both the visit,

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 7: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Deterring illegal downloading 117

t(330)¼ 3.24, p< 0.01, and download,t(330)¼ 2.89, p< 0.01, variables (M’s¼ 2.37versus 2.73 and 2.38 versus 2.72, respectively).Thus, strong threats resulted in lower inten-tions to download in the future than did weakthreats. Hypothesis 1 is supported. Neither themain effect for harm attribution nor theinteraction between threat appeal and harmattribution was significant for either the visit ordownload measures (F’s< 1.0). Hypothesis 3ais not supported. And, because H3b iscontingent upon a significant finding forH3a, we did not test H3b.

To test H2, we estimated a three-way(threat) analysis of covariance (ANCOVA)model controlling for the number of songsdownloaded unlawfully in the past (a continu-ous measure). First, we established thatintended visits and downloads following threatappeals (dependent variables) were related topast illegal downloading behavior (covariate)by collapsing the two dependent variablesacross the threat appeal conditions andcorrelating them with the number of songsdownloaded previously and illegally (r’s¼0.334 and 0.199 for intended visits anddownloads, respectively, p’s< 0.05). Recallthat H2 predicted that the effect of threatappeal will diminish when controlling for pastillegal downloading behavior. Results revealthat the effect of threat appeal remains whencontrolling for past illegal downloading beha-vior for both the visit, F(2, 309)¼ 4.98,p< 0.01, and download, F(2, 309)¼ 4.66,p< 0.01, measures. In addition to the similarityof F-values across the ANOVA and ANCOVAmodels (see above), the pattern of meansacross the threat appeal conditions were alsosimilar in both models. Hypothesis 2 is notsupported.

Discussion of study one

Summary

Study one investigated two commonly usedtactics on the part of the recording industry tocurb illegal downloading of music: threatappeals and attributions of (financial) harm.

Copyright # 2007 John Wiley & Sons, Ltd. J

As predicted, findings revealed a strong andsignificant effect of threat appeal such thatstronger, more threatening appeals werefound to be more effective than weaker, lessthreatening appeals in impeding the illegaldownloading of music. Perhaps just as inter-esting as this significant finding, are thenon-significant findings revealed in the study.While high downloaders are still apt to down-load in the future, their past downloading wasdone in the absence of the threats that wedescribed in this study. The fact that werejected H2 indicates that, in the presence ofthreats, high downloaders were equallyresponsive to the persuasive message as werethe low downloaders.

Although we found that prior downloadingbehavior was significantly related to futureintentions, the significant effects of strongversus weak threat appeals remained aftercontrolling for previous illegal downloadingbehavior. Alternatively stated, our analyses ofthreat appeals and who is harmed were nodifferent when controlling for prior illegaldownloading behavior. There is a substantialbody of work suggesting that prior behavior isan important consideration in predictingbehavioral intentions generally and in thecase of threat appeals specifically (Bentlerand Speckart 1979; Fishbein and Ajzen 1975;Sutton and Eiser 1990; Umeh, 2004; Witte,1992, 1998). Thus, we designed anotherexperiment to re-consider past behavior inthis context to determine the validity of ournull findings.

Fishbein-Ajzen’s (1975) model of reasonedaction includes subjective norms as anotherpredictive variable inherent in the relationshipbetween attitude and intentions as they lead tobehavioral intentions. Subjective norms meas-ure the influence of the social environment onbehavior and are often summarized as an indi-vidual’s feelings, beliefs, and/or perceptions asto how one’s peers might judge a particularaction. Bentler and Speckart (1979) found thatdistinct from prior behavior and attitudes,subjective norms ‘impart a causal effect onintentions’ (p. 461) for both typical (e.g.,alcohol consumption) and atypical (e.g., hard

ournal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 8: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

118 Aron M. Levin et al.

drug use) behaviors. Given that downloadingmusic illegally is a fairly new, relativelyinnocuous, and technically-savvy form ofdeviant behavior, we posit that one’s peersmight play a substantial role in determining thelikelihood to illegally download music. Thusstudy two includes this potentially importantconstruct in order to understand more fully theeffects of fear appeals in this context.

