You are on page 1of 7

Journal of Business Research 116 (2020) 535–541

Contents lists available at ScienceDirect

Journal of Business Research


journal homepage: www.elsevier.com/locate/jbusres

An evolutionary perspective on adoption-diffusion theory T


Mario Gruber
King's College London, United Kingdom

A R T I C LE I N FO A B S T R A C T

Keywords: Innovations increasingly comprise combinations of different technologies, institutions, and user behaviors that
Adoption interdepend and co-evolve. Accordingly, extant adoption-diffusion theories – based on variance and process
Diffusion approaches – are increasingly limited in explaining or enabling understanding of the adoption and diffusion of
Radical radical, system-dependent innovations, such as electric mobility. Drawing on the multi-level perspective, an
Multi-level perspective
evolutionary system-approach theory, this article first identifies the limitations of extant adoption-diffusion
theories. It then uses this perspective to propose an understanding of innovation adoption and diffusion that
accounts for factors and developments beyond extant theories’ typical focus. In particular, it highlights the
evolutionary elements, such as co-evolution and whole system behavior, that should be considered for radical
innovations, whose adoption and diffusion depend on co-evolving user preferences, technology, and policy.
Overall, the article presents how the evolutionary perspective can enable researchers to better understand
adoption and diffusion and incorporate adaptations of theory in future research.

1. Introduction behavior and skills, infrastructure, and regulation (Dijk, Kemp, &
Valkering, 2013; Kanger, Geels, Sovacool, & Schot, 2019). These in-
Adoption-diffusion (AD) theories (e.g., Davis, Bagozzi, & Warshaw, novations are embedded in a context of sociotechnical (ST) landscapes,
1989; Rogers, 2003; Venkatesh, Thong, & Xu, 2012) were constructed so their diffusion hinges on the development of complex systems, en-
to explain, predict, and understand the adoption and diffusion of in- tailing drastic technological and social change (Geels et al., 2018;
novations. Most prominently, researchers either follow a variance ap- Kasmire, Korhonen, & Nikolic, 2012). This implies that the benefits of
proach, aiming to build specific, parsimonious theory (Weber, 2012), or radical innovations cannot be assumed static, but that they emerge over
formulate models to explain and predict diffusion (Peres, Muller, & time, limiting the ability of the variance approach to describe those
Mahajan, 2010). Typically, AD theories have been constructed in the benefits and thus explain or predict individual adoption.
context of organizations and with respect to specific technologies or This article posits that for radical innovations research needs not
focus on individuals’ adoption. However, scholars have increasingly only statistically relevant measurements of AD factors or processes but
argued that AD theories need adapting to fit their context (Venkatesh, also an understanding of on-going, multi-faceted change, emergence of
Thong, & Xu, 2016). The term “context” is understood as “the sur- properties, and system-specific influences (Burton-Jones et al., 2015;
roundings associated with phenomena which help to illuminate that Mignon & Bergek, 2016). Recent calls to extend the context of AD re-
phenomena, typically factors associated with units of analysis above search (e.g., Venkatesh et al., 2016) or to integrate more advanced
those expressly under investigation” (Cappelli & Sherer, 1991, p. 56). process and system approaches are still mostly confined to information
However, the usefulness and quality of AD theories depend on how well systems (Ortiz de Guinea & Webster, 2017). Meanwhile, related fields
their concepts reflect their studied phenomena, i.e., adoption and dif- such as markets and consumer research have recently called for the
fusion (Burton-Jones, McLean, & Monod, 2015). advancement of process and system perspectives: they advocate moving
Extant theories are challenged by increasing complexity in their beyond set assumptions and static, linear conceptualizations to study
contexts, as illustrated by radical innovations, such as in sustainability iterative interactions, struggles, conflicts, and system-specific patterns
and transport: e.g., electric mobility or shared autonomous electric of change (Giesler & Thompson, 2016; Vargo et al., 2017).
vehicles (Geels, Schwanen, Sorrell, Jenkins, & Sovacool, 2018; Sovacool This article draws on the multi-level perspective (MLP) (Geels,
& Axsen, 2018). Radical innovations are radical in the sense that their 2005) – a system approach allowing researchers to build theory on
adoption and diffusion depend on the co-evolution and inter- micro, meso, and macro levels of analysis – as a fitting system per-
dependence of diverse elements, including technology and culture, user spective to analyze how AD theory is limited and might be adapted with

E-mail address: Mario.Gruber@kcl.ac.uk.

