SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

This exploratory study applies and extends a new model of media attendance to examine factors that determine current levels of sharing files through peer-to-peer networks among college students, and to predict downloaders’ intentions to discontinue the behavior in the future. In a multiple regression analysis, downloading activity was found to be positively related to deficient self-regulation and the expected social outcomes of downloading behavior. Downloading activity was lessened by dissatisfaction with poor quality downloads. Those who are willing to discontinue downloading are motivated by fear of punishment, but skilled and habitual downloaders were unlikely to discontinue. Normative beliefs also affected downloading. The perception that downloading was morally unacceptable was positively related to intentions to discontinue downloading, while beliefs that the behavior was morally acceptable were positively related to current downloading activity.


The downloading problem

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

Downloading media files over the Internet is a new and highly controversial form of media consumption behavior. Beginning with the Napster phenomenon in 1999, the craze peaked in the spring of 2003 around the time that the record industry in the United States first made good on threats to prosecute downloaders. At that time, some 35 million U.S. adults used the Internet to download music. Four percent said they did so on an average day. Downloading music has been especially popular among young adults, with 52% of those aged 18–29 involved (Rainie & Madden, 2004). The U.S. Supreme Court’s decision in the Grokster case, which held file sharing companies liable for copyright infringement, did not eliminate the behavior (CacheLogic, 2005), although the decision did drive further innovations in piracy, such as “dark networks” that use encryption to evade the prying eyes of the music industry (Chmielewski, 2005).

Statistics such as those quoted above are the result of millions of individual decisions about whether or not to engage in downloading behavior, and draw upon theories of media consumption behavior. At the same time, downloading is a widely debated social and public policy issue, calling our attention to new media attendance motivations that highlight normative and ethical concerns.

The file sharing controversy provides the conceptual context for the present study. Many of the files that are “shared,” or made available for downloading, are provided without the permission of their copyright owners. The industry had a convenient scapegoat in Napster: Its success coincided with a drop in CD (compact disc) sales, leading to the supposition that the phenomenon was being driven by the substitution of free online music for CD purchases (Stern, 2000). The industry responded with efforts to curb file sharing, led by the Recording Industry Association of America (RIAA). By mid-2001, the industry succeeded in shutting down the original Napster—which by that time claimed 50 million users—through court rulings that enforced the Copyright Term Extension Act of 1998.

Napster, however, was soon replaced by new file sharing programs that used peer-to-peer (P2P) technology. Unlike the original Napster, P2P programs act as user interfaces for end users, facilitating direct exchanges between them instead of coordinating the exchanges through a central file server. That distinction eluded the anticircumvention provision of the new copyright law that had tripped up Napster. A variety of P2P client programs appeared, including KaZaa, Grokster, and Gnutella, which were blamed for further declines in CD sales of 8.7 percent in 2002 and 2.1 percent in 2003 (Glasner, 2003; Pesola, 2004).

With the P2P software companies beyond the reach of the law, the RIAA responded by suing their users, with an initial wave of 261 lawsuits against individual copyright infringers in early September 2003, immediately preceding the collection of the data reported here. Other industry information campaigns (e.g., http://www.riaa.com/issues/piracy/default.asp, http://www.respectcopyrights.org) stressed the economic harm suffered by artists and the industry and the dangers (e.g., virus infection) of downloading.

File sharing grew especially popular on college campuses, where there were large numbers of music fans with limited budgets, high speed Internet connections, and time on their hands. Historically, universities were “common carriers” under the law, with no direct control over the behavior of individuals on the network: While individuals themselves would be liable, the university would typically escape harm. The Copyright Term Extension Act of 1998 reinforced that protection for universities, provided they took down offending content in a timely manner after receiving a notice of infringement. Because of the sheer amount of file sharing activity at colleges, file sharing opponents confronted universities, such as when the RIAA filed copyright infringement suits naming selected major universities as codefendants (Carlson, 2000). Later the RIAA sent a threatening letter (Rosen, 2002) to some 2,000 college and university presidents enlisting them in the policing of copyright infringement on their own campuses. For their part, universities had their own reasons to discourage file sharing, notably the strain it placed on overloaded computer networks. In order to avoid costly legal battles, as well as to restore the integrity of their own data networks, many universities actively sought to discourage downloading. Universities used tactics ranging from supplying identities of infringers to the industry, mounting their own information campaigns, and threatening their students with disciplinary actions if they did not stop file sharing (Carlson, 2003).

The ultimate goal of these efforts was to create a culture shift among file-sharers (Borland, 2003). Evidence of their success is thus far mixed, however. One survey found that downloading music from peer-to-peer services climbed 14% in November 2003 compared to September (NPD Group, 2004), while other data showed that the proportion of Internet users downloading music dropped by half between April and December 2003 (Bonne, 2004; Rainie & Madden, 2004). Indeed, the sales data that initially led to the conclusion that downloading was harming the industry are themselves controversial: A downturn in the general economy, an upswing in college tuition costs, and a dearth of new music all pose competing explanations for declining music sales. Clearly, more insightful research is needed about online music consumption behavior and its effect on the music industry.

A fundamental question is what motivates downloading behavior? Are industry executives, public policy makers, and college administrators using the most appropriate and effective strategies in their efforts to deter file sharing? Do they properly understand the usage patterns and motivations for this new form of media consumption aside from the narrow perspective of their own institutional self-interests? The present exploratory research develops and tests a model of file sharing behavior that examines the factors related to current downloading activity and downloaders’ intentions to discontinue file sharing in the future.

Understanding downloading behavior

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

The motivations for online media consumption have been a frequent topic of recent communication research in the Uses and Gratifications tradition (Charney & Greenberg, 2001; Flanagin, & Metzger, 2001; Kaye, 1998; Korgaonkar & Wolin, 1999; LaRose, Mastro, & Eastin, 2001; Lin, 1999; Papacharissi & Rubin, 2000; Parker & Plank, 2000; Stafford & Stafford, 2001). Unfortunately, the studies that followed the conventional Uses and Gratifications approach most closely (e.g., Kaye, 1998; Papacharissi & Rubin, 2000) repeated a rather disappointing pattern of findings common to the “old media” studies that preceded them: an inability to explain very much (between 1 and 15%) of the variance in media consumption behavior (cf. Palmgreen, Wenner, & Rosengren, 1985).