Study two

Purpose

Although each of our initial predictions (maineffects as well as interactions) is re-tested instudy two, the study specifically enables us tore-examine both the harm variable (manipula-tion) as well as the prior downloading behaviorvariable. In a replication of H3a, the interactioneffect of threat appeal (fear) and harm isre-tested with a more pronounced harmmanipulation. Recall that H3a predicted aninteraction effect between threat appeal andwho is harmed such that the effect of threatappeals on intentions to downloading musicillegally would be more pronounced when arecording artist versus a company is said tobe financially harmed. A possible reason thatthe threat appeal X harm interaction wasnon-significant in our first study was due to aweak harm manipulation (see below).Additionally, study two affords us the oppor-tunity to re-test the effects of prior illegaldownloading behavior. Substantial theory andextant empirical findings suggest that priorbehavior—prior illegal downloading behaviorin this case—should play a significant role as toone’s intentions to download music illegally inthe future. Therefore, in study two, pastdownloading behavior is treated as an inde-pendent variable (versus a covariate) and ismedian-split (versus treated as a continuousvariable) in order to keep our analyticalframework intact across the two studies (usingANOVA models and procedures, etc.).

Finally, and perhaps most importantly, studytwo allows us the opportunity to introduce andtest a new and potentially important indepen-

Copyright # 2007 John Wiley & Sons, Ltd. Jo

dent variable—subjective norms. The theoryof reasoned action (Fishbein and Ajzen 1975)indicates that subjective norms play a poten-tially important role in predicting behavioralintentions. We examine whether subjects’perceptions of their peers’ feelings regardingthe illegal downloading of music will signifi-cantly affect their intentions to illegally down-load music in the future. We test the effects ofsubjective norms in the form of a main effect aswell as interaction effects (interacting with theother independent variables in the study). Thebasic prediction with respect to this newindependent variable is that the more positiveone perceives his/her peers’ perceptions to bewith respect to illegally downloading music(e.g., peers think it’s ‘cool’ to download musicillegally), the more likely one is to downloadmusic illegally in the future. Additionally, weexpect subjective norms to interact withthreat, harm, and prior downloading behaviorsuch that higher levels of the subjective normsvariable (more subjective influence) willattenuate effects of fear and harm on intentionsto download music, and intensify effects ofpast behavior on intentions to downloadmusic.

Study design and methodology

Corresponding with the discussion above,study two serves largely as a replication ofstudy one with certain enhancements regard-ing the design of the study. First, the harmmanipulation is intensified in order to capturebetter the potential effects of this independentvariable. Rather than tell subjects that down-loading causes the artist or company to loseone dollar per song downloaded (as in studyone), in study two, the cover sheet emphasizedthat downloading results in $150 million in lostprofits for artists or companies.

Because study two adds the variable of socialnorms, it is of interest to see if heavydownloaders will be as responsive as lightdownloaders to differences in social norms aswell as differences in threat level. As indicatedabove, in study two we include high versus low

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 9: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Deterring illegal downloading 119

downloaders as a factor in the design andexamine its interactions with the other factors.In addition, study two includes a newindependent variable representing subjectivenorms. Based on the theory of reasoned action,a subjective norms measure was created for thepurposes of the study. The measure consists offour items (following a five-point Likert-typeresponse format) assessing subjects’ percep-tions of what their peers think about illegallydownloading music (e.g., my peers thinkdownloading music illegally is cool, down-loading music illegally will harm the wayothers view me, etc.). These items werecombined to form the subjective normsvariable (alpha¼ 0.71). Higher scores on themeasure indicate that one believes his/herpeers approve of illegally downloading musicand that his/her image would thereby beenhanced (i.e., social approval or influence)while lower scores indicate that one thinks his/her peers do not approve of illegally down-loading music and that his/her image wouldnot be enhanced (lack of social approval orinfluence). This independent variable was alsomedian split (high versus low). Althoughrelated (r¼ 0.28, p< 0.001), the prior down-loading behavior and subjective norms vari-ables appear to be adequately distinct forpurposes of the analyses.