https://doi.org/10.1016/j.jbusres.2020.02.024
Received 27 November 2018; Received in revised form 10 February 2020; Accepted 14 February 2020
Available online 14 March 2020
0148-2963/ © 2020 Elsevier Inc. All rights reserved.
M. Gruber Journal of Business Research 116 (2020) 535–541

respect to radical innovations. In particular, it uses the MLP to con- choice but as on-going behavior and activities, such as iteratively
ceptualize the interdependence and reciprocal interactions on these learning and adapting needs. As these limitations are critical for radical
three levels that form the co-evolution and diffusion of radical in- innovations due to their dependence on contextual developments of the
novations (Geels & Johnson, 2018). Furthermore, this article shows complex systems in which they are embedded, research can benefit
how AD theory could benefit from more qualitative research in order to from incorporating more process approaches (Hojnik & Ruzzier, 2016).
capture complexities, rather than reducing them. The behavior of complex systems changes and emerges as the out-
Recent literature on adoption and diffusion using ST approaches come of many interacting components, whereby the whole is greater
(e.g., Geels & Johnson, 2018; Kanger et al., 2019; Sovacool, 2017) has than the sum of its parts (Mitchell, 2009). Scholars have suggested
contrasted the two theoretical fields, but no prior study has reported in applying a system approach to AD theory in order to incorporate system
detail how the MLP can benefit AD research. effects and multi-dimensional interactions between parts and wholes
Radical innovations such as in sustainability are critical to solving (Burton-Jones et al., 2015). For example, radical innovations require
climate change (Geels et al., 2018), while the potential value of shared researchers to understand not only adoption behavior or aggregate at-
autonomous vehicles has been estimated at potentially trillions of titudes (i.e., the parts) but also how the system (i.e., the whole) of
dollars per year (Fagnant & Kockelman, 2015). However, radical sus- transport industry, infrastructure, culture, and policy, affect behavior,
tainability and transport innovations ultimately depend on adoption by and vice versa (Geels, 2018).
end users. This article attempts to answer recent calls to adapt AD It is also important to note that diffusion research, despite including
theory, particularly for radical innovations, and so makes important the time dimension, often still focuses on producing factors rather than
theoretical and practical contributions. processes. This is exemplified by an international review of diffusion
This article is structured as follows. Sections 2–4 present the con- models for electric vehicles (Gnann et al., 2018): these models most
stituent elements and theoretical approaches of AD theory, as well as commonly aim to forecast diffusion or identify its most important fac-
key ST approaches. Section 5 then highlights the limitations of extant tors, such as charging infrastructure or policy incentives.
AD theory based on the traits of its approaches, as well as potential
improvements from the evolutionary ST perspective. Finally, section 6 4. The evolutionary perspective on radical innovations
presents suggestions for how future research in the AD field can be
adapted. Researchers of system transitions argue that neoclassical economic
assumptions of stable factors, rational actors, or complete information
2. The limitations of the factor approach in AD theory may produce oversimplified conceptualizations, especially for radical
innovations: for instance, the view that social acceptance will occur if
Variance (or factor) theories in AD research follow a reductionist an innovation is better than alternatives (Geels et al., 2018).
philosophy and, thus, aim to summarize myriad factors as independent By contrast, evolutionary economics understands outcomes as re-
variables to explain or predict individuals’ adoption behavior with high sulting from the cumulative effects of multiple factors, events, and
statistical precision and parsimony (Burton-Jones et al., 2015; Weber, choices (which are often unpredictable and unconscious) that reinforce
2012). However, lacking the time dimension, factor-approach theories one another towards a shared direction or trajectory (Nelson & Winter,
can be understood as a “snapshot of a situation,” thus limiting re- 1982; Williams & Edge, 1996). The resulting path dependence (Arthur,
searchers’ understanding of how and why adoption and diffusion occur 1988) implies that transport infrastructure, investments, laws, and
(Ortiz de Guinea & Webster, 2017, p. 150). Moreover, variance theories consumer preferences lay the tracks for many future developments and
have been criticized for potentially producing overly deterministic thus favor incremental innovation (Whitmarsh, 2012). While AD
constructs and overly linear and mechanistic relationships, both in- scholars have previously emphasized the importance of evolutionary
sufficiently reflective of real-world phenomena to explain, predict, and elements, they have still only incorporated these concepts according to
understand adoption in some cases (Burton-Jones et al., 2015; Ortiz de the variance approach (Fichter & Clausen, 2016), leaving evolutionary
Guinea & Webster, 2017). For example, while the concepts of variance perspectives underexplored in AD theory.
theories can change in value (e.g., from high to low), their meanings
cannot change, which limits researchers’ understanding of how things 4.1. The ST approach to studying the diffusion of radical innovations
emerge or change (Langley, Smallman, Tsoukas, & Van de Ven, 2013).
Lastly, in cases where many enablers and barriers need to be accounted The ST approach is used to study and theorize change in and the
for, the number of factors needed to accurately explain adoption can stability of sociotechnical systems, i.e., combinations of elements such
become “theoretically bewildering” and contradictory (Shove, 2010, p. as technology, behavior, regulation and infrastructure that satisfy a
1277). societal function (e.g., transport; Kanger & Schot, 2016). Change and
Venkatesh et al. (2016) suggested that future research should better stability refer to the alignment of sociotechnical elements in a system
contextualize by introducing a multi-level framework with micro, meso, (i.e., the whole), of which adoption and diffusion can be considered
and macro levels of analysis, as well as introducing a “time attribute” to parts or sub-phenomena (Geels et al., 2018).
distinguish between adoption, initial use, and post-adoptive use. A central theory in the ST field is the MLP (Geels, 2005; Rip & Kemp,
However, the limitations of the variance approach largely remain, and 1998), which is used to study processes and patterns of system transi-
will be elaborated in Section 5. tions and radical, system innovations within them (e.g., from horse-
drawn carriages to automobiles) on three different analytical levels.
3. Towards better process approaches in adoption and diffusion First, at the macro level, the so-called sociotechnical landscape sum-
marizes factors external to an innovation or its industry, and the macro
The process approach, focusing on “how” and “why” innovations (economic, technological, and social) trends establishing “gradients of
are adopted or diffused, allows researchers to identify sequences of force” that make certain developments or actions more likely than
critical events: e.g., adopters being persuaded by innovation-specific others (Geels & Schot, 2007, p. 403): examples include rising pro-en-
information or by a certain price-performance ratio (Montalvo & Kemp, vironmentalism among the public or the improvement of battery
2008). However, when innovations and adopters cannot be assumed technology.
stable but change simultaneously, rather than sequentially, a process Second, at the meso level is the sociotechnical regime where tech-
approach may be too linear or deterministic (Lyytinen & Damsgaard, nologies, user preferences, practices, institutions, and regulations col-
2001). For example, albeit in an organizational context, Makkonen, lectively form a stable functioning whole: e.g., the current regime of the
Johnston, and Javalgi (2016) conceptualized adoption not as a final privately-owned internal combustion engine vehicle (Geels, 2005;

