Closing the Technology Adoption

Abstract

A firm may readily subscribe to a new technology but then fail to use it. This article advances existing technology diffusion theory by bringing in a new construct that can explain the likelihood of technology use after adoption. The authors define contiguous user bandwagon and show how this information diffusion mechanism can help in explaining the time to technology use. They test their hypotheses using data on the adoption and use of e-procurement technology (N = 3,158) in the early phase of its diffusion. The authors find support for the hypothesis that contiguous user bandwagon is a strong antecedent of time to technology use.

Keywords

technology adoption–technology use divide contiguous user bandwagon information technology adoption and use

New technologies have the potential to trigger many changes: They can change existing firms’ business models and change the level and characteristics of demand, and they can greatly affect the competitive positions of different industry players. Indeed, the introduction of new technologies can sometimes mark the beginning of the end for established companies that have been successful for decades (Christensen, 1992) and trigger the emergence of whole new industries with a corresponding wave of new entrants (Anderson & Tushman, 1990). Existing literature has explained the diffusion of technology by emphasizing the role of information diffusion (Abrahamson & Rosenkopf, 1993, 1997; Ajzen, 1991; Bass, 1969; Davis, 1989; Rogers, 2003; Venkatesh, Speier, & Morris, 2002), that is, the process by which information about an innovation is transmitted (Rogers, 2003). Yet, despite vast research to date—Rogers counts more than 5,200 articles on the topic—a number of issues need to be investigated more carefully if we are to get a better understanding of technology diffusion. These issues have to do with important assumptions that are incorporated, explicitly or implicitly, in the existing literature on technology diffusion. We have identified two such areas of concern that we address more formally in this article.

First, the existing technology diffusion literature has, for the most part, equated technology adoption (i.e., the purchase of a technology) to technology use. For instance, Geroski (2000) states that “early adopting individuals (or firms) have evidently chosen to use the technology” (p. 606). A closer examination of companies suggests that technology use does not necessarily follow from technology adoption. In many industries, new technologies are sometimes adopted and then used very little or not at all. Indeed, this practice is so common in software that the business press has coined the phrase “shelfware”—software that, once purchased, is put on a shelf and never used (“At your service,” 2003). In other words, use after adoption might even be the exception rather than the rule, at least in some industries and for some technologies. The scant attention to the differences between technology adoption and technology use implies that existing technology diffusion literature has also failed to identify and distinguish the type of organizational actors that are involved in the separate decisions to adopt and to use a new technology. As we elaborate below, the organizational actors behind these decisions tend to be different and have different responses to information diffusion mechanisms.

Second, despite significant advancements in the management literature on the dynamics of technology, existing technology diffusion theory implicitly assumes that a technology being launched and adopted in the market emerges at once in its final form. This treatment of technology as an exogenous variable that “appears” in the market at one point and does not change with time should come as no surprise if we consider that the theoretical roots of technology diffusion theory are in economics. Traditional economics theory has long been criticized for providing such exogenous treatment of technology (e.g., Nelson & Winter, 1982), and many of the pioneer authors in the study of technology diffusion used economic tools and economic perspectives to build their frameworks. However, modern economics and management theory has emphasized that technology does evolve over time (Abernathy & Utterback, 1978; Christensen, 1992; Dasgupta & Stiglitz, 1980; Kahl, 2007; Langlois, 1992). Even after a new technology is introduced, R&D investment continues and firms keep improving on the existing technological designs (Anderson & Tushman, 1990; Utterback & Suarez, 1993). Technologies evolve through a “design hierarchy” wherein key decisions set an improvement trajectory for other less critical decisions to come (Clark, 1985). Indeed, the early expressions of a technology in the market are often “rudimentary” and can differ substantially from the technology form that ends up being adopted by a larger market in a later stage (Foster, 1986). We argue that by failing to incorporate the notion that technology evolves over time, the existing technology diffusion literature misses an important fact: the value that organizational decision makers attach to a specific piece of information changes over time.

In this article, we tackle these two areas of concern by bringing to technology diffusion literature a new construct, contiguous user bandwagon, that we argue can help explain the time to technology use after adoption has taken place. Going beyond the traditional treatment of bandwagons in the management literature (e.g., Abrahamson & Rosenkopf, 1993; Fiol & O’Connor, 2003), we define contiguous user bandwagon as the number of new users of the technology at the time of adoption by a firm. This construct is an important addition to the literature for at least four reasons. The first reason stems from the observation that it formally incorporates into the technology diffusion literature the notion that different organizational actors are involved in the decisions to adopt and use a technology—senior management and the organization’s technical layer, respectively. We expand upon existing technology diffusion theory (e.g., Rogers, 2003) by arguing that the traditional adopter bandwagon mechanisms (introduced to explain technology adoption) should be complemented with those related to user bandwagon to explain technology use. User bandwagons represent information conveyed from actual to potential users of a technology and are likely to activate rational-efficiency or fad-based mechanisms (Abrahamson & Rosenkopf, 1993) that act as powerful enabling factors for an organization’s decision to use a technology after having adopted it (e.g., Attewell, 1992; Burt, 1982; Geroski, 2000; Rogers, 2003). We posit that user bandwagons (as opposed to adopter bandwagons) influence the time elapsed between the adoption of a technology and its actual use.

The second reason stems from our observation that the value of a given piece of information decreases with time. This observation implies a nontrivial change because most of the existing diffusion theory is built upon notions of “cumulative” bandwagon effects, in other words, effects that start to accumulate from the first day the technology is in the market up to the time of an agent’s decision. For instance, the notion of network effects and the resulting “excess inertia” advantages (insurmountable advantages to those players who have amassed a large installed base of users; see Farrell & Saloner, 1985) rest precisely in this “timeless” notion of cumulative bandwagons. In this traditional notion, each adopter, no matter when or where it adopts, has the same informational value for today’s decision-making agent. However, recent literature has started to improve upon this notion. For instance, Suarez (2005), studying the process of mobile telecommunication standards adoption, suggests that not all adopters weigh the same in the decision of an agent; in other words, not all adopters have the same informational value. Information obsolescence is not exclusively produced by technological change but is particularly marked in fast-moving industries. In particular, as technology and the way technology is used change over time (Kahl, 2007), information from periods further back in time may become less relevant for today’s decisions. It follows that our contiguous user bandwagon construct can speak to the fact that prospective users are more likely to base their use decision on information coming from recent users of the technology.

The third reason follows from the observation that by implicitly equating adoption and use, technology diffusion theory overlooks the fact that the competitive implications of technology adoption and technology use can be quite different. As noted above, technology adoption will not improve a firm’s competitiveness unless the adopted technology ends up being used. Moreover, if technology adoption requires a large investment that does not finally result in technology use, the effect can even be negative. For instance, FoxMeyer Corporation is reported to have gone bankrupt due mainly to a failed and costly enterprise resource planning (ERP) implementation (Mabert, Soni, & Venkataramanan, 2001). Therefore, in addition to extending the existing diffusion theory, our focus here on technology use bandwagons has important implications for organizational performance. Senior managers often decide on the adoption of new technologies and can have some degree of influence over technology use decisions via changes in management practices such as incentives and training. By improving their understanding of the external information diffusion mechanisms that influence the use of a technology within their organizations, senior managers can develop even more effective strategies to foster usage and avoid wasteful spending.

Fourth, by explicitly bringing technology use into the technology diffusion literature, we shed additional light on technology diffusion patterns. Only technologies that come to be used can have a long-lasting diffusion. Our theoretical extension here prompts researchers to consider jointly adoption and usage dynamics if we are to improve on the overall predictive power of diffusion theory. For instance, our theory can help explain why a technology, despite enjoying phenomenal initial adoption, may then fail to achieve wide diffusion if it fails to attract users.