Finally, we combined the two dependentvariables from study one to form a singledependent variable in study two (the samedependent measures/items were used acrossthe two studies). That is, instead of having twoseparate dependent variables (two itemsmeasuring how likely subjects are to visit a‘free’ music website where downloadingmusic illegally is possible, and two other itemsmeasuring how likely subjects are to downloadmusic illegally from a ‘free’ website) as in studyone, the two measures were combined in studytwo to created a single, more parsimoniousbehavioral intentions dependent variable(comprised of four items; alpha¼ 0.88).

Aside from the differences presented above,the design and methodology of studies one andtwo were similar. Both were paper and pencilstudies administered to college students, with

Copyright # 2007 John Wiley & Sons, Ltd. J

the key dependent variables assessing subjects’likelihood to download music in the future.The second study yielded a sample size of 211.

Results

With the exception of the predictions regard-ing the newly measured subjective normsvariable (see above), the predictions for studiesone and two are virtually the same. Theanalyses in study two were, however, slightlydifferent compared with those in study one.Specifically, the past behavior variable ismedian split and treated as an independentvariable (versus a covariate) and the newlymeasured subjective norms variable is added tothe analyses and also treated as an independentvariable. In this way, we conducted a3� 2� 2� 2 (threat�harm attribution�pre-previous behavior� subjective norms)ANOVA of behavioral (illegally downloadingmusic) intentions, in which previous behavior(high versus low) and subjective norms (highversus low) were dichotomous factors. Theanalysis revealed a significant main effect forthreat appeal, F(2, 207)¼ 3.31, p< 0.04, pastbehavior, F(1, 207)¼ 55.51, p< 0.001, andsubjective norms, F(1, 207)¼ 8.49, p< 0.004.Results also revealed a significant three-wayinteraction between these three variables(fear, past behavior, and subjective norms),F(2, 207)¼ 3.53, p< 0.03.

The main effect of threat appeal found instudy two is identical to the findings of studyone. For instance, an estimate of Scheffe’spost-hoc t statistic reveals that the high threatcondition compared with the low threatcondition is significantly different (p< 0.04)in terms of intentions to download musicillegally in the future variable (M’s¼ 2.45versus 2.77 respectively). The effect of threatappeals on behavioral intentions (to illegallydownload music) appears to be fairly robust.

The main effects of both previous behavior(past downloading behavior) and subjectivenorms are unique to study two. Measured as acontinuous variable, previous downloadingbehavior was treated as a covariate in our

ournal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 10: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Low Downloaders

0

1

2

3

4

Low Med High

Threat

Lik

elih

ood

to D

ownl

oad

HSN

LSN

High Downloaders

0

1

2

3

4

Low Med High

Threat

Lik

elih

ood

to D

ownl

oad

HSN

LSN

Figure 1. HSN¼High Social Norms: those who believetheir peers are supportive of illegally downloading music;LSN¼ Low Social Norms: those who believe their peersare not supportive of illegally downloading music.

120 Aron M. Levin et al.

initial study, and, contrary to predictions, didnot appear to play a significant role regardingthe relationship between threat appeals andintentions to illegally download music. In thesecond study, previous downloading behaviorwas again measured (in the same way as instudy one) yet this time was median split andtreated as an independent variable in order todetermine if this variable interacts with othervariables in the model. In support of predic-tions, we find that previous downloadingbehavior has a significant positive effect onintentions to download music such that highdownloaders (those who have exhibited highlevels of illegal downloading behavior pre-viously) are more likely to download musicillegally in the future compared with lowdownloaders (those who have not exhibitedhigh levels of illegal downloading behaviorpreviously) (M’s¼ 3.11 versus 2.19, respect-ively). The main effect of subjective norms alsoconforms to prior predictions and suggeststhat those high in subjective norms (those whobelieve that their peers are supportive ofillegally downloading music) are associatedwith higher levels of intended downloadingbehavior compared with those who are low insubjective norms (those who believe that theirpeers are not supportive of illegally down-loading music) (M’s¼ 2.88 versus 2.34,respectively).

