Illegal Adoption of a New Product: a Model of Software Piracy Behavior

ABSTRACT - This paper is concerned with the decision process of computer software adopters, especially pirates. Based on past research, a model is proposed that contains two general normative constructs that influence the general attitude toward piracy. Other specific variables, e.g., perceived utility of the software, the tendency to rely on complementary products and on others (in order to reduce learning costs) are also included and the impact of all these variables on an individual's propensity to pirate a software program is investigated. Finally, implications, particularly managerial, of this analysis are presented.

Citation:

Madhavan Parthasarathy and Robert A. Mittelstaedt (1995) ,"Illegal Adoption of a New Product: a Model of Software Piracy Behavior", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 693-698.

Advances in Consumer Research Volume 22, 1995      Pages 693-698

ILLEGAL ADOPTION OF A NEW PRODUCT: A MODEL OF SOFTWARE PIRACY BEHAVIOR

Madhavan Parthasarathy, University of Nebraska

Robert A. Mittelstaedt, University of Nebraska

ABSTRACT -

This paper is concerned with the decision process of computer software adopters, especially pirates. Based on past research, a model is proposed that contains two general normative constructs that influence the general attitude toward piracy. Other specific variables, e.g., perceived utility of the software, the tendency to rely on complementary products and on others (in order to reduce learning costs) are also included and the impact of all these variables on an individual's propensity to pirate a software program is investigated. Finally, implications, particularly managerial, of this analysis are presented.

INTRODUCTION

Modern communication and copying technologies have greatly enhanced the potential for the theft of intellectual property. Swinyard, Rinned and Keng Kau (1990) estimate that it now accounts for an annual loss to the American economy of over $40 billion. One form of this activity in which most average consumers can participate is the pirating of software. In spite of the impact of piracy has on computer software adoption, little attention has been given to this form of consumer behavior or to the consequences which follow from it.

The purpose of this paper is to investigate the variables which affect an individual's propensity to pirate. A model is proposed that contains general attitudinal and normative constructs regarding piracy; other constructs are suggested by Conner and Rumelt (1991) and Parthasarathy and Hampton (1993). These authors suggest that non-monetary costs associated with piracy (e.g., the time spent learning a specific software package and the efforts expended to acquire it illegally) could often overwhelm the monetray costs associated with the software. One of the objectives of this paper is, therefore, to ascertain if such non-monetary costs are systematically evaluated by individuals when they make a decision whether or not to pirate a particular software package. Thus, the proposed model contains general constructs regarding normative and attitudinal factors, and specific factors regarding the particular software being considered that, together, are expected to influence an individual's propensity to pirate a particular software package. The model is then tested using structural equation modelling with LISREL-7. Finally, the implications of the model are discussed.

THE NATURE OF PIRACY BEHAVIOR

For the purpose of this study, software piracy is defined as the act of making a nonlegal copy of a copyrighted or patented software for one's own use. Thus, it is a form of adoption behavior whose alternative is purchase. On the face of it, piracy carries a zero (or nearly zero) direct price to the pirate. However, there are other nonmonetary costs involved in adopting a software; to receive the benefit of the product one must invest time and effort in acquiring the copy and, more importantly, learning how to use it.

Parthasarathy and Hampton (1993) divided nonmonetary costs into those related to acquiring the software and the those related to learning the software. The cost of acquiring the software may be significant. The would-be pirate must locate a copy of the software and, possibly, some further software which makes it capable of being copied in a usable fashion. Since the person from whom one copies the program may have purchased it, and since pirating is both illegal and unethical, there may be some embarassment or other negative feeling associated with the behavior. Note that these costs are born only by the pirate, not by a purchaser. These costs would be lower if a person knew others from whom he/she could copy the software. Thus, interpersonal influence and support among the adopters should be an important factor in deciding whether one pirates or purchases.

In a similar vein, learning a new software may involve significant nonmonetary costs. Manuals, tutorials, and on-line help services help to reduce these costs to the purchaser; generally, the cost of providing this form of help is part of the price of the software and, thus, is monetized to the purchaser. For the would-pirate, the market often provides tools such as independently produced manuals that make learning the pirate software easier (called complementary products). Friends who are well versed with the working of the program may also help reduce nonmonetary costs. However, the nonmonetary costs will be greater for the pirate and, for a relatively new and complex program, are likely to be very much higher.

In addition to the nonmonetary costs, piracy is a behavior which is both illegal and unethical, and the decision to pirate would be expected to be influenced by a person's subjective norms about the behavior. In sum, piracy is a conscious adoption decision in which the benefit of owning the software is influenced by the nonmonetary costs of using the product and the subjective norms about the nature of the behavior itself.