536
M. Gruber Journal of Business Research 116 (2020) 535–541

Kanger et al., 2019; Schot, Kanger, & Verbong, 2016). Where path de- approach of ST theory as a lens to understand how and why AD factors
pendence and lock-in preserve high regime stability, drastic social and and processes emerge, change, and lead to adoption and diffusion. More
technological change are needed for radical innovations to become specifically, these interactions occur at all levels of the MLP and in four
diffused: individuals are much less likely to adopt radical innovations in broad dimensions (i.e., parts of the system): (a) the users (adopters and
stable regimes (Geels et al., 2016; Schot et al., 2016; Whitmarsh, 2012). user groups), (b) policy (including regulations and politicians), (c) in-
Lastly, at the micro level is the sociotechnical niche where socio- dustry (firms and infrastructure), and (d) culture (society) (Kanger
technical elements such as user preferences or practices and technolo- et al., 2019; Mylan, Morris, Beech, & Geels, 2019).
gies are not aligned but co-evolve (Geels, 2005). Radical innovations’ Adoption factors, explaining and predicting adoption by individuals
benefits accumulate, or increasingly become perceived as beneficial, as at the micro level of analysis, can be better understood by studying the
firms increase investments, experiment and learn about adaptations, as events and interactions in the system that lead to their emergence
infrastructure and regulation are established and as legitimization (Burton-Jones et al., 2015). For example, a radical innovation’s (per-
among potential users increases (Kanger et al., 2019; Schot & Geels, ceived) benefits (i.e., “performance expectancy”; Venkatesh et al.,
2008; Schot et al., 2016). 2012) are not static, but change and emerge through events, activities,
As a system approach, the MLP allows researchers to investigate and opposing forces in the niche, regime, and the landscape (Kanger
multi-dimensional interactions, such as feedback mechanisms, e.g., et al., 2019; Mylan et al., 2019). In the illustrative case of electric ve-
enthusiastic users providing legitimacy or political support for radical hicles, benefits crucially depend, for example, on support in the policy
innovations, and conflicts between actors (Geels, 2010; Roberts & dimension, such as tax breaks (Dijk, Weels, & Kemp, 2016). However,
Geels, 2019; Smith & Raven, 2012). Similarly, the MLP provides a lens without a landscape pressure on the regime to break institutional or
to study how landscape pressures destabilize regimes and create win- political lock-in, such as through growing cultural support for climate
dows of opportunity for actors to increase the momentum of niches change intervention, tax breaks may not be introduced (Raven, Kern,
(Figenbaum, 2017). Researchers using ST approaches such as the MLP Verhees, & Smith, 2016; Roberts & Geels, 2019). Using this system
typically apply qualitative research methods – particularly case studies approach, any factor – e.g., a driver for or barrier to adoption – could
– to capture complexities in rich narratives, rather than reducing them therefore be conceptualized as a trigger of positive and negative in-
(Flyvbjerg, 2006; Geels et al., 2018; Tyfield, Ely, & Geall, 2015). teractions or feedback mechanisms that support or oppose the adoption
Recently, some researchers have discussed applying ST approaches or diffusion of a radical innovation (Dijk et al., 2016). It is, therefore,
to study adoption and diffusion or contrasted them with AD theory critical to consider how the ST approach can be used for con-
(Geels et al., 2018; Mignon & Bergek, 2016; Sovacool & Hess, 2017; textualization, i.e., to ensure the context is sufficiently conceptualized
Sovacool, 2017); others have applied them to analyze the diffusion of to study the adoption and diffusion of radical innovations.
radical innovations, such as sustainable energy systems (Geels & Attempts by AD theorists to contextualize phenomena, such as the
Johnson, 2018) and electric mobility (Kanger et al., 2019). However, no multi-level framework by Venkatesh et al. (2016), thus directly contrast
prior research has considered how ST approaches, such as the MLP, can with the MLP. For example, Venkatesh et al. (2016) suggest “environ-
be used to adapt AD theory. The next section builds on and advances mental attributes” to summarize macro-level contextual factors. How-
the existing literature by contrasting in detail AD- and ST-specific ever, landscapes “do not determine, but provide deep-structural ‘gra-
concepts. dients of force’” (Geels & Schot, 2007, p. 403) that allow researchers to
gain deeper understanding of how, for example, cultural or technolo-
5. Towards potential adaptations of AD theory for radical gical developments affect the likelihood of a change in individuals’
innovations behavior regarding adoption (Geels et al., 2018; Shove et al., 2012).
Potential adopters’ perceptions or preferences regarding, for ex-
5.1. Context adaptations ample, the benefits of a radical innovation cannot be assumed static as
they are not fully known ex-ante and also change over time (Schot
One category of limitations stems from the static assumptions about et al., 2016). As radical innovations carry no shared perceptions or
the elements comprising the context of AD theories. Rather than a cultural meanings in their emergence, ST scholars draw on system-level
single technology, an innovation can often be better understood, from a frames – i.e., collective sense-making that guides perceptions – to re-
more systemic view, as multiple, complementary technologies in com- duce the complexity of heterogenous consumer perceptions (Bergman,
bination with changing user practices (Kanger & Schot, 2016). In the 2017; Geels & Verhees, 2011; Sovacool & Axsen, 2018). For example, as
illustrative case of electric mobility, this would include not only electric people want cars that are safer and more functional, they tend to buy
vehicles but also apps and new transport behaviors such as sharing bigger vehicles, thereby contradicting their own societal concerns over
(Sovacool, Noel, & Orsato, 2017). Regarding “adopters,” the ST ap- the environment or oil dependence (Sovacool & Axsen, 2018). There-
proach allows researchers to capture their role beyond that of pur- fore, by studying frames and potential adopters’ internal perception
chasers or even prosumers (Schot et al., 2016), identifying how they co- struggles, researchers can better understand how potential adopters
construct innovations or influence other dimensions in the system, such perceive innovations (Sovacool & Axsen, 2018).
as policy, industry, or culture (Geels & Johnson, 2018; Kanger et al., Researchers have argued that typical diffusion processes can in-
2019), e.g., by legitimizing electric vehicles. Rather than studying troduce biases toward, for example, information and communication
“adoption” as decisions or one-time events, ST approaches investigate (Straub, 2009) and that AD theory could benefit from including multi-
behaviors multi-dimensionally as patterns of practices, meanings, and dimensional interactions instead of linear conceptualizations (Burton-
skills constrained by the regime stability formed by technologies, in- Jones et al., 2015). For instance, how potential users become persuaded
frastructure or regulations (Kanger & Schot, 2016; Shove, Pantzar, & by a radical innovation depends not only on diffusion-specific processes
Watson, 2012; Tyfield et al., 2015). – i.e., the communication of information about a technology – but also
contextual, higher-level system developments (Geels et al., 2018). In-
5.2. Theoretical approach adaptations stead of studying how individuals become persuaded, ST scholars in-
vestigate how innovations and their related niche changes attain
Based on the limitations regarding the context of AD theories, an- broader cultural legitimacy, i.e., persuading different relevant actors
other category of limitations stems from the inherent inability of typical and social groups that these changes are desirable (Kanger et al., 2019;
AD theories to capture the full extent of multi-dimensionality, dynamics Schot et al., 2016). For radical innovations, potential adopters are not
and interactions in the contexts necessary for radical innovations, i.e., simply persuaded in linear fashion; the legitimizing of radical innova-
sociotechnical systems. Here, we can draw on the evolutionary system tions and underlying cultural changes have to be understood as