A variety of innovative operational and conceptual approaches, however, have emerged through Internet-related research that holds the promise of creating more robust models of media consumption behavior. LaRose and Eastin (2004) analyzed these innovations within the framework of Social Cognitive Theory (SCT) (Bandura, 1986), a theory of human behavior. It is best known in the annals of communication research in its earlier incarnation as social learning theory (Bandura, 1977), through which the mechanism of vicarious learning was introduced to media effects research. SCT, as it is now known, is a broad theory of human behavior and so may be applied to media consumption behavior as well as to the behavioral effects of media consumption. As such, it shares many features with other behavioral theories, such as the Theory of Planned Behavior (Ajzen, 1985). But while there are many parallels between the two (cf. Bandura, 1997, pp. 125–128), SCT embraces a wider range of explanatory mechanisms.

SCT posits that the performance of a behavior is determined in part by its expected outcomes. The expectations may be learned through observation of the consequences visited upon others (the mechanism of vicarious learning), but the consequences learned from direct personal experience (the mechanism of enactive learning) are also powerful. Whether the consequences, or outcomes, of behavior (e.g., downloading) are personally experienced (e.g., by noting that money is saved by downloading a CD instead of buying it) or vicariously experienced through others (e.g., by hearing others boast of their savings on CD purchases as a result of downloading), beliefs about the expected outcomes of future behavior are continually modified in light of experience. The expected outcomes of behavior then determine future behavior, although other mechanisms (see below) modify the behavioral response. Expected positive outcomes of a specific online consumption behavior should cause future consumption while expected negative outcomes should deter it. For example, the expectation of saving money on CD purchases, a positive outcome, encourages downloading, while the fear of being expelled from school, a negative outcome, discourages it. In short, our experience with the media shapes our expectations about the outcomes of future media consumption and so determines our willingness to engage in it.

LaRose and Eastin (2004) distinguished expected outcomes from the conventional gratifications-sought/gratifications-obtained formulation (Palmgreen, Rosengren, & Wenner, 1985). They argued that studies that employed prospective measures of expected outcomes (e.g., Charney & Greenberg, 2001; LaRose, Mastro, & Eastin, 2001; Lin, 1999) produced improved predictions of behavior compared to conventional approaches, typically doubling or tripling the amount of variance explained in media consumption behavior compared to conventional approaches. LaRose and Eastin also noted that prior Uses and Gratifications studies may have over-relied on lists of gratifications developed in old media contexts (e.g., Rubin, 1984). Those lists consistently under-represented important categories of incentives for human behavior known to SCT, particularly status, and novel sensory and monetary incentives (cf. Bandura, 1986, pp. 232–240). Research that expanded upon those lists by adding new theory-based dimensions (e.g., Charney & Greenberg, 2001; LaRose et al., 2001) also produced improved results.

SCT introduces the concept of self-efficacy that could further explain media consumption behavior. Self-efficacy is belief in one’s ability to complete a particular course of action to achieve important attainments (Bandura, 1986). The concept is relevant for users who have yet to master the computer skills necessary to complete important tasks online. Past research has linked self-efficacy to the overall amount of Internet usage (LaRose et al., 2001).

Self-efficacy, however, may also be understood in regard to specific subcategories of media consumption behaviors, such as the use of file-sharing software. Indeed, file sharing presents an interesting case since the competencies required are something of a moving target. The users of file-sharing services must continually hone their skills to overcome the barriers to successful utilization, such as the avoidance of corrupted or mislabeled music files and the evasion of media industry investigators.

A subcategory of self-efficacy known as coping self-efficacy (Bandura, 1997), or the perceived ability to cope with negative life events, may play a role in file-sharing behavior. In the present context it would be the individual’s perceived ability to avoid the negative consequences of file sharing—the perception that one would not get caught, in other words.

The self-regulation mechanism of SCT (Bandura, 1991) describes how individuals observe their own behavior, judge it according to personal and societal standards, and then self-administer incentives to change their behavior. This mechanism prevents individuals from blindly following the dictates of external reinforcement. When self-regulation is deficient, increased media consumption may be expected, perhaps out of habit (LaRose, Lin, & Eastin, 2003).

Habit may be conceptualized as a form of automaticity (Aarts, Verplanken, & van Knippenberg, 1998; Bargh & Gollwitzer, 1994), a recurring behavioral pattern (e.g., downloading music daily) that follows a set cognitive schema. The pattern is triggered by an environmental stimulus (e.g., connecting to the Internet) or by recalling a goal (e.g., the desire to possess every Frank Sinatra recording ever made) and performed without further self-instruction. Users may well evaluate their behavior actively the first few times they download songs, but by downloading session number 100, they no longer do. This is a failure of the self-observation subfunction as understood in SCT.

The manifestation of repeated behavior, however, does not distinguish between a behavior that stems from active decision-making or habit. Rather, the distinction rests on the degree to which the behavior is under self-control. Reliable operational procedures have been developed (e.g., Aarts et al., 1998; LaRose et al., 2003) to assess the degree of self-regulation exercised over specific behaviors.

Repetition makes users inattentive to the original reasoning behind their behavior—the behavior becomes a goal in itself, no longer the result of any actively imagined expected outcomes or desired gratifications. This might explain accounts of seemingly impossibly large collections of music files (see results section, below) that are larger than even the most avid music fan might actually enjoy. In the extreme, the failure of conscious self-control (deficient self-regulation) has been suggested as an explanation for problematic Internet use (LaRose et al., 2003). The lack of self-regulation makes media usage a goal unto itself, no longer the product of expected outcomes, but rather a habitual response to a negative effect. In turn, unregulated media behavior can interfere with normal life activities, producing negative real world consequences (e.g., faltering relationships, failing grades) that in turn lead to deeper negative effects and a spiral of mounting media usage.