One of the most interesting findings in thesecond study is the three-way interactionbetween threat appeal, past behavior, andsubjective norms. Illustrated in Figure 1, thisinteraction reveals an interesting as well asappealing (with respect to the literature)pattern of findings. Under conditions of weakthreat, the results suggest that subjectivenorms play a meaningful role in deterringillegal downloading intentions among colle-ge-aged consumers who have engaged in highlevels of downloading. Again utilizing Scheffe’st statistic, results suggest that low priordownloaders who do not perceive high levelsof subjective norms are significantly less likelyto illegally download in the future comparedwith high past downloaders who perceiveeither high subjective norms (p< 0.001;

Copyright # 2007 John Wiley & Sons, Ltd. Jo

M’s¼ 1.79 versus 3.24, respectively) or lowsubjective norms (p< 0.01; M’s¼ 1.79 versus3.06, respectively). Thus, for the low threatcondition, subjective norms have much moreinfluence for those with little downloadingexperience.

Under conditions of moderate threat, thepattern of findings seems to suggest thatprevious downloading behavior is what mat-ters most in terms of curbing intentions toillegally download music. In this moderatethreat condition, for example, the Scheffe t

statistic reveals that among those who experi-ence high subjective norms, those low in pastdownloading behavior are significantly(p< 0.03) less likely to download in the futurecompared with those high in past down-loading behavior (M’s¼ 2.11 versus 3.27,respectively; see Figure 1).

Finally, in the strong threat condition, thepattern of findings seems to suggest that bothsubjective norms and past downloading beha-vior are important in terms of decreasingintentions to illegally download music. Under

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 11: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

Deterring illegal downloading 121

conditions of high threat, for instance, theScheffe t statistic suggests that those low inboth subjective norms and past downloadingbehavior are significantly (p< 0.05) less likelyto download illegally in the future comparedwith those high in both subjective norms andpast downloading behavior (M’s¼ 2.05 versus2.92, respectively; see Figure 1).

Overall Summary/discussion

In a series of two experiments, we examinedthe impact of a variety of factors on collegestudents’ likelihood to illegally downloadmusic. The results of the two studies indicatethat stronger threat appeals lessened thelikelihood that students will download musicin the future, and students who have down-loaded many songs in the past are more likelyto download in the future. Contrary to ourpredictions, we found that attribution of harmhad no impact on students’ intentions todownload in the future. It is quite possible thatcollege students, who are typically not yetfinancially independent, have a hard timebelieving that music artists and/or companiestruly need the money that is lost due todownloading music. It is also possible thatcollege students simply do not care whetherthe artist or company is financially harmed. Asample of working adults would possibly yieldquite different findings. The most interestingfinding from study two was that social normshave a significant impact on future down-loading behavior. In other words, collegestudents do care what others think aboutthem downloading songs without paying.

Whereas previous literature tended to lookat the variables of interest in isolation, thecurrent research makes a contribution to theliterature and to theory development byexamining the relations among the fourvariables (threat appeals, attribution of harm,previous behavior, and social norms). Inparticular, this research illuminates the poten-tial importance of social norms in discouraginga type of undesirable consumer behavior butshows that this occurs only under a restricted

Copyright # 2007 John Wiley & Sons, Ltd. J

set of conditions: when threat is low and theconsumer is not a habitually high downloader.For these consumers, the extra incentive ofdoing what others think is ‘cool’ compensatesfor low levels of threat but not for higher levelsof threat. Current theories concerning the roleof social norms in curbing undesirable forms ofconsumer behavior must consider the moder-ating role of factors like past behavioralpatterns.