537
M. Gruber Journal of Business Research 116 (2020) 535–541

“contested processes” (Geels & Verhees, 2011, p. 910). 6. Synthesis: usefulness vs. complexity
These changes thus occur as discursive struggles between the niche
and the regime, including industry coalitions, policy makers, user 6.1. Assessing research goals and benefits
groups, and other actors (Rosenbloom, Berton, & Meadowcroft, 2016).
Therefore, ST scholars study discourses (e.g., on the future of mobility) How exactly to synthesize the AD and ST approaches depends on
and frames on specific topics (e.g., battery technology) relevant to ra- researchers’ goals – i.e., explaining or understanding the diffusion of a
dical innovations (Bergman, 2017; Sovacool & Hess, 2017). These radical innovation – and theories’ ability to reflect phenomena in a
frames range from media to politics and technology, and unite different suitable manner for these goals. Therefore, researchers must balance
stakeholders (Brown & Sovacool, 2017). For example, proponents of additional benefits against increased complexity (Burton-Jones et al.,
sustainability in industry, media, politics, and the public understand it 2015).
as a solution for climate change, whereas opponents delegitimize such To guide such decisions, researchers – as well as managers and
innovations by arguing that batteries are environmentally harmful or policy makers – should recognize two main benefits of the evolutionary
that tax subsidies are too expensive (Roberts, 2017; Rosenbloom et al., system perspective using the MLP. First, it provides specialized con-
2016). Proponents also engage in creating positive discourses, visions, ceptualizations and rich narratives, enabling deeper understanding of
and future expectations about the niche, aiming to further improve a how radical innovations become adopted and diffused. Besides usefully
shared understanding and even enthusiasm that legitimize the radical informing policy making or business strategy, this also facilitates ex-
innovation and niche (Bergman, 2017; Geels & Verhees, 2011). In turn, plaining and predicting adoption, e.g., when “the basis of the ex-
growing legitimacy and user adoption create a positive feedback me- planation is path dependence” (Ortiz de Guinea & Webster, 2017, p.
chanism and self-reinforcing patterns that increase investments in a 149). Second, researchers can benefit from the ST approach by using it
radical innovation and boost the niche’s political power, e.g., as firms’ as a lens to guide potential adaptations to AD theory. However, they
lobbying positions are strengthened (Roberts & Geels, 2019). Hence, in must also consider to what degree additional complexity and abstrac-
order to understand the communication process of diffusion and how tion should be included. The following subsection offers guidance for
potential users become persuaded of radical innovations, researchers decisions about adapting AD theory in future research.
need to consider the visions, discourses, and framing contests on the
MLP’s multiple levels of analysis. 6.2. Suggestions for future research
Capturing such contradictions and discursive struggles is essential
as complex systems are socially constructed; accordingly, research ap- 6.2.1. Methodological adaptations
proaches prioritizing single truths or attempting to reach closure should This article suggests future research to build new theory through a
be replaced by approaches incorporating different accounts of reality system perspective that views adoption and diffusion of radical in-
and continuous change (Arthur, 2015; Ismail, 2015). This is crucial for novations as configurations of micro-, meso-, and macro-level elements,
radical innovations as they are embedded in complex systems, such that and allows exploration of how these elements align and interact.
what constitutes a beneficial innovation is often contested and contra- Repeating configurations of relationships between multiple elements
dictory. Human affairs are more open, soft, and probabilistic than sta- constitute patterns, e.g., patterns of diffusion or feedback cycles (Geels,
tistical truth; since human affairs occur continuously and on multiple 2005; Montalvo & Kemp, 2008). Incorporating a system perspective
levels of analysis, system approaches offer an alternative to variance does, however, require researchers to make methodological changes,
approaches or linear process approaches (Burton-Jones et al., 2015). such as moving from measuring co-variance and sequential processes to
mapping patterns (Gleick, 1987; Vargo et al., 2017). Future research
5.3. Limitations of the MLP could draw on research methods typically used in ST approaches, such
as qualitative exploratory and explanatory methods with single or
As this article focuses on adapting AD theory, particularly in the multiple cases as units of analysis. Case studies are particularly useful to
context of radical innovations, care must be taken not to consider too study complex phenomena dependent on their real-world contexts (Yin,
much of the “whole”, as the level of abstraction becomes higher when 2017), such as the adoption and diffusion of radical innovations as
taking the system perspective (Vargo et al., 2017). The conceptualiza- explained in sections 1 and 5. For example, researchers using the MLP
tions of ST theories are, in fact, often metaphorical and coarse-grained, commonly analyze documents – e.g., parliamentary decisions and white
rather than fine-grained and specific (Roberts & Geels, 2019). More- papers, historical or contemporary news articles (Figenbaum, 2017) –
over, the system approach has limitations in understanding specific but also conduct and analyze in-depth interviews with experts and key
concepts and relationships that require the statistical precision offered actors (Nykvist & Nilsson, 2015), providing rich and varied data to
by the variance approach (Burton-Jones et al., 2015). Therefore, it is build theory with reliability and validity (Eisenhardt & Graebner, 2007;
important to preserve the advantages of AD theories. This raises the Yin, 2017).
question to what degree evolutionary elements – i.e., uncertainty, As a key strength of extant AD theory lies in measuring AD factors
multi-dimensional interactions, and co-evolution – should be added. with high statistical accuracy, such factors could be used to establish
While the MLP offers a lens to study micro-contexts, e.g., particular focus on phenomena of interest and thus reduce the complexity and
projects (Geels et al., 2016), it is not specialized for the demand side abstraction inherent in the MLP. As Geels and Verhees (2011, p. 916)
alone (Shove et al., 2012). What must be identified are the patterns of pointed out, “the aim is not to ‘tell the story as it really was’ but to have
change regarding adoption and diffusion, rather than those behind a ‘more selective analytical focus’”. MLP-based studies frequently also
transitions of entire sociotechnical systems. Therefore, evolutionary use quantitative data, for example to identify tipping points or as
elements and system-perspective insights should be related to phe- temporal backbones (Roberts & Geels, 2019).
nomena of interest, such as why individuals “adopt” – become users of – However, while AD and ST theories have been combined (e.g.,
electric mobility or how electric vehicles develop perceived benefits Sovacool, 2017), more research is necessary to blend the different ap-
among potential users. Hence, the challenge for researchers lies in how proaches’ respective advantages and to overcome their disadvantages
exactly to combine the different approaches and methodologies as well (Ortiz de Guinea & Webster, 2017). The following subsection suggests
as their advantages and disadvantages. The next (and final) section of key research questions that combine the approaches’ strengths and
this article discusses several tentative suggestions for how to synthesize focus on adoption and diffusion related phenomena.
the different approaches in future research, while Table 1 summarizes
the main strengths and limitations of the three different theoretical 6.2.2. Research questions
approaches of AD theory with respect to radical innovations. Researchers can use the system perspective of the MLP to “zoom