We hypothesize below that the stronger the contiguous user bandwagon, the shorter the time to technology use, after adoption has occurred. We test our hypotheses in the context of early e-procurement diffusion, a technology introduced in the mid- to late 1990s. The term e-procurement describes an information technology that enables the use of electronic marketplaces in different stages of the purchasing process, from identification of requirements to payment and contract management. In this article, e-procurement technology refers to a prepackaged standard software product used to facilitate an organization’s interaction with such electronic marketplaces. We have data on 3,158 firms that adopted e-procurement technology from October 1999 to November 2000 (59 weeks), and we observed them until May 2002. Therefore, the overall observation period spans more than 136 weeks (October 1999 to May 2002). In this time period, e-procurement technology was still in its infancy and rapidly evolving (Hoffman, Keedy, & Roberts, 2002). For the purpose of our analyses, we say that there is technology use when any of the functionality embedded in the adopted e-procurement software package has been used for the first time. We say that there is technology adoption when a firm purchases the e-procurement software. Data about firms’ adoption and use were obtained from one of the largest European e-procurement providers, which agreed to collaborate with our study. We find that only about 13.5% of the sampled early e-procurement adopters end up being also early users. This is consistent with the shelfware story above and lends additional support to the main concern of this article. We find strong empirical support that contiguous user bandwagon is a powerful antecedent of time to technology use. We draw managerial and policy implications from our theoretical propositions and empirical findings.

Theory and Hypotheses

Technology Diffusion Literature

The spread of a new technology in a market or user community is commonly referred to as diffusion (Cooper & Zmud, 1990; Loch & Huberman, 1999). The technology diffusion literature has pointed to information diffusion as a key force that enables the spread of new technologies. In particular, technology diffusion—often measured as adoption rates or adoption time—has been defined as a function of mass media communication (Fourt & Woodlock, 1960) and information diffusion (Bass, 1969; Mansfield, 1961). Information diffusion processes can take different forms (Geroski, 2000), including broadcasting and information provision, epidemics and word-of-mouth processes, and information cascades.

Building upon research findings in economics, sociology, and cognitive and behavioral theories, Abrahamson and Rosenkopf (1993) and Fiol and O’Connor (2003) have introduced the construct of bandwagons into technology diffusion theory. Bandwagons refers to a positive feedback loop in which increases in the number of adopters create a stronger bandwagon, and a stronger bandwagon, in turn, causes further increases in the number of adopters (Abrahamson & Rosenkopf, 1997). Bandwagon-influenced behaviors have been described as ranging from rational behavior based on positive externalities (e.g., Katz & Shapiro, 1986) to behavior to conform with the sheer number of organizations that have already adopted the technology (e.g., Abrahamson & Rosenkopf, 1993; Meyer & Rowan, 1977; Strang & Macy, 2001; Tolbert & Zucker, 1983). Many researchers have used bandwagon theories to explain technology diffusion for different technologies, industries, and geographical zones (Gurbaxani, 1990; Kumar & Kumar, 1992).

Existing diffusion literature has also pointed to two other clusters of variables to explain technology diffusion patterns. First, research has shown that the diffusion of a specific technology depends on some features of the technology itself, such as its relative advantage vis-à-vis existing technologies (Loch & Huberman, 1999), the technology’s compatibility with existing products (Farrell & Saloner, 1985; Katz & Shapiro, 1994) and complementary technological infrastructures (Katz & Shapiro, 1994; Shy, 2001), the technology’s complexity, and its ability to be trialed and observed (Rogers, 2003); diffusion also depends on whether the new technology is a product or a process technology (Bass, 1969; Cabral & Leiblein, 2001). Second, existing literature has pointed to firm resources and capabilities to explain technology diffusion, but their precise role remains a subject of controversy. Some studies have postulated a positive relationship with technology adoption timing (i.e., the higher the level a firm’s resources, the later a firm adopts a new technology), while others have proposed a negative relationship. The argument for a positive relationship suggests that firms with a high level of resources emphasize formal roles and control systems and tend to become more rigid. Bureaucracy research (Blau, 1970) and organizational ecology studies (Hannan & Freeman, 1989) concur, indicating that the level of firm resources, often operationalized as a firm’s size, is related to higher organizational inertia, higher formalization and standardization, and more structural rigidity. Christensen, Anthony, and Roth (2004) suggest that, as organizations grow, they increasingly rely on processes that over time become embedded in hard-to-change organizational routines and values. These conditions prevent large, well-endowed firms from being early adopters of technology. The argument for a negative relationship suggests that resource-rich organizations are more likely to be early adopters of technology because of slack resources (Nohria & Gulati, 1996) or formal innovation management practices (Van de Ven, 1986) or because their resources translate into higher absorptive capacity (Cohen & Levinthal, 1990).

Extending the Theory of Technology Diffusion

One major area of concern in the existing technology diffusion literature is that it has tended to equate technology use with technology adoption when in fact these are two distinct phenomena that respond to different dynamics. We argue that extending technology diffusion theory to account for the distinction between adoption and use is not straightforward, as each decision involves different organizational actors. These differences may be large enough to grant a more careful theoretical and empirical treatment. Indeed, it is through using a new technology that a firm can trigger changes in its existing business models, value chains, and interfirm relationships. We therefore argue that technology diffusion theory can increase its value and predictive power if we widen its scope to encompass the antecedents of technology use. The technology diffusion and technology use literatures have evolved by and large independently, and although both can be said to rely on information diffusion, the specific mechanisms by which each of them occur are quite different.

Contiguous User Bandwagon and the Time to Technology Use

In existing diffusion theory, information bandwagons typically refer to the number of adopters, without consideration to how many of those adopters come to actually use the technology. For instance, Katz and Shapiro (1986) and Shy (2001) outline that the extent of adoption externalities depends on the expected final network sizes—a phenomenon that is often also called “adopter bandwagon” (Abrahamson & Rosenkopf, 1993). However, as noted earlier, within an organization the decisions to adopt or use a technology are typically made by different stakeholders. For instance, Leonard-Barton and Deschamps (1988), in their study of the introduction of a new software package (an “expert system”), make a clear distinction between the “top management ‘authority decision’ to adopt the innovation” and the “target end-user’s adoption decisions” (p. 1253). Similarly, after surveying many cases of ERP implementation, Mabert et al. (2001) conclude that the “adoption of ERP was generally a top-down decision” (p. 75). Senior managers tend to be responsible for the decision to adopt a new technology because adoption requires the approval of significant capital expenditures (e.g., ERP, customer relationship management) and sometimes even a strategy change that can be endorsed only by an organization’s senior level. Thus, their decisions to adopt a new technology tend to be influenced by adopter bandwagons and are typically based on rational efficiency (Westphal, Gulati, & Shortell, 1997), the “symbolic value” of the new technology (DiMaggio & Powell, 1983; Meyer & Rowan, 1977), “managerial improvisation” (Orlikowski, 1996), call options (Miller & Folta, 2002), or simply “me too” behavior (Strang & Macy, 2001).