THE PROPOSED MODEL

General variables: The model on which this research is based contains elements corresponding to: the behavioral intention, the propensity to pirate (PP); the attitude toward piracy (APIRACY); and two subjective norms (NORMS1) and (NORMS2). NORMS1 involves the perception of the ethical nature of piracy itself. NORMS2 recognizes the possibility of "situational ethics" being involved in the decision. Preliminary research suggested that some people believe that the pricing policies of software developers are, themselves, unethical. NORMS2 captures these beliefs, which a person might use to justify piracy to him/herself.

This part of the model has some similarities to the theory of reasoned action (Ajzen and Fishbein 1980; Fishbein and Ajzen 1975) in that it contains a behavioral intention construct as well as attitudinal and normative constructs. However, in contrast to the original theory of reasoned action, the model proposed in this paper presumes that attitudes are largely a function of the normative factors. Indeed, research supports the contention that the theory of reasoned action is inadequate when applied to moral behavior and other behaviors (e.g., contraceptive use) that are largely private and concerned with non-visible consumption (e.g., Doll and Orth 1993; Vallerand et. al. 1992). These authors argue that in such a situation the normative factor would causally relate to the attitudinal factor since attitudes are largely a function of the norm. In addition, the normative variables used in this research are concerned with ethical beliefs and are therefore qualitatively different from those prescribed by the theory of reasoned action. Further, the attitudinal component is a general component (not specific to the particular software as would be indicated by the reasoned action model). Since the normative factors used in the model are general and will not vary between software packages, the attitudinal construct (which is a function of these normative beliefs) will have more meaning if meaured generally rather than specifically. However, the authors realize that specific factors associated with the particular software being tested could have strong influence on an individual's propensity to pirate that software package. Therefore, the model proposed also incorporates several specific factors, as discussed below.

FIGURE 1

PROPOSED MODEL

Specific Factors: In addition to the general variables mentioned above, the proposed model includes two variables representing the nonmonetary costs of piracy. The perceived availability and usefulness of complementary products (as defined above) is incorporated into the variable "reliance on complementary products" (RCOMP.) The availability of informed other persons, and one's willingness to ask them for help, is incorporated into the variable "reliance on others" (ROTH.) A final variable (UTILITY) captures the perceived usefulness of the particular software under consideration.

Thus, the proposed model contains three general decision elements, the desirability of the particular software, the magnitude of nonmonetary costs of acquiring and learning how to use it, and the attitude toward piracy, with its normative components. Figure 1 shows the proposed model and the direction of the hypothesized relationships among the several constructs. The relationships among these variables are discussed, taking the three general decision elements in order.

Nature of the product: The greater the perceived desirability of the software, the more a consumer would wish to adopt it for use, thus enhancing the utility of both piracy and purchase. However, it is likely that, as a particular program is seen as being more desirable, its value to the purchaser is enhanced more than its value to the pirate. The key to this prediction lies in the nature of nonmonetary costs.

To the extent that programs which are more useful are also more powerful, the nonmonetary costs of adopting them are also rather high. Just as so-called "shareware" programs are relatively low-powered and generally very easy to use, powerful word processors or spreadsheet programs come with, and require, a sizeable package of support services. Thus, for highly desirable software packages, the nonmonetary costs are likely to be very high for the would-be pirate. Further, what often enhances the perceived utility of the program is the presence of the factors which are available to the purchaser as part of the program's price. Given this, greater desirability is likely to increase the value (i.e., the ratio of program benefit to the total of monetary and nonmonetary costs) for the purchaser more than for the potential pirate. Therefore, it would be expected that perceived utility of the particular program (UTILITY) would be inversely related to the propensity to pirate (PP).

Conversely, the greater the perceived usefulness of the software, the greater will a person be inclined to use tutorials to learn the software and ask others for help about it. Therefore, UTILITY should be expected to be positively related to both the reliance on others (ROTH) and the reliance on complementary products (RCOMP).

Attitudes toward Piracy: In keeping with the theory of reasoned action, the subjective norms about piracy (NORMS1 and NORMS2) combine with the generalized attitude toward piracy to affect the propensity to pirate in a straightforward way. Further, because the presence of knowledgable others and/or complementary products are almost necessary for one to make use of pirated software, the willingness to use these sources of help become the "situational" attributes which make "situational ethics" possible. Therefore, the subjective norms about piracy might be expected to be related to both the reliance on others and the reliance on complementary products.

TABLE 1

QUESTIONNAIRE ITEMS

Reliance on Others and Complementary Products: Both the availability of complementary products and the presence of knowledgable others enhance the ability of the pirate to successfully use the product. Both would be expected to exert a positive influence on the propensity to pirate. However, the presence of others may substitute for complementary products, and vice versa; in this sense these two factors may be substitutes for each other and the relationship between the two could be inverse. As noted before, these variables would be affected by the perceived usefulness of the software and the subjective norms about pirating.