538
M. Gruber Journal of Business Research 116 (2020) 535–541

out” of the context of individual adopters and micro phenomena (i.e., Sovacool, 2018). Hence, a final, tentative suggestion for future research
parts) in order to understand higher-level influences at the system-level is to investigate the question: What alignment patterns lead to accelerated
(i.e., the whole). Hence, a key future research question is: How do diffusion rates?
system-level phenomena affect micro-level phenomena regarding the adop-
tion or diffusion of radical innovations?
The system perspective can also provide deeper understanding of 7. Conclusion
AD-specific factors, such as “perceived benefits” and “facilitating fac-
tors.” Future research could, for example, investigate how culture, This article presents an evolutionary perspective on AD theory and
policy, and industry affect adoption factors of electric vehicles, such as suggests how the MLP – an evolutionary system perspective – can
range, charging infrastructure, and other perceived range issues com- provide a deeper understanding of adoption and diffusion, as well as
monly synopsized as “range anxiety.” Research shows that the majority facilitate the adaptation of AD theory for radical innovations (e.g.,
of people drive less than 20 km a day (Sovacool, Kester, Noel, & electric mobility). Typical AD theories, which explain or predict
Zarazua de Rubens, 2018, p. 90), but also indicate that the peak of adoption or diffusion through factors (e.g., range anxiety) or sequences
potential adopters’ willingness to pay for range is at around 150 km of critical events (e.g., learning about an innovation’s benefits), are
(Noel et al., 2019) and, similarly, that above a range of 200 miles more increasingly challenged in the context of radical innovations. The
than half of potential users consider buying an electric vehicle (Beard adoption or diffusion of radical innovations depend on a broader con-
et al., 2019). This could imply that while people would buy or use text, i.e., a complex system where socio-technical elements ranging
electric vehicles with a lower range compared to conventional vehicles, from technology, culture and user preferences to policy and industry co-
and therefore a considerably lower purchase price, most manufacturers evolve.
would prioritize shorter ranges. Similarly, studies suggest that the vast This article suggests that the MLP provides a useful lens to con-
majority of people currently charge their electric vehicles at home textualize the phenomena of radical innovations’ adoption and diffu-
(Figenbaum, 2018), yet some manufacturers (e.g., Tesla) argue that sion by transcending taken-for-granted assumptions, static factors, and
rapid chargers are a priority (Tesla Inc., 2018). Researchers could, thus, linear processes, and incorporating multi-dimensional interactions. For
use the MLP to explore how different approaches by industry or policy example, macro-level landscape pressures, such as growing public
regarding range (e.g., shorter vs. longer range in the same price class) support for environmentalism, can destabilize meso-level regime stabi-
or charging (e.g., slow or fast), as well as shared cultural meanings lity, such as political or cultural support for internal combustion engine
(frames) of automobility affect potential adopters’ range anxiety and vehicles. Adoption-specific factors, e.g., the “perceived benefits” of ra-
ultimately their adoption or use of electric vehicles. This raises another dical innovations, emerge and change over time, and are thus influ-
question for future research: How do system-level influences affect micro- enced by higher-level interactions in culture, policy, and industry.
level adoption or diffusion factors? System-level phenomena, such as changes in policy, can also be influ-
By capturing, rather than reducing, complexities in narratives, re- enced, in turn, by the feedback mechanisms of increasing diffusion and
searchers can investigate the framing struggles, motivations, and cultural legitimacy of an innovation. Studying adoption and diffusion as
choices of different actors, as well as contradictions in patterns of configurations of co-evolving multi-dimensional elements requires
change (Langley et al., 2013), in order to gain a deeper understanding methodological changes, for example from measuring co-variances to
of, for example, AD-specific events. For example, to better understand mapping patterns. Furthermore, by case-studying the adoption and
why policy or industry choose different actions to support or oppose a diffusion of radical innovations, researchers could gain more in-depth
radical innovation, future research should raise the question: Why do understanding, as rich narratives reveal not only single truths but also
different adoption-diffusion relevant phenomena at the system-level occur? the contradictions and struggles of different actors – capturing, rather
The system perspective could also be used in order to study how than reducing, complexities.
increasing diffusion affects other elements in the system, such as how This article hence advances the literature in several crucial ways.
enthusiasm and cultural legitimacy form around specific innovation First, it answers recent calls by AD theorists to use a system approach by
features and provide a feedback mechanism for culture, industry, and showing how the MLP can be used as a lens to study adoption and
policy. This raises the question: How do micro-level developments affect diffusion (Burton-Jones et al., 2015; Ortiz de Guinea & Webster, 2017).
system-level phenomena? Second, while scholars have previously used ST approaches to study
Scholars could use longitudinal research methods to investigate how adoption and diffusion (e.g., Kanger et al., 2019), and combined both
properties or phenomena evolve in changing contexts and different approaches (e.g., Sovacool, 2017), this article specifically compared
time periods. For example, they could explore how a radical innova- key concepts and traits of the respective approaches to discuss in detail
tion’s perceived benefits emerge or change over time, and in co-evo- how extant AD theory can be adapted for radical innovations. Third,
lution with the context (i.e., its system). Using the MLP could, therefore, this article illustrates how adoption and diffusion can be conceptualized
help to answer: How do properties emerge and change over time? as configurations of micro-, meso-, and macro-level elements, and
By studying adoption and diffusion as configurations and how suggests several research questions to guide future explorations of the
micro-level elements align or interact with system-level elements (e.g., patterns of adoption and diffusion of radical innovations. For example,
innovation characteristics and landscape trends or opportunities), re- identifying the patterns of feedback mechanisms, breakthroughs, or
searchers could conceptualize AD-specific alignment patterns. This accelerators for diffusion would be useful to researchers, firms, and
prompts a suggestion for future research to explore: What are the policy makers. Finally, this article offers suggestions for combining
alignment patterns for adoption and diffusion? The alignment of critical theoretical approaches of extant AD and ST theory, and recommends
elements can increase momentum of niches and lead to accelerated suitable research methods, such as exploratory case studies to construct
diffusion (Roberts & Geels, 2019; Kester, Noel, Zarazua de Rubens, & new theory.