The results of these two studies have clearimplications for marketers of electronic media(music, and possibly movies) who wish todissuade consumers from pirating theirmaterial on the Internet. The use of severethreats seems to be an effective way todiminish the likelihood of downloading, whileattributing financial harm to either the com-pany or artist was not found to be particularlyeffective. In addition, music/movie companiesmight make better use of their de-marketingefforts by focusing their message on consu-mers who have already tried illegal down-loading, but are not yet ‘habitual’ downloaders.Finally, the use of social norms as a disincentiveto download music may be particularlyeffective for certain consumers. PSA’s for otherundesirable behaviors such as underage drink-ing and smoking often use the approach that itis ‘uncool’ to engage in these behaviors.Perhaps a campaign that suggests to youngconsumers that their friends won’t think theyare cool if they pirate songs would be effective.Additional research on high school studentswould be useful to detect when and to whatextent the ‘cool’ factor impacts young con-sumers’ likelihood to illegally download musicand/or movies.

Acknowledgements

The authors wish to thank Irwin Levin at theUniversity of Iowa for his helpful comments onearlier drafts of this manuscript.

References

Boster FJ, Mongeau P. 1984. Fear—arousing per-

suasive messages, In: Bostrom R.N.Westley BH.

ournal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb

Page 12: Deterring illegal downloading: the effects of threat appeals, past behavior, subjective norms, and attributions of harm

122 Aron M. Levin et al.

Eds. Communication Yearbook 8, Sage: Beverly

Hills, CA; 330–375.

Burnett JJ, Oliver RL. 1979. Fear appeal effects in

the field: a segmentation approach. Journal of

Marketing Research 16: 181–190.

Fullerton SD, Kerch KB, Dodge HR. 1996. Consu-

mer ethics: an assessment of individual behavior

in the market place. Journal of Business Ethics

15: 805–814.

LaTour MSHJ, Rotfeld HJ. 1997. There are threats

and (maybe) fear-caused arousal: theory and con-

fusions of appeals to fear and fear arousal itself.

Journal of Advertising 26(3): 45–59.

Levin AM, Conway Dato-on M, Rhee K. 2004.

Money for nothing and hits for free: the ethics

of downloading music from peer-to-peer web

sites. Journal of Marketing Theory and Practice

12(1): 48–60.

McBride S. 2006. Movie Debut: Films for Sale by

DownloadWall Street Journal, April 3, 2006, B1, B6.

Mongeau P. 1989. Another look at fear arousing

messages,. In: Allen M, Preiss R. (eds.), Persua-

sions: Advances Through Meta-Analysis.

Cresskill, NJ; Hampton Press: 53–68.

Copyright # 2007 John Wiley & Sons, Ltd. Jo

Rossiter JR, Thornton J. 2004. Fear-pattern analysis

supports the fear-drive model for antispeeding

road-safety TV ads. Psychology & Marketing

21(11): 945–960.

Sutton SR, Eiser JR. 1990. The decision to wear a

seat belt: the role of cognitive factors. fear and

prior behaviour, Psychology &Health Vol. 4(2):

111–123.

Umeh K. 2004. Cognitive appraisals, maladaptive

coping, and past behaviour in protection

motivation. Psychology & Health 19(6): 719–

725.

Witte K. 1992. Putting the fear back into fear

appeals: the extended parallel process model.

Communication Monographs 59: 329–349.

Witte K. 1998. Fear as motivator, fear as inhibitor:

using the EPPM to explain fear appeal successes

and failures. In: Andersen PA, Guerrero LK.

(eds.), The Handbook of Communication and

Emotion. Academic Press: New York; 423–450.

Witte K, Allen M. 2000. A meta-analysis of fear

appeals: implications for effective public health

campaigns. Health Education & Behavior 27(5):

2000. 591–615.

urnal of Consumer Behaviour, Mar.–Jun. 2007

DOI: 10.1002/cb