TESTING THE MODEL: METHOD

Data were supplied by a sample of 205 undergraduate students at a large Midwestern university. Sixty four respondents failed to answer the questionnaire completely (i.e., they skipped at least one question) and were excluded from the sample in order to keep the analysis unbiased. LISREL is especially sensitive to this issue and, since the sample size after excluding the incomplete questionnaires was still reasonably high (141), this approach was judged by the authors to be the most unbiased. The questionnaire presented the respondents with a hypothetical scenario involving a software package titled Integra that combined word processing, spreadsheet, and database applications in a single integrative package. Since most of the respondents were senior level students, they could identify with the usefulness of the package. It was mentioned that the university was trying to obtain student discounts that would make the package available to the students at $100. Also, it was mentioned that the package was not copy protected, that it was relatively easy to copy, and that the risk of being apprehended in the act of piracy was nil. The aim was to create a hypothetical scenario that reflected, as best as possible, a typical real-world situation. Following the introductory scenario, the the questionnaire requested the respondents to complete Likert type questions for each of the constructs used in the model. There were 26 items on the questionnaire corresponding to the constructs of interest. However, after an initial factor analysis (some items loaded strongly on more than one construct) and reliability tests (items with less than 0.7 coefficient alpha were dropped), sixteen items were used to measure the seven constructs, with each construct being assigned at least two items. As table-2 suggests, the chosen items loaded very strongly with their assigned constructs. The constructs and their respective items are contained in Table 1.

The Figure depicts the constructs and their relationships in graphical form. This model was tested using LISREL-7.

RESULTS

The results from the LISREL analysis are presented in Tables 2 and 3. Table 2 is concerned with the relationship of the constructs to their respective items (i.e., Lambda X and Lambda Y matrices and their significance levels, based on t-values). As an indication of construct validity, all the items are significant and of the expected sign.

Table 3 presents the overall fit of the model as well as the path coefficients and their respective significance levels. Of specific concern are the model chi-square, the goodness-of-fit index (GFI), the adjusted-goodness-of-fit-index (AGFI), and the root mean squared residual (RMR).

TABLE 2

ITEM-CONSTRUCT RELATIONSHIP

The chi-square value, though not of great concern in LISREL analyses, is (112.94 / df=88), thus indicating a reasonable fit of the model to the data. Traditionally, the GFI and RMR have been used as indicators of a model's fit (Tanaka, 1993). Bagozzi and Yi (1988) suggest that the GFI should be above 0.9 and the RMR below 0.05. Both these requirements are satisfied in this model, adding substance to the notion that the model is indeed a good fit.

The paths emanating from UTILITY were all in the correct direction, and those terminating at ROTH and RCOMP respectively were also highly significant. The negative relationship between UTILITY and PP was forecasted, although its strength was weaker than predicted.

A strong path from NORMS-1 to APIRACY was expected and confirmed. It suggests that attitude toward piracy is a function of existing long-term normative perceptions of the ethical nature of piracy (which are likely to vary between societies).

The paths emanating from NORMS-2 were all insignificant although in the direction forecasted. This strongly suggests that "situational ethics" are overshadowed by long-term perception of the nature of piracy itself. Put differently, the perception that computer software packages are overpriced and that software manufacturers make outrageously high profit margins does not seem to significantly enhance (or detract from) an individual's attitude toward piracy, or his/her reliance on either complementary products or on others.

The strong negative path between NORMS-1 and RCOMP was, however, surprising and not forecasted. While pirates have greater nonmonetary costs of adoption and therefore should be positively inclined toward the use of complimentary products, the results suggest that those who normatively support piracy are less likely to rely on complementary products. Perhaps their inclination to save money (by not purchasing complementary products) overshadows their desire to reduce the nonmonetary costs of learning. It must also be remembered that the subjects were students who can easily find help with the operation of the software product, including that from friends. Additionally, most universities have help desks and lab assistants who can help. Therefore it seems reasonable to expect that in spite of a positive inclination to pirate, some of these individuals may favor reliance on others over reliance on complimentary products, especially since the former is free. The fact that most students are extremely price sensitive may add fuel to this contention. The notion that such individuals are likely to seek help from others is supported by the positive path between NORMS-1 and ROTH.

The path from APIRACY to PP, and from ROTH to PP were both positive and significant as predicted, suggesting that positive attitudes toward piracy as well as the degree of reliance on others is directly, and strongly, associated with a person's propensity to pirate. However, the path from RCOMP to PP, though positive as predicted, is insignificant, indicating that the use of complementary products does not necessarily result in enhanced piracy. There could be two reasons for this phenomenon: (1) the pirate's desire to save money (by not purchasing the complementary products) overrides his/her desire to ease the time and cost of learning (nonmonetary costs), and (2) those individuals who have a low propensity to pirate (likely purchasers) also purchase complementary products. Although nonmonetary costs are likely to be higher for the pirate than for the purchaser, they may be quite high for both of them (especially for a relatively complex package like Integra) and therefore the purchaser may also benefit strongly from the use of complementary products.