However, when it comes to using a new technology, the key actors are not senior managers but the “technical layer” of the organization (Bacharach, Bamberger, & Sonenstouhl, 1996)—that is, those who are asked to replace an existing technology and associated processes with new technology, processes, and routines. More often that not, there is a marked disconnect between the senior management’s cognition and the technical layers’ beliefs as to the real need to adopt the new technology and the associated costs and benefits of using it. In his study of how organizations adopt total quality management (TQM), Zbaracki (1998) notes that “after leaders decide to implement TQM, they pass it to other members in the organization” (p. 613). He then describes the frustration of the “other members” when they try to integrate TQM into their daily routines. Similarly, Leonard-Barton and Deschamps (1988) note that the use of a technological innovation is a process that involves “numerous individual ‘secondary’ adoption decisions by target users even after successive layers of management have passed along the ‘authority decision’” (p. 1253). In spite of specific incentives or practices that senior management may put in place to promote the use of a new technology, existing literature suggests that the technical level in charge of putting the new technology to use tends to be cautious and conservative and often does not respond as planned to traditional incentives (e.g., Bresnahan, Brynjolfsson, & Hitt, 2002; Edmondson, Bohmer, & Pisano, 2001; Leonard-Barton, 1992). There is a powerful reason for the technical layer’s reluctance: New technologies disrupt existing roles and routines and are surrounded by high uncertainty and therefore tend to entail high costs for the technical layer (Black, Carlile, & Repenning, 2004). It follows from our argument above that the technical layer of the organization tends to be very careful when it comes to making a decision whether to undertake the significant cognitive and organizational costs necessary to use a new technology.

Adopter bandwagons can point to the existence of a new technology and the fact that other companies have purchased it, but they fail to transmit relevant information regarding the use of the technology (Attewel, 1992). Prospective users in the technical layer of the organization look for information that can speak of the “expected personal outcomes of adopting the innovation. . . . ‘What will its advantages and disadvantages be in my situation?’ ‘How complex will the innovation be for me to use?’” (Agarwal & Prasad, 1998). We argue that users place great value in information coming from other users of the same technology when it comes to assessing the uncertain costs and benefits of using a technology. For instance, prospective users (the technical layer) of complex technologies such as ERP often require vendors to arrange for visits to existing ERP deployments in order to talk directly with other users about their implementation experience. These arguments can also be explained with structural equivalence theory (Burt, 1987; Hedstrom & Swedberg, 1998) that predicts actors’ behavior based on the set of “linkages,” not necessarily ties, existing among actors. Burt (1987) argues, “Two people identically positioned in the flow of influential communication will use each other as a frame of reference for subjective judgments and so make similar judgments even if they have no direct communication with each other” (p. 1293).

We argue that the decision to use a new technology will be affected by a different kind of bandwagon, which we call user bandwagon—that is, the number of technology users.

As shown above, a second area of concern in existing diffusion literature is the treatment of technology itself. Contrary to what existing diffusion theory implies, a new technology does not remain unchanged once it appears in the market, nor is it used always in the same way. Management scholars have long studied the evolution of technology, with particular emphasis on the implications for firm survival and performance. Abernathy and Utterback (1978) characterized the early phase of technology as “fluid,” a phase marked by a high rate of innovation in product features and architecture. Anderson and Tushman (1990) conveyed the same idea when describing the early phase of technology diffusion as an era of “ferment.” During the era of ferment, technology evolves constantly, and technological change adds to the uncertainty that surrounds technology-related decisions by organizations. This uncertainty makes it especially hard for users to commit to technology-specific learning (Carpenter & Nakamoto, 1989; Schmalensee, 1982). As time goes by and technology evolves, new information becomes available that reduces the level of uncertainty surrounding the new technology; for example, particular variations of the new technology may be selected out of the market, or new, improved technologies may be introduced. Not only does technology change over time, but the way that people use it also changes. Kahl (2007) argues that use is a learning mechanism that can generate knowledge to reinforce a particular use as well as generate knowledge about different uses. In this context, information that reduces uncertainty is valued highly, particularly by the technical layer that has to commit significant personal resources to use the new technology effectively.

We argue that, during the diffusion of a new technology, potential users tend to place greater value on new bandwagons, that is, bandwagons formed in the time periods just prior to the adoption decision. We call these bandwagons contiguous bandwagons. We argue that prospective users will be positively influenced by contiguous bandwagons because these may signal new, up-to-date information and convey less uncertainty about the technology’s true potential benefits.

Our focus on contiguous bandwagon expands upon and complements the conventional focus on the cumulative level of adopter bandwagon that has traditionally been associated with technology diffusion (e.g., Abrahmson & Rosenkop, 1993). By considering the cumulative level of adopters from all past periods, existing literature implicitly assumes that information from different periods has the same importance for the organizational actor making a decision about a new technology. We argue that this is not the case.

Combining user bandwagon and contiguous bandwagon, we define the construct of contiguous user bandwagon as the number of new users of the technology at the time of a firm’s technology adoption. As elaborated above, contiguous user bandwagon should act as powerful antecedent of time to technology use because (a) the information relates to users, not adopters, of the technology (as we discussed in the previous paragraph), and it is therefore considered more relevant and reliable by prospective users; and (b) the information is “new,” that is, relates to users realized during the time period of a firm’s technology adoption decision. Hypothesis 1 follows:

Hypothesis 1: The time between technology adoption and technology use (time to technology use) will be inversely related to contiguous user bandwagon.

From our arguments above, it follows that a prospective user’s behavior may also be positively influenced by specific contiguous user bandwagons that relate to users that share similar “structural” characteristics with the prospective user (e.g., Burt, 1987; Lazarsfeld & Merton, 1964). Abrahamson and Rosenkopf (1993: 493) argue that economic actors should not be thought as independent agents; rather, they should be grouped into “collectivities,” groups of agents where information diffuses more easily. Several empirical studies have provided evidence that structurally equivalent collectivities can predict behavior (e.g. Burt, 1987; Harkola & Greve, 1995). There are several ways of grouping organizations into collectivities. One obvious grouping is by geographical proximity. The studies pioneered by Hagerstand (1952) show that “spatial interactions” generally have a positive impact on information diffusion. The basic idea here is that information is more easily transferred and runs a lower risk of integrity loss when the agents are geographically closer. This is particularly important for complex information such as that derived from the use of a new technology. Following Hagerstand’s initial study, the importance of the spatial/geographical dimension in innovation diffusion has been widely studied, including additional research by Krugman (1993) and Audretsch and Feldman (1996). We define contiguous user bandwagon by location as contiguous bandwagons generated by new users located in the same geography of a prospective user. We posit,

Hypothesis 1a: The time between technology adoption and technology use (time to technology use) will be inversely related to contiguous user bandwagon by location.

Another common way of grouping organizations is by industries or groups of organizations that produce close substitutes (Porter, 1980). Within an industry, firms may attempt to match the behavior of rivals in an effort to ease the intensity of competition and keep “competitive parity” (Lieberman & Asaba, 2006). In addition, several empirical studies have provided evidence that industry is an appropriate level of analysis to study technology adoption dynamics (e.g. Powell, Koput, & Smith-Doerr, 1996). We define contiguous user bandwagon by industry as contiguous bandwagons generated by new users of the technology operating in the same industry of a prospective user. We posit,

Hypothesis 1b: The time between technology adoption and technology use (time to technology use) will be inversely related to contiguous user bandwagon by industry.

