Closing the Technology Adoption/Use divide: The role of Contiguous User Bandwagon Gianvito Lanzolla Cass Business School 106 Bunhill Road, Room 4065 London EC1Y 8TZ, UK Tel. 44 (0)20 7040 5243 email: [email protected]Fernando F. Suarez Boston University School of Management 595 Commonwealth Avenue, Room 546-F Boston MA 02215, USA Tel. (617) 358-3572 email: [email protected]27 November 2009
45
Embed
Closing the Technology Adoption/Use divide: The …people.bu.edu/fsuarez/Fernando_Suarez_Website/Publications_files... · 2 Closing the Technology Adoption/Use Divide: The Role of
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
& Lemeshow, 1999) on our data. To draw preliminary insights on time to technology
use, we first ran a non-parametric 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:
24
------------------------------------------------
Please insert Figure 1 about here
------------------------------------------------
Figure 1 shows that the hazard is not constant and displays negative duration
dependence, dλ(t)/dt < 0. Given the shape of the non parametric 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:
With being the hazard function and the shape parameter. We introduce
covariates by defining as a function of a set of regressors:
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 throughout 7 in Table 3 show the results of our estimations.
------------------------------------------------
Please insert Table 3 about here
------------------------------------------------
1 An accelerated failure time form is characterized by its conditional survival function S(t|Z=z) for a duration T, with . is the baseline survivor function, and if it is specified parametrically, we get a parametric model (as in our case). The model used depends on how we define
(we used a log-logistic model here).
25
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<0.001). 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 Appendix2.
Contiguous adopter bandwagon is significant in all models (p <0.001). In contrast,
contiguous mass media communication is significant only in the first two models. The
coefficient of contiguous adopter bandwagon is positive, suggesting that higher level of
contiguous adopter bandwagon increases the time to technology use. The coefficient of
cumulative adopter bandwagon (residuals) is significant (p <0.001) and negative in all
models. Model 2 shows that contiguous user bandwagon is significant (p<.001) and its
coefficient, as expected, is negative3. It follows that contiguous user bandwagon has a
negative impact on the time to technology use – i.e. it shortens the gap between
technology adoption and use.
2 Likelihood-ratio tests whether the parameters used in a given model are significant by comparing the likelihood of this model with the likelihood of a model without parameters using:
under with Q the number of restrictions.
26
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, respectively4. As expected, the coefficients of all contiguous
bandwagon variables are negative and significant (p<0.001). Hypotheses 1a, 1b and 1c
are therefore supported. Model 7 is a full model where all variables of interest are
entered5. In this Model, 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 contiguous user
bandwagon remain 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 paper. 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 – e.g. the effect of marketing campaigns.
Model estimations show that contiguous user bandwagon is a significant predictor
(p<0.001) of time to technology use even after including time dummies.
4 In each of these models, the cumulative adopter bandwagon variable expresses the residuals of cumulative adopter bandwagon against contiguous user bandwagons by industry, contiguous user bandwagons by location, and contiguous user bandwagons by legal status, respectively. 5 In this Model, cumulative adopter bandwagon expresses the residuals of cumulative adopter bandwagon against contiguous user bandwagon.
27
Finally, we ran models with contiguous user bandwagons that incorporate new users
from periods further back in time – i.e. 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 paper that
relates to our “contiguous” measures, i.e. that prospective users tend to discount the
value of information coming from periods further back.
5. Discussion and final remarks
Technology use is an important topic to be investigated; after all, a new technology
can only have an impact on firms and industries if it is used and, as stated above,
technology use does not necessarily follow adoption. This paper advances 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 use6. To address this issue, our proposed new construct
directly addresses two assumptions that existing literature has often made, if only
implicitly: (a) that the antecedents of technology adoption and technology use are the
same; (b) that organizationl actors value all information much in the same way,
irrespective of their sources and “newness” (time period the information comes from) .
6 As noted by an anonymous reviewer, our theoretical contribution can be classified as “invention by extension” (Dubin, 1978).
28
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 responsible for making the adoption decision and for using and
implementing the new technology -- senior management and the firm’s technical layer,
respectively. These actors not only are 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, they 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.
29
The empirical results presented here provide strong 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 come as no big 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 to be influenced by business media communication
than adopters 7. Moreover, our result here further suggests that users do not respond to
the same stimuli than adopters. Overall, these results provide further support to one of
the key ideas of the paper – i.e. the need to differentiate between antecedents of
adoption and use.
7 We thank an anonymous reviewer for pointing out this potential explanation.
30
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 can help not only predict time to use
but also 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.
5.1 Avenues for future research
There are some limitations that apply to this paper 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 paper, we have considered 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 – e.g. by capturing the persistence or effectiveness of
use by different users. Last, building upon existing technology diffusion literature, we
31
assume that prospective users, like adopters, are influenced by mass media
communication. This aspect should be explored more, both theoretically and
empirically.