Appendix A

See Table 1.

539
M. Gruber Journal of Business Research 116 (2020) 535–541

Table 1
Strengths and limitations of the three core approaches to the adoption and diffusion of radical innovations.
Theory Strengths Limitations

Adoption-diffusion (AD) Specific contexts that enable the construction of theory focused on Set assumptions in one context might be too specific to build useful
theory adoption and diffusion. theory for adoption or diffusion cases affected by factors or processes
outside that context.
Parsimony (simplicity) and high specificity of theory. Extensions and contextualizations are needed for different, broader
contexts.
Potential biases on information, communication, rational choice, or
superiority of innovations.

Factor approach in AD High variance and strength of prediction of individuals’ adoption or Static concepts and deterministic relationships in variance models risk
theory rates of diffusion. overlooking contextual system influences. No time dimension, so no
understanding of how independent factors lead to dependent factors;
how factors arise; how they affect each other; how they are affected by
events; or when they are important.

Process approach in AD Understanding how adoption and diffusion occur over time; modeling Sequence of events is often too linear, i.e. cannot account for iterations
theory growth curves and forecasting; identifying important events and and interactions at multiple levels of analysis.
sequences in adoption and diffusion processes. Some measurements are not possible as properties emerge or depend on
complex systems.

System-approach multi- Ability to include evolutionary elements such as uncertainty, Danger of considering too many factors and processes outside the
level perspective cumulative change, interdependence, and co-evolution of elements. relevant context.
Allows capturing contextual factors and processes and structuring them
on multiple levels of analysis.
Building narratives that capture evolutionary elements, contextual
factors and processes, as well as interactions and contradictions,
enabling realistic understanding of adoption and diffusion for radical
innovations.
Useful for AD cases of radical innovations and where system effects
must be considered.

References challenges in socio-technical transitions research and the Multi-Level Perspective.