Basic firm descriptive characteristics are another commonly used way to group organizations based on “structural equivalence” (Abrahamson & Rosenkopf, 1993). Among these characteristics, several authors have focused on the role of firm “status” (Burt, 1997; DiMaggio & Powell, 1983; Granovetter, 1985; Podolny 1994). These studies suggest that companies of the same status are more likely to imitate each other to the point where a status position can easily become “an element of identity” (Phillips & Zuckerman, 2001). In this light, for instance, an international bank’s behavior is more likely to be affected by actions taken by other similar international banks than by local banks’ actions. A company’s legal status—public firm, limited liability, partnership, or incorporated—is an important characteristic often used to categorize firms. It follows from our arguments that, for instance, prospective users from public firms will be more likely to be influenced by bandwagons created by actual users in other public firms. We define contiguous user bandwagon by legal status as contiguous user bandwagons generated by actual technology users that have the same legal status as a prospective user. We posit,

Hypothesis 1c: The time between technology adoption and technology use (time to technology use) will be inversely related to the value of contiguous user bandwagon by legal status.

Method

Sample and Technology Context

We test our hypotheses in the context of e-procurement technology at the early stage of its diffusion. E-procurement describes the use of electronic marketplaces in every stage of the purchasing process, from identification of requirements through payment and contract management. The e-procurement technology considered in this study was offered in standard packages and did not require major customization. The technology vendor did not charge an upfront fee but only a fixed fee per completed transaction on actual use of the technology, which did not vary across firms. We can then reasonably assume that there were no asymmetric disincentives for organizations to experiment with the technology. In addition, interviews conducted with the technology vendor and some adopters suggest that technology implementation time was fairly constant and independent from firm characteristics such as size or industry. We also asked about channels by which users collected information about other users of the technology. In addition to traditional channels (e.g., media and interpersonal communication, where applicable), it also emerged that the e-procurement technology provided users and potential users with access to an electronic database (marketplace) that published real-time information on companies adopting and using the technology. Prospective users had full access to these data and could set up their own search criteria, for example, location, industry, or legal status.

Data for our analysis were obtained from one of the largest European e-procurement providers, which agreed to share with us its database on adoption and use. We sampled 3,158 firms that adopted the e-procurement technology from October 1999 to November 2000 (59 weeks), and we observed them until May 2002. Therefore, the overall observation period spans more than 136 weeks, from October 1999 to May 2002. For each firm adopting the e-procurement technology, the dataset provides some firm-specific data, for example, firm’s location and standard industrial classification (SIC), and tracks the timing of the firm’s activity with the technology, such as time of purchase and time of use. Our sample exhibits right censoring: In Week 136, when we stopped observing our firms, 86.44% of the firms that had subscribed to this technology had not yet used it.

This data set has at least three key strengths. First, data are structured according to a uniform reporting criterion, and therefore they are easily comparable. Second, there is no response bias, because the data capture hard variables that were automatically recorded on the technology vendor database. Third, having sampled firms that have adopted the same e-procurement technology, the sample can be considered self-controlled for differences across technologies.

Measures—Dependent Variable

Time Between Technology Adoption and Technology Use (Time to Technology Use)

Time to technology use is a positive variable calculated as the time elapsed between adoption and first use (see Table 1) or to the end of the observation period if the firm did not use the technology. We say that there is technology use when any of the functionality embedded in the adopted e-procurement software package has been used for the first time. We say that there is technology adoption when a firm purchases the e-procurement software.

Table 1

Sample: Descriptive Statistics

Variable	Mean	Standard Deviation	Minimum	Maximum
Time of first use	124.6	29.70373	22	136
Time of adoption	35.2	15.33683	2	59
Time to technology use	90.4	34.18454	1	135
Censor	0.13	0.342	0	1
Industry
Agriculture and fishing	0.01	0.098606	0	1
Construction	0.002	0.050275	0	1
Mining	0.03	0.178488	0	1
Manufacturing	0.41	0.491334	0	1
Transportation	0.04	0.188133	0	1
Wholesale	0.14	0.342342	0	1
Retail	0.05	0.226346	0	1
Financial services	0.03	0.163703	0	1
Other services	0.27	0.446513	0	1
Public administration	0.002	0.435753	0	1
Legal status
Personal company	0.15	0.357531	0	1
Partnership	0.17	0.374335	0	1
Limited liability	0.41	0.492714	0	1
Incorporated	0.21	0.406654	0	1
Italy	0.98	0.144126	0	1

Note: For each variable, the number of observations is 3,158.

Measures—Independent and Control Variables

Contiguous User Bandwagon

Contiguous user bandwagon is operationalized by computing the number of firms that use the e-procurement technology for the first time the week that a prospective user purchases the technology. Building on this definition, we define contiguous user bandwagon by industry as the number of firms in the same industry of a prospective user that use the e-procurement technology for the first time the week that the prospective user purchases the technology (industry is defined by the first digit of the user’s SIC). We define contiguous user bandwagon by location as the number of firms in the same geographical location of the prospective user that use the e-procurement technology for the first time the week that the prospective user purchases the technology (geographical location is defined at the province level). We define contiguous user bandwagon by firm’s legal status as the number of firms of the same legal status as the prospective user that use the e-procurement technology for the first time the week that the prospective user purchases the technology (see below for a definition of firm’s legal status).

Contiguous Mass Media Communication

As reviewed earlier, mass media communication is a main diffusion mechanism and is considered a rapid and efficient means of informing companies about the existence of new technologies (Fourt & Woodlock, 1960; Mahajan, Muller, & Bass, 1990). We define the variable contiguous mass media communication and compute it as follows. For each period considered in our analysis, we carried out a search in the FACTIVA database counting the number of occurrences of the keywords “e-procurement” and “e-marketplace.” These keywords well describe the technology under consideration. The FACTIVA database searches more than 9,000 sources, including the Wall Street Journal and the Financial Times, and has often been used by other researchers for searches on business-oriented media. Following our reasoning above, contiguous mass media communication is defined as the number of occurrences in the business press of the above keywords for each week included in the analysis.

Contiguous Adopter Bandwagon

Adopter bandwagon is another diffusion mechanism considered in the technology diffusion literature. We compute contiguous adopter bandwagon in any given week as the number of new firms that purchase the e-procurement technology in that week.

Cumulative Adopter (User) Bandwagon

We compute cumulative adopter (user) bandwagon as the total number of technology adopters (users) from the introduction of the technology. It is important to note that contiguous user bandwagon is highly correlated with both cumulative adopter bandwagon and cumulative user bandwagon. However, as we have elaborated extensively above, contiguous and cumulative bandwagons are different constructs. To include both contiguous and cumulative variables in our models and avoid biased estimations that arise from multicollinearity, we used a generated new regressor (Wooldridge, 2002: 115-116), namely, the predicted residuals of a regression of cumulative adopter bandwagon against contiguous user bandwagon. This involves formalizing the relationship to model the overlap between the two measures and using the residuals as predictors. In this way, we can separate out the cumulative adopter effect from the contiguous user effect. This is a standard technique when correcting for multicollinearity (Kennedy, 1992: 210-211).

Industry

We created dummy variables to capture differences across industries. The three main industries are manufacturing (40.7% of the sample), wholesale (13.6%), and services (27.5%).

Censor

We captured right censoring by using a dummy variable that takes the value of 1 if a firm has used the new technology and 0 otherwise.

Legal Status

We created dummy variables representing four types of firms’ legal status: limited liability company, corporation, partnership, and personal company.

Italy

A significant percentage of the sampled firms are based in Italy, and thus a dummy variable captures this country effect.

Time

We created dummy variables for each time period included in the analyses in order to control for time-related effects, including, for instance, cumulative adopter bandwagons, cumulative user bandwagons, marketing campaigns, and stage of adoption.