5.2 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 technology 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, designed to diffuse technology information to current and future users.
Our results point to 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
32
users of a technology in other organizations. In this light, Internet-based social networks
– e.g. blogs, Web 2.0, or more recently Twitter - could provide an important platform to
initiate and trigger these user bandwagons8.
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 re-tool their instruments and
policies used 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 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.
8 We thank an anonymous reviewer for raising this implication.
33
References
Abernathy, W. J., & Utterback, J. M., 1978. Patterns of industrial innovation.
Technology Review, 80, 97–107.
Abrahamson, R., & Rosenkopf, L., 1993. Institutional and competitive bandwagons:
using mathematical modeling as a tool to explore innovation diffusion. Academy of
Management Review, 18: 487-517.
Abrahamson, R., & Rosenkopf, L., 1997. Social Network Effects on the Extent of
Innovation Diffusion: A Computer Simulation. Organization Science, 8: 289-309.
Agarwal, R. & Prasad, J. 1998. The antecedents and consequents of user perceptions in
information technology adoption, Decision Support System, 22(1)
Ajzen I., 1991. The theory of Planned Behavior. Organizational Behavior and Human
Decision Processes, 50: 179-211.
Anderson, P., & Tushman, M.L., 1990. Technological discontinuities and dominant
design: a cyclical model of technological change. Administrative Science Quarterly,
35: 604 – 633.
Attewell, P., 1992. Technology diffusion and organizational learning: the case of
business computing. Organization Science, 3: 1 – 19.
Bacharach, S., & Bamberger, P., Sonenstouhl, W., 1996. The organizational
transformation process: The micropolitics of dissonance reduction and the alignment of
logics of action. Administrative Science Quarterly, 41: 477-506.
Bass, F.M., 1969. A new product growth model for consumer durables. Management
Science, 15: 215-227.
Beaudry A., & Pinsoneault, A., 2005. Understanding user responses to Information
Technology: a copy model for user adaptation. MIS Quarterly, 29: 493 -524.
34
Black, L., Carlile, P., & Repenning, N. Formalizing Theoretical Insights from
Ethnographic Evidence: Revisiting Barley’s Study of CT-Scanning Implementations.
Administrative Science Quarterly, forthcoming..
Blau, P. M., 1970. A Formal Theory of Differentiation in Organizations. American
Sociological Review, 35: 201-218.
Bresnahan, T. E., Brynjolfsson, E., & Hitt, L., 2002. Information Technology,
Workplace Organization and the Demand for Skilled Labor: Firm-level Evidence.
Quarterly Journal of Economics: 117: 339-376.
Burt, R., 1987. Social contagion and innovation: cohesion versus structural equivalence.
American Journal of Sociology, 92: 1287 – 1335.
Cabral, D., & Leiblein, M., 2001. Adoption of a Process Innovation with Learning-by-
Doing: Evidence from the Semiconductor Industry. Journal of Industrial Economics,
49.
Carpenter, G. S., & Nakamoto, K., 1989. Consumer preference formation and
pioneering advantage. Journal of Marketing Research, 26: 285-298.
Christensen, C. M., 1992. Exploring the limits of the Technology S-curve, Part I:
Component Technologies. Product and Operations Management Journal, 1: 334-357.
Christensen, C. M., Roth, E. A. & Antony, S. D. 2004. Seeing what’s next. Boston,
MA: Harvard Business School Press.
Clark, K., 1985. The interaction of design hierarchies and market concepts in
technological evolution. Research Policy, 14: 235-251.
Cleves, M., Gould, W.M., & Gutierrez, R.G., 2002. An Introduction to Survival
Analysis using Stata. College Station, TX: Stata Press.
35
Cohen, W. M., & Levinthal, D.A., 1990. Absorptive capacity: a new perspective on
learning and innovation. Administrative Science Quarterly, 35: 128 – 152.
Cooper, R.B., & Zmud, R.W., 1990. Information Technology Implementation Research:
A Technological Diffusion Approach. Management Science: 36: 123 – 139.
Czepiel, J.A., 1975. Patterns of Interorganizational Communications and the Diffusion
of a Major Technological Innovation in a Competitive Industrial Community. Academy
of Management Journal, 18: 6-24.
Dasgupta, P., & Stiglitz, J., 1980. Uncertainty, Industrial Structure and the Speed of
Rand. Bell Journal of Economics, 11: 1-28.
Davis F.D., 1989. Perceived Usefulness, Perceived Ease of Use, and user Acceptance of
Information Technology. MIS Quarterly, 13: 319 -340.
Davis, F.D. Bagozzi, P., & Warshaw, P.R., 1989. User Acceptance of Computer
Technology: A Comparison of Two Models. Management Science, 38: 475 – 487.
Desanctis, G., & Poole, M. S., 1994. Capturing the complexity in advanced technology