Arguments against file sharing predicated on legal, moral, and ethical grounds (e.g., file sharing is illegal, harms struggling musicians, or is morally wrong) address the second subfunction of self-regulation, the judgmental process. Such arguments may be effective to the extent that they convince users that their downloading is socially unacceptable. This mechanism, however, may also encourage file-sharing behavior if advantageous comparisons are made to lax norms for conduct, such as the perception that “everyone is doing it” or that media conglomerates are undeserving of sympathy, as these perceptions lend an aura of social acceptability.

The judgmental process within sociocognitive theory thus provides a means of understanding the influence of normative beliefs on media consumption behavior. Normative influences on media usage have been examined previously through a closely related paradigm, the Theory of Planned Behavior (Ajzen, 1985). There, normative beliefs are conceptualized and operationalized as beliefs about the desires of specific significant others in relation to the target behavior. For example, beliefs about the approval of relevant significant others have been found to predict consumption of video games (Doll & Ajzen, 1992), although not television viewing (Ouellette & Wood, 1998). Moral norms, or perceptions of the moral correctness or incorrectness of performing a behavior, have been suggested as a distinct variable that taps a different dimension of normative influence (Conner & Armitage, 1998), one that has not been examined in previous media research. Since the debate over downloading has frequently invoked ethical as well as legal arguments, downloading behavior would seem to present a heuristic context in which to examine moral norms.

Hypotheses

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

The SCT view of media behavior suggests that the expected positive and negative outcomes of downloading are important initial causes of behavior. The expected outcomes that users experience at a given point in time should govern both their current behavior and their intentions to perform it in the future. That is, if I expect to save money by downloading music, this expectation will logically be reflected in my current level of downloading activity and also frame my intentions to engage in further downloads from this point forward. From reading popular accounts of the downloading phenomenon, we also expect that the variety of music available and the ability to socialize with like-minded music fans are other positive outcomes that downloaders might expect. On the other hand, the fear of punishment and the poor quality of many downloaded files may be negative outcomes that deter downloading. It is well established that expected outcomes predict behavioral intentions in this way and that behavioral intentions in turn can predict actual future behavior (cf. Ajzen, 1985). Previous research (LaRose & Eastin, 2004) established that expected outcomes predicted overall Internet consumption behavior. The Internet encompasses a wide range of content, however, so perhaps global statements about expected outcomes weakly tap a wide variety of motivations that predict consumption behavior aggregated across that entire, highly diverse medium, but would fail to predict specific behaviors, such as downloading. This possibility has yet to be explored.

H1: Expected positive outcomes of downloading will be positively related to a) current downloading activity and b) future downloading intentions.

H2: Expected negative outcomes of downloading will be negatively related to a) current downloading activity and b) future downloading intentions.

Self-efficacy specific to the use of file-sharing software, or file-sharing self-efficacy, should be positively related to current and future behavior as well. Coping self-efficacy, defined in relation to the ability to avoid the negative consequences of downloading, should also facilitate these behaviors.

H3: File sharing self-efficacy will be positively related to a) current downloading activity and b) future downloading intentions.

H4: Coping self-efficacy will be positively related to a) current downloading activity and b) future downloading intentions.

Expected outcomes and self-efficacy are likely to be especially salient in the early stages of a novel media consumption behavior pattern. Over time, media consumption behaviors are likely to become habitual as self-regulation becomes inattentive and defective:

H5: Deficient self-regulation will be positively related to a) current downloading activity and b) future downloading intentions.

Beliefs about the negative legal, moral, and ethical implications of file-sharing behavior should also have an impact on downloading. In other words, beliefs about the moral unacceptability of file sharing may trigger judgmental comparisons with standards of conduct that restore effective self-regulation and moderate downloading.

H6: Beliefs about the moral unacceptability of file sharing will be negatively related to a) current file sharing activity and b) future downloading intentions.

File sharing users, however, may also judge themselves in relation to lax standards that provide favorable judgments about the appropriateness of personal conduct. Such judgments minimize self-reactive influences, such as feelings of guilt, and hence encourage downloading:

H7: Beliefs about the moral acceptability of file sharing will be positively related to a) current file sharing activity and b) future downloading intentions.

Research methods

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

Respondents

A convenience sample of 265 undergraduate students was recruited from an introductory communication course at a large university located in the Midwestern United States. The sample was 69% white and 67% male with a median age of 19, consistent with the demographics of the class from which the sample was drawn. Sixty-six percent of the sample came from households with incomes over $50,000. In all, 82% of the respondents were active downloaders, defined as those who reported having downloaded files of music, movies, videos, images, video games or software applications. The other 18% reported no downloading activity. The average respondent spent 3 hours and 54 minutes on the Internet each day. The average amount of time the active downloaders in the group spent on that activity each day was one hour and nine minutes. The average respondent had 734 MP3s stored on his or her hard drive as a result of downloading, but that average was skewed by a small number of very large collections, including one respondent who claimed over 30,000 recordings. The median MP3 collection size was 200.

Convenience samples such as the present one violate a key assumption of parametric statistics in that they are not randomly selected from the population of interest and therefore do not provide reliable estimates of population parameters. Their value in exploratory research, aimed at uncovering lawful relationships among variables, is, however, widely accepted. The size of the present sample may be characterized as fair to good for the purposes of multivariate analysis (Comrey & Lee, 1992).

Procedure

Participants completed online surveys from their own personal computers in the last week of November and the first week of December of 2003. Extra credit was offered in exchange for participation with the approval of the university human subjects review committee. In order to preserve participants’ anonymity and to encourage honest responses to potentially incriminating questions about downloading activity, the server log was turned off. This was done so that neither the researchers nor the authorities could match the questionnaires with individual respondents.