Table 1 reports descriptive statistics for our variables, and Table 2 provides a correlation matrix. Our sample contains a significant component of right-censored observations (censor = 0.13). The time of adoption in our sample varies between 2 and 59 weeks, with a mean of 35.2 weeks. Time of first use ranges between 22 and 136 weeks, with a mean of 124.6 weeks. Time to technology use, calculated as time of first use minus time of adoption, has an average value of 90.4 weeks and ranges between 1 and 135 weeks.

Table 2

Correlation Coefficients

Variable (V)	V1	V2	V3	V4	V5	V6	V7	V8	V9
V1: Time of adoption	1
V2: Cumulative adopter bandwagon	.9833	1
V3: Cumulative user bandwagon	.97	.9436	1
V4: Contiguous user bandwagon	.9015	.9164	.8838	1
V5: Contiguous adopter bandwagon	.137	.1374	.115	.2389	1
V6: Contiguous media bandwagon	.7676	.7894	.6633	.7022	−.0184	1
V7: Contiguous user bandwagon (standard industrial classification)	.5839	.5927	.5589	.6478	.1364	.4998	1
V8: Contiguous user bandwagon (location)	.325	.3595	.2967	.4172	.1303	.2632	.3042	1
V9: Contiguous user bandwagon (legal status)	.6858	.6817	.6639	.7393	.1682	.5602	.5036	.3427	1

Note: Correlations greater than |.1| are significant at p < .001.

Models and Results

To test our hypotheses on the role of contiguous user bandwagon in the time to technology use, we fitted survival models (Cleves, Gould, & Gutierrez, 2002; Hoesmer & Lemeshow, 1999) on our data. To draw preliminary insights on time to technology use, we first ran a nonparametric estimation. We estimate the hazard function without covariates, that is, the probability of a firm using the adopted technology within a short time interval, conditional on not having used the technology up to the starting time of the interval. The survivor function is, therefore,

λ (t) = \lim_{h ↓ 0} \frac{P (t \leq T < t + h | T \geq t)}{h} .

Figure 1 shows that the hazard is not constant and displays negative duration dependence, dλ(t)/dt < 0. Given the shape of the nonparametric hazard function, we can estimate our model using the accelerated failure time form¹ of logistic distribution. Using this distribution, the hazard function is given by the following:

λ (t) = λ ρ {(λ t)}^{ρ - 1} [1 + {(λ t)}^{ρ}],

with λ(t) being the hazard function and ρ the shape parameter. We introduce covariates by defining λ(t) as a function of a set of regressors:

λ_{i} = e^{- β x_{i}} .

This model allows us to estimate the effect of each explanatory variable on duration. If a coefficient displays a negative sign, it implies that the variable decreases time to technology use (increases the probability of earlier use). No left censoring is present in our database given that, for each firm, the exact time of adoption is known. Models 1 through 7 in Table 3 show the results of our estimations.

Table 3

Model Estimations

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
Contiguous user bandwagon		−0.0818**	−0.163***				−0.0911*
		(0.0260)	(0.0351)				(0.0456)
Contiguous user bandwagon (legal status)						−0.461***	−0.159*
						(0.0985)	(0.0818)
Contiguous user bandwagon (location)					−0.700***		−0.0964
					(0.152)		(0.0899)
Contiguous user bandwagon (industry)				−0.435***			−0.159
				(0.0942)			(0.0818)
Cumulative adopter bandwagon (residuals)			−0.00124***	−0.000933***	−0.000976***	−0.000851**	−0.00133***
			(0.000372)	(0.000255)	(0.000232)	(0.000259)	(0.000374)
Contiguous adopter bandwagon	0.00925***	0.0126***	0.0117***	0.0126***	0.0126***	0.0127***	0.0113***
	(0.00247)	(0.00265)	(0.00266)	(0.00254)	(0.00254)	(0.00253)	(0.00267)
Contiguous mass media communication bandwagon	−0.0118***	−0.00657**	−0.00312	−0.00273	−0.00307	−0.00262	−0.00243
	(0.00168)	(0.00231)	(0.00252)	(0.00254)	(0.00252)	(0.00253)	(0.00257)
Industry
Agriculture and fishing	−1.300	−3.797	−3.949	−3.922	−3.967	−3.945	−4.100
	(1.442)	(2.203)	(2.192)	(2.192)	(2.193)	(2.191)	(2.188)
Mining	32.15	27.28	28.23	26.51	28.18	25.60	29.79
	(1879)	(1209)	(1548)	(1060)	(1547)	(908.8)	(2182)
Construction	0.214	−2.331	−2.315	−2.317	−2.363	−2.384	−2.410
	(1.014)	(1.955)	(1.945)	(1.945)	(1.947)	(1.946)	(1.944)
Manufacturing	−1.115	−3.696*	−3.776*	−3.560*	−3.800*	−3.843*	−3.542*
	(0.588)	(1.775)	(1.766)	(1.788)	(1.768)	(1.768)	(1.794)
Transportation	−0.652	−3.085	−3.180	−3.158	−3.162	−3.214	−3.315
	(0.843)	(1.873)	(1.863)	(1.863)	(1.864)	(1.864)	(1.862)
Wholesale	−0.805	−3.317	−3.379	−3.316	−3.395	−3.437	−3.420
	(0.656)	(1.799)	(1.790)	(1.792)	(1.792)	(1.792)	(1.792)
Retail	−2.921	−2.892	−2.894	−2.889	−2.917	−2.948	−3.021
	−1.866	(1.857)	(1.847)	(1.847)	(1.848)	(1.848)	(1.846)
Financial services	−1.054	−3.537	−3.587	−3.560	−3.572	−3.596	−3.580
	(0.907)	(1.902)	(1.892)	(1.892)	(1.893)	(1.893)	(1.890)
Other services	−0.360	−2.847	−2.866	−2.730	−2.865	−2.925	−2.740
	(0.611)	(1.782)	(1.773)	(1.781)	(1.774)	(1.774)	(1.783)
Public administration	−4.859*	−7.195*	−7.371**	−7.246**	−7.353**	−7.389**	−7.459**
	(2.281)	(2.821)	(2.790)	(2.803)	(2.793)	(2.800)	(2.804)
Legal status
Personal company	−0.430	−0.325	−0.223	−0.256	−0.257	−0.214	−0.193
	(0.941)	(0.942)	(0.939)	(0.938)	(0.939)	(0.939)	(0.938)
Partnership	−1.593	−1.500	−1.460	−1.479	−1.469	−1.366	−1.285
	(0.909)	(0.910)	(0.907)	(0.906)	(0.907)	(0.909)	(0.909)
Limited liability	−1.441	−1.450	−1.385	−1.430	−1.399	−1.097	−0.828
	(0.877)	(0.879)	(0.877)	(0.876)	(0.877)	(0.904)	(0.918)
Incorporated	−3.220***	−3.258***	−3.214***	−3.256***	−3.212***	−3.030***	−2.823**
	(0.886)	(0.889)	(0.887)	(0.885)	(0.887)	(0.896)	(0.901)
Italy	−1.575	−1.435	−1.435	−1.417	−1.363	−1.412	−1.340
	(1.383)	(1.384)	(1.376)	(1.377)	(1.379)	(1.377)	(1.375)
Constant	14.93***	16.39***	16.71***	16.29***	16.46***	16.19***	16.09***
	(1.547)	(2.250)	(2.248)	(2.254)	(2.235)	(2.243)	(2.254)
/ln_gamma	0.883***	0.878***	0.871***	0.872***	0.872***	0.871***	0.869***
Log likelihood	−2033.3	−2028.3	−2022.8	−2022.7	−2022.6	−2022.1	−2019.5

Note: Standard errors are in parentheses.

p < .05. **p < .01. ***p < .001.