Energy Research & Social Science, 37, 224–231.
Geels, F. W., Kern, F., Fuchs, G., Hinderer, N., Kungl, G., Mylan, J., …, Wassermann, S.
Arthur, W. B. (1988). Competing technologies: An overview. In G. Dosi, C. Freeman, R. (2016). The enactment of socio-technical transition pathways: a reformulated ty-
Nelson, G. Silverberg, & L. Soete (Eds.). Technical change and economic theory (pp. pology and a comparative multi-level analysis of the German and UK low-carbon
590–607). London: Pinter. electricity transitions (1990–2014) Research Policy, 45, 896–913.
Arthur, W. B. (2015). Complexity and the economy. New York: Oxford University Press. Geels, F. W., & Johnson, V. (2018). Towards a modular and temporal understanding of
Beard, G., Durrell, L., Kent, J., Skippon, S., Kinnear, N., Al-Katib, H., & Anable, J. (2019). system diffusion: Adoption models and socio-technical theories applied to Austrian
Consumer uptake trial report: Mainstream consumers’ attitudes and willingness to adopt biomass district-heating (1979–2013). Energy Research & Social Science, 38, 138–153.
BEVs and PHEVs. Transport Research Laboratory. Geels, F. W., & Schot, J. (2007). Typology of sociotechnical transition pathways. Research
Bergman, N. (2017). Stories of the future: Personal mobility innovation in the United Policy, 36(3), 399–417.
Kingdom. Energy Research & Social Science, 31, 184–193. Geels, F. W., Schwanen, T., Sorrell, S., Jenkins, K., & Sovacool, B. K. (2018). Reducing
Brown, G., & Sovacool, B. K. (2017). The presidential politics of climate discourse: Energy energy demand through low carbon innovation: A sociotechnical transitions per-
frames, policy, and political tactics from the 2016 Primaries in the United States. spective and thirteen research debates. Energy Research & Social Science, 40, 23–35.
Energy Policy, 111, 127–136. Geels, F. W., & Verhees, B. (2011). Cultural legitimacy and framing struggles in in-
Burton-Jones, A., McLean, E. R., & Monod, E. (2015). Theoretical perspectives in IS re- novation journeys: A cultural-performative perspective and a case study of Dutch
search: From variance and process to conceptual latitude and conceptual fit. European nuclear energy (1945–1986). Technological Forecasting & Social Change, 78(6),
Journal of Information Systems, 24, 664–679. 910–930.
Cappelli, P., & Sherer, P. D. (1991) The missing role of context in OB: The need for a Giesler, M., & Thompson, C. J. (2016). Process Theorization in Cultural Consumer
meso-level approach. In L. L. Cummings, & B. M. Staw (Eds.), Research in organi- Research. Journal of Consumer Research, 43(4), 497–508.
zation behavior (Vol. 33, pp. 55–110). Gleick, J. (1987). Chaos: Making a new science. New York: Penguin.
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer Gnann, T., Stephens, T. S., Lin, Z., Plötz, P., Liu, C., & Brokate, J. (2018). What drives the
technology: A comparison of two theoretical models. Management Science, 35(8), market for plug-in electric vehicles? - A review of international PEV market diffusion
982–1003. models. Renewable & Sustainable Energy Reviews, 93, 158–164.
Dijk, P., Wells, & Kemp, R. (2016). Will the momentum of the electric car last? Testing an Hojnik, J., & Ruzzier, M. (2016). What drives eco-innovation? A review of an emerging
hypothesis on disruptive innovation. Technological Forecasting & Social Change, literature. Environmental Innovation and Societal Transitions, 19, 31–41.
105, 77–78. Ismail, A. (2015). A review of theoretical approaches on diffusion analysis: Discussing
Dijk, M., Kemp, R., & Valkering, P. (2013). Incorporating social context and co-evolution issues involved in the adoption of ICT services in a complex socio-economic context.