Operational definitions

The current downloading activity dependent variable was a composite of the amount of time, in hours and minutes, spent downloading in the typical weekday and in the typical weekend day in the month prior to the survey. The weekday and weekend day times were converted to minutes and subjected to a log10 (value +1) transformation before being summed to minimize the effects of outliers. The means, standard deviations, and alpha coefficients of the variables are shown in Table 1. Eight respondents who reported out of range numbers (e.g., 30 hours per day) were removed from the database to arrive at the final sample size of 265. Intentions to discontinue downloading were assessed with a single behavioral intention item which asked how likely the respondent was to discontinue downloading entirely in the next month. Intentions were rated on a 7-point scale ranging from very likely (scored as 7) to very unlikely (scored as 1). The reliability and validity of behavioral intention items is well established (Ajzen, 1985).

Table 1.  Pearson product-moment correlation coefficients, means, standard deviations, and reliabilities of variables.
Variable123456789101112Hi–LoMeanS.D.
  1. Note: Boldface indicates significant correlation coefficients, p < .05. Diagonal entries are Cronbach alpha coefficients for multi-item indices.

1. Download Activity.88 0–5.72.461.71
2. Intentions to stop−.16n/a 1–72.591.68
3. Variety.23−.17.70 3–2117.03.37
4. Punishment−.01.39−.05.83 3–2110.44.41
5. Poor Quality−.01.19.04.23.52 3–2111.73.68
6. Social.34.10.17.19.21.66 3–2111.24.27
7. Economic.12−.03.52−.06.06.10.69 6–2117.92.95
8. Deficient Self Regulation.40−.06.03.04.24.36−.02.70 4–2713.15.00
9. Self-efficacy.23−.36.56−.21−.09−.01.31.08n/a 1–75.61.45
10. Coping efficacy.24−.22.21−.21.06.10.19.32.30.73 2–148.32.97
11. Unacceptability−.09.23−.15.26.21.07−.01.21−.18−.08.66 4–2815.54.77
12. Acceptability.18−.06.32.04.04.23.26.11.22.08−.18.585–2114.93.41

Statements about the expected outcomes of downloading were developed from popular press accounts of the file-sharing phenomenon and were rated on 7-point Likert-type scales (scored 7 for strongly agree, 1 for strongly disagree). Since the expected outcomes of downloading behavior were not examined in previous research, an exploratory factor analysis was appropriate. The principal components method with varimax rotation was used in order to maximize the interpretability of the variables in relation to the underlying factors and thus clarify their relationship to theoretical constructs represented by the factors. A five-factor solution explaining 58% of the variance was obtained. The outcome factors were Variety,1 Punishment,2 Poor Quality,3 Social,4 and Economic.5 SCT includes a priori assumptions about the categories of incentives that frame outcome expectations and motivate behavior (cf. Bandura, 1986, pp. 232–240), around which outcomes may be expected to factor. The variety factor coincides with novel sensory incentives, which motivate behavior through exposure to novel sights and sounds. The poor quality factor is a negative instance of the enjoyable activity incentive, in which behavior is motivated by the expectation of taking part in pleasing activities. The economic factor exemplifies monetary incentives, while the punishment factor is a negative instance since it involves penalties (lawsuits, suspension from school) that entail a significant financial burden. Finally, the social factor characterizes social incentives, which prompt behavior through the prospect of rewarding social interactions. Thus, all of the resulting factors were interpretable within the social cognitive framework and all were retained in the analysis despite the fact that one (poor quality) displayed substandard reliability (cronbach alpha = .52).

A 4-item measure of Deficient Self-Regulation was adapted from LaRose and Eastin (2004),6 where evidence of its reliability and validity were previously established. File-sharing self-efficacy was a single-item Likert-type scale, “I know how to use file sharing software,” while coping self-efficacy was assessed with two Likert-type items.7 Both of these were measured following well-established procedures from the social cognitive research tradition, although they had never before been applied to downloading specifically. Four beliefs about the moral unacceptability of downloading8 and three statements attesting to its moral acceptability9 were also rated with Likert-type scales.

The latter two variables were not included in previous research and so the alpha coefficients and correlations presented in Table 1 provide preliminary evidence of their reliability and validity. The means, standard deviations, and reliability coefficients of these variables are shown in Table 1. The moral acceptability measure narrowly missed a commonly accepted rule of thumb for internal consistency (Nunnally, 1967) in exploratory research, a Cronbach alpha of .6, while the moral unacceptability measure fell within it. Both were retained to provide preliminary evidence about the moral acceptability construct. One would expect the two to be negatively correlated to one another (r(261) =−.177, p < .05), providing initial evidence of their construct validity.

Data analysis

All analyses were performed using the Statistical Package for the Social Sciences, Version 11 (SPSS, 2003). Exploratory factor analysis, Pearson product-moment correlations, and multiple regression procedures were used. Missing data on the independent variables (between 0 and 1.5% of the cases) were replaced with mean values. An alpha level of .05 was adopted for the present research. Alpha levels of .05 or lower are generally considered acceptable for exploratory research (Wimmer & Dominick, 2003). Fourteen individual hypotheses were tested through correlation coefficients, raising the probability that familywise error could have yielded one significant finding by chance. Corrections for familywise error reduce statistical power, however, and are no longer considered an essential practice. They are seldom applied in multiple correlation studies (O’Keefe, 2003).

The analysis of future downloading intentions was limited to a subset of the sample who were currently engaged in downloading, defined as those who had an archive of downloaded music, video, film, game, software, or image files. There were 218 such respondents, or 82% of the sample. Of these, 3.4% said they were very likely to stop file sharing altogether in the next month, 3.4% likely, and 5.7% somewhat likely. Another 8.7% indicated that they were neither likely nor unlikely to stop. In contrast, 32.6% rated themselves very unlikely to discontinue, 27.5% unlikely and 16.1% somewhat unlikely, for a total of 100% of those who had an archive of downloaded material. There was a significant negative correlation (r(218) =−.164, p < .05) between intentions to stop downloading and current downloading activity among those currently engaged in downloading by virtue of maintaining an archive of files.