Model 1 in Table 3 is a baseline model containing our control variables, contiguous mass media communication and contiguous adopter bandwagon. In Model 2, our key construct, contiguous user bandwagon, is entered. In Model 3, we add cumulative adopter bandwagon. Models 4 through 6 include additional explanatory variables in steps: contiguous user bandwagon by industry, contiguous user bandwagon by location, and contiguous user bandwagon by legal status. Model 7 contains all variables of interest.

All models are highly significant, as shown by their log likelihood (p < .001). Using Model 1 as a baseline, all additional variables in Models 2 through 6 are significant when taken together, according to the likelihood ratio tests reported in the appendix.² Contiguous adopter bandwagon is significant in all models (p < .001). In contrast, contiguous mass media communication is significant only in the first two models. The coefficient of contiguous adopter bandwagon is positive, suggesting that a higher level of contiguous adopter bandwagon increases the time to technology use. The coefficient of cumulative adopter bandwagon (residuals) is significant (p < .001) and negative in all models. Model 2 shows that contiguous user bandwagon is significant (p < .01), and its coefficient, as expected, is negative. It follows that contiguous user bandwagon has a negative impact on the time to technology use—that is, it shortens the gap between technology adoption and use.

In Model 3, and hereafter, we add the residuals of cumulative adopter bandwagon against contiguous user bandwagon. Contiguous user bandwagon is still highly significant (p < .001) and negative. Hypothesis 1 is therefore supported.

We test the effect of contiguous user bandwagons by industry, location, and legal status in Models 4 to 6, respectively.³ As expected, the coefficients of all contiguous bandwagon variables are negative and significant (p < .001). When tested individually, Hypotheses 1a, 1b, and 1c are therefore supported. Model 7 is a full model where all variables of interest are entered.⁴ In Model 7, the coefficients of all contiguous bandwagon variables have the expected negative sign. Contiguous user bandwagon and contiguous user bandwagon by legal status are significant (p < .05). This suggests that the effect of contiguous user bandwagon remains significant even after the other contiguous bandwagon measures are entered in the model.

We performed several other analyses to check for the robustness of our results (not included in this article). We replicated the analyses above by testing contiguous user bandwagon against cumulative user bandwagon (using the residual approach noted above for cumulative adopters), and our key results were confirmed. We then estimated our models by entering time dummies for all periods. Time dummies provide an alternative way to control for cumulative adopter bandwagon, cumulative usage bandwagon, and other time-related trends, for example, the effect of marketing campaigns. Model estimations show that contiguous user bandwagon is a significant predictor (p < .001) of time to technology use even after including time dummies.

Finally, we ran models with contiguous user bandwagons that incorporate new users from periods further back in time, that is, not only new users at a firm’s time of technology adoption. In particular, we ran Model 3 with several redefined contiguous user bandwagon measures that incorporated users from two periods, three periods, and four periods back, respectively. Model estimations show that the coefficients of the lagged contiguous user bandwagons rapidly decrease in their magnitude as the information refers to periods further back (the variable still retains significance and its negative sign). This test provides some support to an important claim in our article that relates to our contiguous measures, that is, that prospective users tend to discount the value of information coming from periods further back.

Discussion and Final Remarks

Technology use is an important topic to be investigated; after all, a new technology can have an impact on firms and industries only if it is used, and as stated above, technology use does not necessarily follow adoption. This article advances the existing literature on technology diffusion by proposing a new construct, contiguous user bandwagon, and showing theoretically and empirically how this construct can help explain the time to technology use.⁵ To address this issue, our proposed new construct directly addresses two assumptions that the existing literature has often made, if only implicitly: (a) that the antecedents of technology adoption and technology use are the same and (b) that organizational actors value all information much in the same way, irrespective of their sources and “newness” (time period the information comes from).

We have shown that by incorporating contiguous user bandwagon in technology diffusion theory—that is, when used to complement cumulative constructs that have been the focus of this theory to date—we can develop a more comprehensive and useful theory of how technology spreads and influences firms and industries. One of the main reasons underpinning the technology adoption–technology use divide is that different organizational actors are likely to be responsible for making the adoption decision and for using and implementing the new technology—senior management and the firm’s technical layer, respectively. Not only are these actors different, but they respond differently to information stimuli. We argue that adopter bandwagons (and the buzz and hype that typically surrounds them) tend to exert a larger influence on senior management than they do on the technical layer of the organization (Abrahamson & Rosenkopf, 1993). Drawing from different theoretical perspectives, we argue that the technical layer tends to have a more conservative approach when it comes to technology, given the fact that, when a new technology is adopted, those in the technical layer have to go through a painful process of change in routines, processes, and cognitive maps. Ultimately, only technologies that are used can enjoy long-lasting diffusion, and our theory explicitly urges researchers (and practitioners) to jointly consider adoption and usage to gain a better understanding of long-term technology diffusion patterns. We have argued that decision makers may tend to heavily discount information from earlier periods, a feature that existing literature on technology diffusion has largely overlooked. This consideration led us to move beyond the traditional conceptualization of bandwagons. Our contiguous user bandwagon construct departs from conventional wisdom in diffusion theory and bandwagon studies that define and operationalize bandwagons as the cumulative level of adopters—that is, information coming from all time periods since the launching of a technology.

The empirical results presented here provide some support for the hypothesis that contiguous user bandwagon is an important antecedent of time to technology use. Furthermore, we find that contiguous user bandwagon by location is also a significant antecedent of time to technology use. Our models also include other contiguous bandwagon control variables. Particularly interesting are our results for contiguous adopter bandwagons. This variable shows a significant effect on time to technology use yet its net effect is positive, increasing the time to use. This result may apparently look surprising. Yet, our arguments above made it clear that adopters have different motivations and costs when compared to users; thus, it should be no surprise that users may actually react with some skepticism to bandwagons triggered by adopters. For instance, users may perceive management’s decision to adopt as a “fad” or simply a “me too” behavior (Strang & Macy, 2001) and resist the use of the new technology.

Contiguous mass media communication does not reach significance in most of our models. It is interesting to consider that, in the time span considered in our analysis, mass media communication created very high expectations regarding the advantages and promises of e-procurement and other Internet-based technologies. Yet, this bandwagon may have affected the behavior of adopters but not that of users. As we argued earlier, users are less likely than adopters to be influenced by business media communication.⁶ Moreover, our result here suggests that users do not respond to the same stimuli as do adopters. Overall, these results provide further support to one of the key ideas of the article: the need to differentiate between the antecedents of adoption and use.

Finally, our analyses confirm that our theory does add predictive power to the extant bandwagon literature by showing that contiguous user bandwagon does have a significant (negative) impact on time to use even after controlling for the cumulative level of adopters (or users). Indeed, our theory not only can help predict time to use but also can improve our predictive power in terms of the overall technology adoption and diffusion patterns of a given technology. Only technologies that are used can have a long-lasting diffusion; our theory prompts research and practitioners to consider both aspects (adoption and use) of the overall diffusion process.