Finally the strong and negative path from RCOMP to ROTH confirms the prediction that individuals who rely more on complementary products rely less on others. To a large extent, these sources of reducing the nonmonetary costs of adoption are substitutable for each other; an individual who uses complementary products to learn the package may not have the need to seek help from others.

TABLE 3

RESULTS OF THE PATH MODEL

Although the model described above was satisfactory from a goodness-of-fit perspective, several theoretically meaningful paths were insignificant. While these may have enhanced understanding of the pirate adoption process, the quest for an even more parsimonious solution led to a partial trimming of the model through the removal of the NORMS-2 construct and the three nonsignificant paths that emanated from it. The results were more or less the same as that for the a priori model in that the significant paths in the previous model continued to remain significant (at the same levels) in the trimmed model and the insignificant paths (e.g., from UTILITY to PP and RCOMP to PP) were still insignificant. However, the chi square value dropped to 80.60 (p=0.014), the GFI went up to 0.921, the AGFI to 0.870, and the RMR to 0.041. This suggests that the trimmed model was a slightly better fit than the original one, although it changed none of the relationships.

DISCUSSION AND CONCLUSIONS

The general outline of the proposed model was confirmed. The propensity to pirate is influenced by the potential pirate's attitudes toward piracy, subjective norms about the act of piracy, the perceived utility of the software package, and the willingness to seek help from others or complementary products to reduce the nonmonetary costs.

Of course, the generalizability of the results is limited by the nature of the sample. While students are software users, their demographic and cultural homogeneity is likely to restrict the range of observed attitudes and, especially, subjective norms about piracy. Indeed, about three fourths of the respondents were willing to pirate the software in question and about two thirds saw no ethical problems with piracy. However, the strong effects of the desirability of the product itself and the role of the means for reducing nonmonetary costs suggest that there is little reason to believe that the nature of the relationships among the constructs would change much in a different sample.

To the extent the general nature of the model holds, the implications for software producers are not encouraging. First, the belief that software prices are unjustified did not appear to affect attitudes toward piracy nor encourage the propensity to pirate. Promotional efforts to justify prices would appear to be a relatively ineffective strategy to reduce piracy. Second, the a-priori model hypothesized that the relationship between perceived product utility and piracy would be negative, suggesting that enhancing product utility would be a piracy reducing strategy. However, the relationship, although in the right direction, was not significant. Rather, it seemed to affect the propensity to pirate only by acting through the constructs which reflect the ability to reduce the nonmonetary costs of pirating.

This brings us to the final issue; the roles of complementary products and knowledgable others appear to have a strong influence on piracy. Obviously, as a software product gains in popularity, the availability of complementary products and the likelihood of finding someone who can be of help in learning how to use the product increase. Further, one of the strongest paths in the tested model found that these two are seen as substitutable means for reducing nonmonetary costs. While a software manufacturer might take some means to limit the availability of complementary products, there is no way to limit access to helpful other persons. This may be another example of the winner's curse; one of the penalties of success in the software industry appears to be the increased likelihood of piracy.

REFERENCES

Ajzen, Icek and Martin Fishbein (1980), Understanding Attitudes and Predicting Social Behavior, Englewood Cliffs, NJ: Prentice Hall.

Bagozzi, Richard P. and Youjae Yi (1988), "On the Evaluation of Structural Models," Journal of the Academy of Marketing Science, 16 (Spring), 74-94.

Conner, Kathleen Reavis and Richard P. Rumelt (1991), "Software Piracy: An Analysis of Protection Strategies," Management Science, 37 (February), 125-39.

Fishbein, Martin and Ichek Ajzen (1975), Belief, Attitude, Intention,Behavior: An Introduction to Theory and Research, Ma: Addison-Wesley.

Parthasarathy, Madhavan and Ronald Hampton (1993), "The Role of Piracy in the Diffusion of a Software Product: A Propositional Framework", paper presented at the Winters Educators Conference of the American Marketing Association, Chicago: IL.

Sheppard, Blair H., Jon Hartwick, and Paul R. Warshaw (1988), "The Theory of Reasoned Action: A Meta-Analysis of Past Research with Recommendations for Modifications and Future Research", Journal of Consumer Research, 15 (December), 325- 343.

Swinyard, William R., Heikke Rinne and Ah Keng Kau (1990), "The Morality of Software Piracy: A Cross Cultural Analysis," Journal of Business Ethics, 9 (August), 655-64.

Tanaka, Jeff S. (1993), "Multifaceted Conceptions of Fit in Structural Equation Models," in Testing Structural Equation Models, Sage Publications, Bollen and Long (eds), 11-39.

----------------------------------------