Results

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

Current downloading behavior

The results for current downloading behavior are presented first, followed by those for future intentions (see Table 1). There were significant positive correlations between the amount of downloading activity and variety (r(265) = .23, p < .001) and social expected outcomes (r(265)= .34, p < .001), but not economic (r(265) = .12, p = .053) expected outcomes, partially confirming H1a. Two categories of negative outcomes had negative relationships with downloading activity, in the direction hypothesized, but these were not significant, so H2a was not confirmed. As predicted, there were significant positive correlations between downloading activity and file sharing self-efficacy (r(265) = .23, p < .001), coping self-efficacy (r(265) = .24, p < .001), and deficient self-regulation (r(265) = .40, p < .001), confirming H3a, H4a, and H5a, respectively. The relationship of file sharing activity to perceptions of the moral unacceptability of the behavior were in the predicted negative direction but was not significant, so H6a was not confirmed. Perceived moral acceptability did have a positive relationship to file sharing activity (r(265) = .179, p < .01), supporting H7a.

The regression of the independent variables on current downloading activity produced a significant result (F(10,254) = 21.74, p < .001, R = .530, adjusted R2= .25, Table 2). After controlling for the other variables, only expected social outcomes (β= .22, p < .001) and deficient self-regulation (β= .34, p < .001) remained as positive predictors, while poor download quality (β=−.12, p < .05) and moral unacceptability (β=−.12, p < .05) emerged as significant negative predictors of current downloading activity.

Table 2.  Multiple regression of predictor variables on current downloading activity
 Unstandardized CoefficientsStandardized Coefficients
 BStd. ErrorBetatSig.
  1. F(10,254) = 21.74, p < .001 R = .530, adjusted R2= .253

(Constant)−.867.800 −1.083.280
Variety.044.037.0881.186.237
Punishment.012.023.030.511.610
Poor Quality−.054.027−.116−2.021.044
Social.086.024.2163.571.001
Economic.011.037.019.303.762
Deficient Self Reg..116.022.3385.323.001
Self-efficacy.118.081.1001.466.144
Coping efficacy.037.035.0641.058.291
Unacceptability−.044.021−.121−2.048.042
Acceptability.007.030.013.225.822

Future downloading intentions

H1b received only partial support; only variety outcomes were significantly negatively related to intentions to discontinue downloading (r(218) = .17, p < .05). H2b was supported; both the expectation of punishment (r(218) = .39, p < .001) and the expectation of poor quality downloads (r(218) = .19, p < .01) were directly related to intentions to stop. File sharing self-efficacy (r(218) =−.36, p < .001) and coping self-efficacy (r(218) =−.22, p < .001) were inversely related to discontinuance, confirming H3b and H4b. H5b, however, was not supported; there was no relationship between deficient self-regulation and intentions to stop downloading in the next month, although the relationship was in the direction predicted, with higher levels of deficient self-regulation tending to be associated with lower levels of intentions to discontinue file sharing (r =−.06, p = .390). Perceptions of moral unacceptability had the expected direct relationship (r(218) = .23, p < .001) while perceptions of moral acceptability did not, so H6b was confirmed but not H7b.

The regression of the independent variables on intentions to discontinue downloading produced a significant result (F(10,217) = 7.75, p < .001 R = .522, adjusted R2= .24). In this analysis, the expectation of punishment was the only expected outcome with a significant relationship to discontinuance (β= .22, p < .001). After controlling for the other variables, deficient self-regulation (β=−.22, p < .01) and file sharing self-efficacy (β=−.22, p < .01) emerged as significant predictors with relationships in the directions originally hypothesized (by H5b and H3b).

Post hoc analyses of the results showed significant but small correlations between the belief that one is likely to be sued by the industry and intentions to discontinue file sharing (r(218) = .25, p < .05) and between the belief in the consequence of losing one’s campus computer account and intentions to discontinue (r(218)= .34, p < .05). Belief in the consequence of expulsion from school was moderately correlated to discontinuance intentions (r(218) = .43, p < .001).

Expected economic outcomes were not related to the level of current downloading activity or intention to discontinue downloading in the future. Furthermore, post hoc analyses revealed no significant relationships between the reported family incomes of the respondents and either their current (r(265) = .02, p > .05) or projected future downloading (r(218) =−.08, p > .05).

Discussion

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References

The present results supported the hypothesized model of downloading activity derived from SCT. It extends previous research (LaRose & Eastin, 2004) by demonstrating that the sociocognitive model can predict specific types of Internet usage as well as the overall amount of attendance to the medium. In the case of downloading behavior, variety outcomes—the anticipation of obtaining new, rare, or customized music selections—were significant predictors of current downloading, as were expectations of social interactions with other file sharers. Self-efficacy in using file sharing software and the perceived ability to avoid punishment for doing so were also related to the amount of downloading individuals engaged in. The perceived moral acceptability of file sharing was also a contributing factor. The most important predictor of downloading was deficient self-regulation—the perception that one’s file sharing was out of control.

After controlling for the interactions among variables in multiple regression analysis, social outcomes, and deficient self-regulation remained as important predictors. The expectation of poor quality downloads and beliefs that file sharing is morally unacceptable (i.e., because it is morally wrong and harms artists and recording companies) also emerged as significant negative predictors. The role of deficient self-regulation implied that there is some degree of automaticity in file sharing behavior (Aarts et al., 1998). File sharing becomes a habit rather than a purposive behavior, with files accumulating on college students’ computers that they may never even open or listen to. These results thus supported previous work (LaRose & Eastin, 2004; LaRose et al., 2001) that related Internet usage, deficient self-regulation, and expected outcomes.

The present research suggests an important addition to the sociocognitive model of media attendance proposed by LaRose and Eastin (2004): comparisons to standards of conduct. Specifically, the moral unacceptability of the media consumption behavior in question was related to the performance of that behavior. This taps a new element of the self-regulatory mechanism that was previously unexamined within SCT, the judgmental process. The previous formulations included constructs that tapped the other two subprocesses of self-regulation: namely self-observation and self-reactive influence, which are examined here through the deficient self-regulation variable. These previous formulations, however, did not address judgmental process.