Avenues for Future Research

There are some limitations that apply to this article that open interesting avenues for further research. We have conducted our study in a specific context (e-procurement technology), which can be considered a process technology. Although we believe that our propositions are general enough to be applicable to other contexts, this should be done with the usual caveats. Further studies could focus on the antecedents of technology use in other phases of the technology diffusion cycle (we have focused on the early diffusion phase) or replicate our study with other technologies to provide interesting comparisons. Also, our measure of technology use itself may be improved upon. In this article, we consider use as a discrete event: A firm either uses or does not use the technology during the time of analysis. We measured this by looking at the time of first use of the e-procurement system. However, it could be argued that there are different degrees and forms of use, and a future study could try to provide that extra granularity in the analysis—for example, by capturing the persistence or effectiveness of use by different users. Building upon the existing technology diffusion literature, we assume that prospective users, like adopters, are influenced by mass media communication; this aspect should be explored more, both theoretically and empirically. Finally, caution should also be taken in extending our results outside the original sample; despite the fact that our analysis is largely exempt from nonresponse bias, there could still exist nonadoption bias because we did not observe the whole population of adopter and nonadopters.

Managerial and Policy Implications

The results of our study suggest that technology producers and technology vendors should pay attention to contiguous bandwagons, particularly in what relates to “managing” the use characteristics of their new technologies so as to enable contiguous user bandwagons. For instance, technology vendors could create strategic action plans to increase user bandwagons. To some degree, this is already happening in sectors such as software. Nearly all software product companies have set up user and developer groups that are designed to diffuse technology information to current and future users.

Our results point to the contiguous user bandwagon as an important mechanism for the long-term success of the new technology. A boom in early sales of a new technology might be followed by a sudden drop if the technology is not accepted by its potential users. Technology producers and vendors should be aware that the timing of introducing a new technology is a key variable and should critically consider the implications of rushing a new technology to the market if that can have negative implications for user acceptability. For senior managers making technology adoption decisions, our research flags a warning to the risk of shelfware that can result from choices that do not take into account the antecedents of technology use. Decision-making managers should strive to have a better understanding of users within their own organizations and the implementation risks and obstacles associated with new technologies and should facilitate the communication between prospective uses in their organization with actual users of a technology in other organizations. In this light, Internet-based social networks—for example, blogs, Web 2.0, or more recently Twitter—could provide an important platform to initiate and trigger these user bandwagons.⁷

Our results also have implications for policy makers. When it comes to new technologies, policy makers often want to find ways of accelerating diffusion, but diffusion of a technology is not successful unless the technology is both purchased and used. Policy makers should increase their awareness of the differences between adopters and users of new technologies and correspondingly retool their instruments and policies to support effective technology diffusion. For instance, policy makers could place greater emphasis on policies that foster and promote technology use and not just on technology adoption. As the technical layer of an organizations is crucial in technology use decisions, policies designed to ease the transition of these organizational actors from the old to the new technology could include training, user workshops, and other policies destined to promote user awareness and exchange. Technology use, and not technology adoption, should be the final aim of policy makers.

Footnotes

Appendix

Appendix 1

Joint Likelihood Ratio Tests

Likelihood-ratio test	LR chi2(1) = 10.11
(Assumption: model1 nested in model2)	Prob > chi2 = 0.0015
Likelihood-ratio test	LR chi2(2) = 21.21
(Assumption: model1 nested in model3)	Prob > chi2 = 0.0000
Likelihood-ratio test	LR chi2(2) = 21.28
(Assumption: model1 nested in model4)	Prob > chi2 = 0.0000
Likelihood-ratio test	LR chi2(2) = 21.50
(Assumption: model1 nested in model5)	Prob > chi2 = 0.0000
Likelihood-ratio test	LR chi2(2) = 22.49
(Assumption: model1 nested in model6)	Prob > chi2 = 0.0000
Likelihood-ratio test	LR chi2(5) = 27.69
(Assumption: model1 nested in model7)	Prob > chi2 = 0.0000
Likelihood-ratio test	LR chi2(3) = 6.49
(Assumption: model3 nested in model7)	Prob > chi2 = 0.0901

Acknowledgements

This article was accepted under the editorship of Talya N. Bauer.

Notes

References

Abernathy

W. J.

Utterback

J. M.

1978. Patterns of industrial innovation. Technology Review, 80: 97-107.

Abrahamson

Rosenkopf

1993. Institutional and competitive bandwagons: Using mathematical modeling as a tool to explore innovation diffusion. Academy of Management Review, 18: 487-517.

Agarwal

Prasad

1998. A conceptual and operational definition of personal innovativeness in the domain of information technology. Information Systems Research, 9: 204-215.

Agarwal

Prasad

1998. The antecedents and consequents of user perceptions in information technology adoption. Decision Support Systems, 22(1): 15-29.

Ajzen

1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50: 179-211.

Anderson

Tushman

M. L.

1990. Technological discontinuities and dominant design: A cyclical model of technological change. Administrative Science Quarterly, 35: 604-633.

Attewell

, 1992. Technology diffusion and organizational learning: The case of business computing. Organization Science, 3: 1-19.

At your service. 2003. Economist, 367: 8323.

Audretsch

D. B.

Feldman

M. P.

1996. R&D spillovers and the geography of innovation and production. American Economic Review, 86: 630-640.

10.

Bacharach

Bamberger

Sonenstouhl

1996. The organizational transformation process: The micropolitics of dissonance reduction and the alignment of logics of action. Administrative Science Quarterly, 41: 477-506.

11.

Bass

F. M.

1969. A new product growth model for consumer durables. Management Science, 15: 215-227.

12.

Black

Carlile

Repenning

2004. A dynamic theory of expertise and occupational boundaries in new technology implementation: Building on Barley’s study of CT scanning. Administrative Science Quarterly, 49: 572-607.

13.

Blau

P. M.

1970. A formal theory of differentiation in organizations. American Sociological Review, 35: 201-218.

14.

Bresnahan

T. E.

Brynjolfsson

Hitt

2002. Information technology, workplace organization and the demand for skilled labor: Firm-level evidence. Quarterly Journal of Economics, 117: 339-376.

15.

Burt

R. S.

1982. Towards a structural theory of action. Cambridge, MA: Harvard University Press.

16.

Burt

1987. Social contagion and innovation: Cohesion versus structural equivalence. American Journal of Sociology, 92: 1287-1335.

17.

Burt

R. S.

1997. The contingent value of social capital. Administrative Science Quarterly, 42: 339-365.

18.

Cabral

Leiblein

2001. Adoption of a process innovation with learning-by-doing: Evidence from the semiconductor industry. Journal of Industrial Economics, 49: 269-280.

19.

Carpenter

G. S.

Nakamoto

1989. Consumer preference formation and pioneering advantage. Journal of Marketing Research, 26: 285-298.

20.

Christensen

C. M.

1992. Exploring the limits of the technology S-curve. Part I: Component technologies. Production and Operations Management Journal, 1: 334-357.

21.

Christensen

C. M.

Antony

S. D.

Roth

E. A.

2004. Seeing what’s next. Boston: Harvard Business School Press.

22.

Clark

1985. The interaction of design hierarchies and market concepts in technological evolution. Research Policy, 14: 235-251.

23.

Cleves

Gould

W. M.

Gutierrez

R. G.

2002. An introduction to survival analysis using Stata. College Station, TX: Stata Press.

24.

Cohen

W. M.

Levinthal

D. A.

1990. Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35: 128-152.

25.

Cooper

R. B.

Zmud

R. W.

1990. Information technology implementation research: A technological diffusion approach. Management Science, 36: 123-139.

26.

Dasgupta

Stiglitz

1980. Uncertainty, industrial structure and the speed of Rand. Bell Journal of Economics, 11: 1-28.

27.

Davis

F. D.

1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13: 319-340.

28.

DiMaggio

P. J.

Powell

W. W.

1983. The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48: 147-160.

29.

Dubin

1978. Theory building. New York: Oxford University Press.

30.