This finding opens a potentially fruitful avenue for further development of theories of media attendance. Judgmental process was conceptualized to involve comparisons of one’s behavior to moral standards for conduct, in this case, beliefs about the moral justification for the behavior in question. The judgmental process, however, may utilize a wide variety of referent beliefs. For example, in the Theory of Planned Behavior (TPB) (Ajzen, 1985), perceived social norms are referenced to the beliefs of specific significant others (e.g., one’s parents or roommate) and the motivation to comply with their wishes. Sociocognitive theory suggests yet other standards for comparison, including personal comparisons with one’s own prior behavior, social comparisons with individual members of relevant reference groups (e.g., one’s online “buddies”), and collective comparisons (e.g., with one’s online “music sharing community”), as well as norms for morally acceptable behavior. Nor do moral norms exhaust all the possibilities for behavioral standards. Standards may also be set by knowledge of patterns of behavior that are typical for one’s station in life, such as by stories in the media about the level of downloading activity typical of college students today.

There appears to be little precedent for examining normative influence on media attendance in the annals of communication research. An earlier TRA-based study of video game usage (Doll & Ajzen, 1992) found that perceived social norms were a significant predictor of usage, but the “significant other” in that case was the experimenter himself. We can imagine a wide range of media consumption behaviors that might be affected by the judgmental process mechanism, ranging from those who avoid online pornography because they believe it is morally wrong, to those who read articles in the Journal of Computer Mediated Communication because they believe that their professors wish them to do so.

Individuals who use file sharing heavily may do so to meet not only musical needs but social needs as well. The benefits of involvement with a community of other file sharing peers, combined with the habitual nature of file sharing, may perhaps outweigh their fears of punishment. The countervailing forces of potentially poor quality files and the social unacceptability of file sharing may discourage but not stop file sharers because of the much greater social benefits from group membership. Among these social benefits are a sense of community and the chance to know what other file sharers are listening to. Musically, it is a way to find out what is popular well before seeing the video on MTV. Seeing others’ play lists also affords the chance to find music before it enters the mainstream and to indulge tastes that are never part of the mainstream. On the surface, the finding that individuals engage in a media behavior to fulfill social needs is nothing new. Uses and gratifications research (Palmgreen et al., 1985) has long recognized social needs as important motivating factors in media usage in general and for the Internet in particular (e.g., Papacharissi & Rubin, 2000).

The nature of the file sharing experience, however, brings a fresh theoretical perspective to social motivations for media usage. Consider the statement, “I get to see the play lists others have so I know what is hot,” one of the items in the measure of social outcomes used in the present study. How many lists? How hot? Which others? In more formal terms, these are the issues of size of community, asymmetry of resources, heterogeneity of interests and resources, and task interdependence that are thought to drive adoption of interactive media in theories of critical mass (Markus, 1990) and discretionary data bases (Connolly & Thorn, 1990). Do these same factors influence utilization of interactive media once the initial adoption decision has been made? And, if one wished to encourage the rejection of downloading, rather than its adoption, how could critical mass be reversed?

A different pattern of variables affected intentions to discontinue downloading in the future. Expectations of finding varied music again encouraged downloading, as did self-efficacy in using file sharing software and coping with the risk of detection. The expectation of punishment, however, was related to intentions to discontinue, as were expectations of poor quality downloads and perceptions of social unacceptability. In multivariate analyses, only the expectation of punishment remained as a factor supporting discontinuance, while file sharing self-efficacy remained a negative factor.

After the effects of the other variables were controlled, deficient self-regulation again emerged as a negative predictor of intentions to discontinue file sharing. The bivariate correlation between deficient self regulation and discontinuance was nonsignificant, although in the predicted direction, but was moderately correlated to social outcomes (r (257) = .36, p < .001). Post hoc analysis showed that among those who scored above the median on perceived social outcomes there was no relationship between deficient self-regulation and intentions to discontinue file sharing (r (106) = .10, p = .294). The hypothesized negative correlation (r (108) =−.27) was found only among those with low levels of perceived social outcomes. Thus, the expectation of deriving social benefits from downloading moderated the effect of deficient self-regulation on intentions to discontinue file sharing. We speculate that through social interaction with other file sharers some of those who are most involved in downloading become aware that their behavior is both excessive and potentially harmful and so plan efforts to discontinue it. Or frequent participants in interactions with other music fans might make file sharers more aware of alternatives, such as pay music services and fan sites that don’t provide downloads, which make the prospect of discontinuance more palatable than among those with minimal social interactions. Either possibility could be partially offset by learning how to better avoid detection through social interaction with other downloaders, producing the nonsignificant, positive correlation between deficient self regulation and discontinuance intentions observed in the high social outcome group.

Another possibility is that an unaccounted for suppressor variable may have attenuated the hypothesized relationship between deficient self regulation and future downloading intentions. Media researchers often introduce demographic variables such as gender and age in their analyses, apparently because such variables are so widely used by practitioners in media industries. Within social cognitive theory, however, demographic differences are the result of social cognitive processes. For example, male-female differences in preferred leisure activities should be the result of differences in expected outcomes of those activities produced by differential socialization of males and females (cf. Bussey & Bandura, 1999). The limited range of age and education in the present student sample ruled these out as important variables in the current research, but gender, ethnicity, and household income were included in initial analyses. None of these variables were significantly correlated to downloading activity and so were not retained for regression analysis. Gender, but not ethnicity or household income, had a small negative correlation with intentions to discontinue downloading (r(217) =−.147, p < .05), indicating that males were slightly less likely to discontinue. Gender, though, was not a significant predictor when added to multiple regression analysis, nor did it change the pattern of significant predictor variables.

In breaking the downloading habit, we must ask ourselves: Are fear appeals the answer, and if so, fear of whom? Among the handful (3.4%, or 7 in all) of users in the survey who said they were very likely to give up file sharing in the next month, the industry’s scare tactics may have discouraged two of the heaviest downloaders, including one individual with over 30,000 MP3s on file and another with 2,500. In contrast to this, the other five who were very likely to discontinue downloading had less than the average number of 734 files and other heavy downloaders with 10,000 or more MP3s were found in the “very unlikely to discontinue” group. Thus, for the relatively light downloader, fear may be an effective tactic that may induce some to give up the practice. The fear of industry lawsuits, however, may not be the most effective scare tactic. Rather, the present results suggested that fear of sanctions from one’s university may be more effective.