Duplaga

E. A.

Astani

2003. Implementing ERP in manufacturing. Information Systems Management, 20(3): 68-75.

31.

Edmondson

A. C.

Bohmer

R. M.

Pisano

G. P.

2001. Disrupted routines: Team learning and new technology adaptation. Administrative Science Quarterly, 46: 685-716.

32.

Farrell

Saloner

1985. Standardization, compatibility, and innovation. Rand Journal of Economics, 16: 70-83.

33.

Fiol

C. M.

O’Connor

E. J.

2003. Waking up! Mindfulness in the face of bandwagons. Academy of Management Review, 28: 54-70.

34.

Foster

1986. Innovation: The attacker’s advantage. New York: Summit Books.

35.

Fourt

L. A.

Woodlock

J. W.

1960. Early prediction of market success for new grocery products. Journal of Marketing, 25: 31-38.

36.

Granovetter

1985. Economic action and soda! structure: The problem of embeddedness. American Journal of Sociology, 91: 481-510.

37.

Geroski

P. A.

2000. Models of technology diffusion. Research Policy, 29: 603-625.

38.

Gurbaxani

1990. Diffusion of computing networks: The case of Bitnet. Communications of the ACM, 33: 65-75.

39.

Hagerstand

1952. The propagation of innovation waves. Lund Studies in Geography No. 4. London: Royal University of Lund.

40.

Hannan

Freeman

1989. Organizational ecology. Cambridge, MA: Harvard University Press.

41.

Harkola

Greve

1995. Diffusion of technology: Cohesion or structural equivalence. Academy of Management Best Papers Proceedings.

42.

Hedstrom

Swedberg

1998. Social mechanisms. Cambridge, UK: Cambridge University Press.

43.

Hoesmer

Lemeshow

1999. Applied survival analysis. New York: John Wiley.

44.

Kahl

2007. Utilizing use: The potential impact of customer use patterns on technology and industry evolution. In AOM Best Paper Proceedings. Briarcliff Manor, NY: Academy of Management.

45.

Katz

Shapiro

1986. Technology adoption in the presence of network externalities. Journal of Political Economy, 94: 822-841.

46.

Katz

Shapiro

1994. Systems Competition and Network Effects. Journal of Economic Perspectives, 8: 93-115.

47.

Kennedy

1992. A guide to econometrics. Cambridge, MA: MIT Press.

48.

Krugman

P. R.

1993. Geography and trade. Chicago: University of Chicago Press.

49.

Kumar

1992. Technological innovation diffusion: The proliferation of substitution models and easing the user’s dilemma. IEEE Transactions on Engineering Management, 39: 158-166.

50.

Langlois

R. N.

1992. External economies and economic progress: The case of the microcomputer industry. Business History Review, 66: 1-50.

51.

Lazarsfeld

P. F.

Merton

R. K.

1964. Friendship as social process: A substantive and methodological analysis. In Berger

Abel

Page

C. H.

(Eds.), Freedom and control in modern society (pp. 18-66). New York: Van Nostrand.

52.

Leonard-Barton

1992. Core capabilities and core rigidities: A paradox in managing new product development. Strategic Management Journal, 13: 111-125.

53.

Leonard-Barton

Deschamps

1988. Managerial influence in the implementation of new technology. Management Science, 34(10): 1252-1265.

54.

Lieberman

M. B.

Asaba

2006. Why do firms imitate each other? Academy of Management Review, 31: 366-385.

55.

Loch

C. H.

Huberman

B. A.

1999. A punctuated equilibrium model of technology diffusion. Management Science, 45: 160-177.

56.

Mabert

V. A.

Soni

Venkataramanan

M. A.

2001. Enterprise resource planning: Common myths versus evolving reality. Business Horizons, 44(3): 69-76.

57.

Mahajan

Muller

Bass

F. M.

1990. New product diffusion in marketing: A review and directions for research. Journal of Marketing, 54: 1-26.

58.

Mansfield

1961. Technical change and the rate of imitation. Econometrica, 29: 741-766.

59.

Meyer

J. W.

Rowan

1977. Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83: 340-363.

60.

Miller

K. D.

Folta

T. B.

2002. Option value and entry timing. Strategic Management Journal, 23: 655-665.

61.

Nelson

R. R.

Winter

S. G.

1982. An evolutionary theory of economic change. Cambridge, MA: Belknap of Harvard University Press.

62.

Nohria

Gulati

1996. Is slack good or bad for innovation? Academy of Management Journal, 39: 1245-1264.

63.

Orlikowski

W. J.

1996. Improvising organizational transformation over time: A situated change perspective. Information Systems Research, 7: 63-92.

64.

Phillips

D. J.

Zuckerman

E. W. E. W.

2001. Middle-status conformity: Theoretical refinement and empirical demonstration in two markets. American Journal of Sociology, 107: 379–429.

65.

Podolny

J. M.

1994. Market uncertainty and the social character of economic exchange. Administrative Science Quarterly, 39: 458-483.

66.

Porter

M. E.

1980. Competitive strategy. New York: Free Press.

67.

Powell

W. W.

Koput

K. W.

Smith-Doerr

1996. Interorganizational collaboration and the locus of innovation: Networks of learning. Administrative Science Quarterly, 41: 116-145.

68.

Rogers

E. M.

2003. Diffusion of innovation. New York: Free Press.

69.

Schmalensee

1982. Product differentiation advantages of pioneering brands. American Economic Review, 72: 349-368.

70.

Shy

2001. The economics of network industries. New York: Cambridge University Press.

71.

Strang

Macy

M. W.

2001. In search of excellence: Fads, success stories, and adaptive emulation. American Journal of Sociology, 107: 147-182.

72.

Suarez

2005. Network effects revisited: The role of strong ties on technology selection. Academy of Management Journal, 48: 710-720.

73.

Tolbert

P. S.

Zucker

L. G.

1983. Institutional sources of change in the formal structure of organizations: The diffusion of civil service reform: 1880-1935. Administrative Science Quarterly, 28: 22-39.

74.

Utterback

J. M.

Suarez

F. F.

1993. Innovation, competition, and industry structure. Research Policy, 22: 11-21.

75.

Van De Ven

A. H.

1986. Central problem in the management of innovation. Management Science, 32: 590-607.

76.

Venkatesh

Speier

Morris

M. G.

2002. User acceptance enablers in individual decision making about technology: Toward an integrated model. Decision Science, 33: 297-316.

77.

Westphal

Gulati

Shortell

1997. Customization or conformity? An institutional and network perspective on the content and consequences of TQM adoption. Administrative Science Quarterly, 42: 366-394.

78.

Wooldridge

J. M.

2002. Econometric analysis of cross section and panel. Cambridge, MA: MIT Press.

79.

Zbaracki

M. J.

1998. The rhetoric and reality of total quality management. Administrative Science Quarterly, 43: 602-638.

Closing the Technology Adoption–Use Divide

Abstract

Keywords

Theory and Hypotheses

Technology Diffusion Literature

Extending the Theory of Technology Diffusion

Contiguous User Bandwagon and the Time to Technology Use

Method

Sample and Technology Context

Measures—Dependent Variable

Time Between Technology Adoption and Technology Use (Time to Technology Use)

Measures—Independent and Control Variables

Contiguous User Bandwagon

Contiguous Mass Media Communication

Contiguous Adopter Bandwagon

Cumulative Adopter (User) Bandwagon

Industry

Censor

Legal Status

Italy

Time

Models and Results

Discussion and Final Remarks

Avenues for Future Research

Managerial and Policy Implications

Footnotes

Appendix

Acknowledgements

Notes

References