The weak and insignificant correlations between expected economic outcomes and current or future downloading activity, as well as the weak and insignificant correlation between current or future downloading activity, contradict the often-heard economic explanations of the music industry to the effect that file sharing is really about the unauthorized appropriation of copyrighted music. It could be, however, that beliefs in the economic advantages of file sharing are so widespread that they are not distinguishable between heavy and light users, thus creating a ceiling effect, and it is important to note that the relationship between economic outcomes and current activity did closely approach significance. The present research also did not include questions about CD purchases, so the question of economic competition between free online music and recorded music available on CDs cannot be answered here.

Limitations

It bears repetition that the sample used in this study was not representative of the general population of college students. The present study is an exploratory examination of the downloading phenomenon and of new variables that may inspire future research. College students may also differ from other populations of downloaders. As a group, college students have relatively high levels of depression (Rich & Scovel, 1987), a contributor to poor self-regulation (Bandura, 1991) that may heighten the impact of deficient self-regulation compared to other populations. Survey research (Pew Research, 2002) has shown that students have a particular reliance on the Internet for social interaction and amusement activities, so results in adult populations may differ. The low reliabilities of two of the variables (download quality and social acceptability) and the use of single item measures also limit the reliability of our findings.

Implications

The implications of the present research are limited by the convenience sample that was used, but the findings nonetheless pose some intriguing challenges to current thinking about the downloading phenomenon and methods of controlling it. Specifically, the present results call into question the effectiveness of the bare knuckle legal tactics that are the centerpiece of the current recording industry strategy to counter file sharing. Enlisting universities to do the “dirty work” of tracking down and intimidating file sharers would appear to be more effective than filing lawsuits against isolated users, in that expectations of university reprisals were more strongly related to intentions to discontinue than expectations of law suits. Or universities may opt out of being enforcers by buying group subscriptions to music services, as some have started to do. Another approach might be to make legal pay music services more attractive to downloaders, specifically by highlighting the social aspects of downloading. The new pay version of Napster offers an example. It provides message boards, browsing of user collections, and musical suggestions.

Future research could examine the adoption of pay music services and the conditions under which they supplant free downloading. The disadoption of a widely-performed media behavior in favor of another with two seemingly important relative disadvantages (i.e., no longer free, no longer communal) would make an interesting case study in the (dis)adoption of innovations, a process which has also be described in social cognitive terms (Bandura, 1994). The interaction of CD purchases with both file sharing and pay music services could also be examined through the lens of that theory.

The present model of file sharing might also be improved. The present research neglected important categories of behavioral incentives that could further explain file sharing behavior: status, enjoyable activity, and self-reactive incentives. For example, skill in evading the perils of downloading or the possession of an impressive MP3 collection might confer social status on the user. The use of downloaded music to dispel feelings of boredom or anxiety (a self-reactive incentive in social cognitive terms) or to provide an enjoyable musical experience was also not examined here. Also, what is the impact of vicarious learning? Does reading about RIAA lawsuits against innocent grandmothers have the same effect as knowing a fellow student who has been sued? Following the logic of social cognitive theory, the more that the targets of the lawsuits resemble the downloaders themselves and the more certain downloaders are of being punished, the greater the impact of the scare tactics should be. Finally, as mentioned previously, further dimensions of social norms should be explored. These include beliefs about the desires of specific significant others and social comparisons with individuals and groups engaged in file sharing.

Notes
  • 1

    I can make my own custom mixes; I will get exposed to new music; I can find rare songs.

  • 2

    I am likely to be sued by the industry; I will lose my campus computer account; I might get thrown out of school.

  • 3

    I will have a hard time finding what I want; I will get poor quality files; I waste a lot of time.

  • 4

    I meet new friends online; I feel a sense of community with the people I share with; I get to see the play lists others have so I know what is hot.

  • 5

    I can save money; I can sample music before I buy it; I can have backup copies of the music I own.

  • 6

    The following items were rated on a seven-point Likert-type scale ranging from strongly agree (7) to strongly disagree (1): I sometimes download files without thinking about why I want them; My downloading is out of control; I spend so much time downloading that it is interfering with my life; Downloading is a habit I have gotten into.

  • 7

    I know how to avoid detection by the authorities when I file share; I am confident I won’t get caught downloading illegal files.

  • 8

    Downloading is cutting into music sales; Copyright piracy is morally wrong; Artists are hurt by file sharing; The industry is right to sue people who download illegally.

  • 9

    Downloading music is just like borrowing a CD; High CD prices force people to download music; Everyone else is downloading music. It’s OK for me to do it.

References

  1. Top of page
  2. Abstract
  3. The downloading problem
  4. Understanding downloading behavior
  5. Hypotheses
  6. Research methods
  7. Results
  8. Discussion
  9. References
  • Aarts, H., Verplanken, B., & Van Knippenberg, A. (1998). Predicting behavior from actions in the past: Repeated decision making or a matter of habit? Journal of Applied Social Psychology, 28, 13551374.
  • Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J.Kuhl & J.Beckman (Eds.), Action-Control: From Cognition to Behavior (pp. 1139). Heidelberg: Springer.
  • Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
  • Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall.
  • Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior and Human Decision Processes, 50, 248287.
  • Bandura, A. (1994). Social cognitive theory of mass communications. In J.Bryant & D.Zillmann (Eds.), Media effects: Advances in theory and research (pp. 6190). Hillsdale, NJ: Lawrence Erlbaum.
  • Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.
  • Bargh, J. A., & Gollwitzer, P. M. (1994). Environmental control of goal directed action: Automatic and strategic contingencies between situation and behavior . In W. D. Spaulding (Ed.), Nebraska Symposium on Motivation, Vol. 41 (pp. 71124). Lincoln, NE: University of Nebraska Press.
  • Bonne, J. (2004, January 4). Big drop seen in music downloads. MSNBC.com. Retrieved September 7, 2005, from http://www.msnbc.msn.com/id/3860823
  • Borland, J. (2003, September 10). RIAA seeks to alter file-trading culture. CNETNews.com. Retrieved September 7, 2005, from http://news.com.com/2100-1027_3-5073786.html?tag=st_rn
  • Bussey, K., & Bandura, A. (1999). Social cognitive theory of gender development and differentiation. Psychological Review, 106, 676713.
  • CacheLogic. (2005). First ever, real time traffic analysis of file form crossing peer-to-peer released by CacheLogic. Retrieved September 7, 2005, from http://www.cachelogic.com/news/pr090805.php
  • Carlson, S. (2000, April 28). Metallica sues universities and Napster, charging that students engage in music piracy. The Chronicle of Higher Education, p. A50.
  • Carlson, S. (2003, May 23). A president tries to settle the controversy over file sharing. The Chronicle of Higher Education, p. A27.
  • Charney, T., & Greenberg, B. (2001). Uses and gratifications of the Internet. In C.Lin & D.Atkin (Eds.), Communication, technology and society: New media adoption and uses. Cresskill, NJ: Hampton Press.
  • Chmielewski, D. (2005, August 5). Darknets rising to expand file sharing. Knight Ridder Tribune Business News.
  • Comrey, A. L., & Lee, H. B. (1992). A First Course in Factor Analysis (2nd Edition). Hillsdale, NJ: Lawrence Erlbaum.
  • Conner, M., & Armitage, C. I. (1998). Extending the theory of planned behavior: A review and avenues for future research. Journal of Applied Social Psychology, 28, 14291464.
  • Connolly, T., & Thorn, B. K. (1990). Discretionary data bases: Theory, data and implications. In J.Fulk & C.Steinfield (Eds.), Organizations and communication technology (pp. 219234). Newbury Park, CA: Sage.
  • Doll, J., & Ajzen, I. (1992). Accessibilty and stability of predictors in the theory of planned behavior. Journal of Personality and Social Psychology, 63, 754765.
  • Flanagin, A. J., & Metzger, M. J. (2001). Internet use in the contemporary media environment. Human Communication Research, 27, 153181.
  • Glasner, J. (2003, September 9). Fear may not spur CD sales. Wired.com. Retrieved September 7, 2005, from http://www.wired.com/news/business/0,1367,60350,00.html
  • Kaye, B. K. (1998). Uses and gratifications of the World Wide Web: From couch potato to Web potato. New Jersey Journal of Communication, 6, 2140.
  • Korgaonkar, P., & Wolin, L. (1999). A multivariate analysis of Web usage. Journal of Advertising Research, 39, 5368.
  • LaRose, R., & Eastin, M. (2004). A social cognitive explanation of Internet uses and gratifications: Toward a new theory of media attendance. Journal of Broadcasting and Electronic Media, 48(3), 458477.
  • LaRose, R., Lin, C. A., & Eastin, M. S. (2003). Unregulated internet usage: Addiction, habit, or deficient self-regulation? Media Psychology, 5, 225253.
  • LaRose, R., Mastro, D. A., & Eastin, M. S. (2001). Understanding Internet usage: A social cognitive approach to uses and gratifications. Social Science Computer Review, 19, 395413.
  • Lin, C. A. (1999). Online-service adoption likelihood. Journal of Advertising Research, 39, 7989.
  • NPD Group. (2004, January 16). The NPD Group notes recent increase in peer-to-peer digital music file sharing. Press release. Retrieved September 7, 2005, from http://www.npd.com/press/releases/press_040116.htm
  • Markus, M. L. (1990). Toward a “critical mass” theory of interactive media. In J.Fulk & C.Steinfield (Eds.), Organizations and communication technology (pp. 219234). Newbury Park, CA: Sage.
  • Nunnally, J. (1967). Psychometric theory. New York: McGraw-Hill.
  • O’Keefe, D. J. (2003). Against familywise alpha adjustment. Human Communication Research, 29, 431447.
  • Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124, 5474.
  • Palmgreen, P., Wenner, L., & Rosengren, K. (1985). Uses and gratifications research: The past ten years. In K.Rosengren, L.Wenner, & P.Palmgreen (Eds.), Media Gratifications Research (pp. 1137). Beverly Hills, CA: Sage Publications.
  • Parker, B., & Plank, R. (2000). A uses and gratifications perspective on the Internet as a new information source. American Business Review, 18(2), 4349.
  • Papacharissi, Z., & Rubin, A. M. (2000). Predictors of Internet usage. Journal of Broadcasting and Electronic Media, 44, 175196.
  • Pesola, M. (2004, January 2). U.S. music-sales decline seen slowing. MSNBC.com. Retrieved September 7, 2005, from http://msnbc.msn.com/id/3860239/
  • Pew Research Center (2002, October 20). The internet goes to college: How students are living in the future with today’s technology. Retrieved September 7, 2005, from http://www.pewinternet.org/reports/toc.asp?Report=71
  • Rainie, L., & Madden, M. (2004, January). Pew Internet project and comScore Media Matrix data memo. Retrieved September 7, 2005, from http://www.pewinternet.org/reports/pdfs/PIP_File_Swapping_Memo_0104.pdf
  • Rich, A. R., & Scovel, M. (1987). Causes of depression in college students: A cross-lagged panel correlational analysis. Psychological Reports, 60, 2730.
  • Rosen, H. (2002, October 3). Recording Industry Association of America chairman and CEO letter to college and university presidents. Retrieved September 7, 2005, from http://www.riaa.com/pdf/Universityletter.pdf
  • Rubin, A. (1984). Ritualized and instrumental television viewing. Journal of Communication, 34, 6777.
  • SPSS (2003). Statistical package for the social sciences, Version 12.0. Chicago, SPSS, Inc.
  • Stern, C. (2000, June 13). Record firms say Napster hurt sales. The Washington Post, p. E01.
  • Wimmer, R. D., & Dominick, J. R. (2003). Mass media research (7th Edition). Belmont, CA: Wadsworth.