in an innovation diffusion model – with an application to cleaner vehicles. Journal of Australian Journal of Sustainable Business and Society, 1(1), 97–108.
Evolutionary Economics, 23, 295–329. Kanger, L., Geels, F., Sovacool, B. K., & Schot, J. W. (2019). Technological diffusion as a
Eisenhardt, K. M., & Graebner, M. E. (2007). Theory building from cases: Opportunities process of societal embedding: Lessons from historical automobile transitions for
and challenges. Academy of Management Journal, 50(1), 25–32. future of electric mobility. Transportation Research Part D: Transport and Environment,
Fagnant, D. J., & Kockelman, K. (2015). Preparing a nation for autonomous vehicles: 71, 47–66.
Opportunities, barriers and policy recommendations. Transportation Research Part A: Kanger, L., & Schot, J. (2016). User-made immobilities: A transitions perspective.
Policy and Practice, 77, 167–181. Mobilities, 11(4), 598–613.
Fichter, K., & Clausen, J. (2016). Diffusion dynamics of sustainable innovation – insights Kasmire, J., Korhonen, J. M., & Nikolic, I. (2012). How radical is a radical innovation? An
on diffusion patterns based on the analysis of 100 sustainable product and service outline for a computational approach. Energy Procedia, 20, 346–353.
innovations. Journal of Innovation Management, 4, 30–67. Kester, J., Noel, L., Zarazua de Rubens, G., & Sovacool, B. K. (2018). Policy mechanisms to
Figenbaum, E. (2017). Perspectives on Norway’s supercharged electric vehicle policy. accelerate electric vehicle adoption: A qualitative review from the Nordic region.
Environmental Innovation and Societal Transitions, 25, 14–34. Renewable & Sustainable Energy Reviews, 94, 719–731.
Figenbaum, E. (2018). Electromobility status in Norway. TØI-rapport 1627/2018. Langley, A., Smallman, C., Tsoukas, H., & Van de Ven, A. H. (2013). Process studies of
Flyvbjerg, B. (2006). Five misunderstandings about case study research. Qualitative change in organization and management: Unveiling temporality, activity, and flow.
Inquiry, 12, 219–245. Academy of Management Journal, 56(1), 1–13.
Geels, F. W. (2005). The dynamics of transitions in socio-technical systems: A multi- level Lyytinen, K., & Damsgaard, J. (2001). What’s wrong with the diffusion of innovation
analysis of the transition pathway from horse-drawn carriages to automobiles theory? In M. A. Ardis, & B. L. Marcolin (Eds.). Diffusing software product and process
(1860–1930). Technology Analysis & Strategic Management, 17(4), 445–476. innovations (pp. 173–190). New York: Springer.
Geels, F. W. (2010). Ontologies, socio-technical transitions (to sustainability), and the Makkonen, H., Johnston, W. J., & Javalgi, R. G. (2016). A behavioral approach to or-
multi-level perspective. Research Policy, 39, 495–510. ganizational innovation adoption. Journal of Business Research, 69(7), 2480–2489.
Geels, F. W. (2018). Disruption and low-carbon system transformation: Progress and new Mignon, I., & Bergek, A. (2016). System- and actor-level challenges for diffusion of

540
M. Gruber Journal of Business Research 116 (2020) 535–541

renewable electricity technologies: An international comparison. Journal of Cleaner Shove, E. (2010). Beyond the ABC: Climate change policy and theories of social change.
Production, 128, 105–115. Environment and Planning A, 42(6), 1273–1285.
Mitchell, M. (2009). Complexity: A guided tour. New York, NY: Oxford University Press. Shove, E., Pantzar, M., & Watson, M. (2012). The dynamics of social practice: Everyday life
Montalvo, C., & Kemp, R. (2008). Cleaner technology diffusion: Case studies, modeling and how it changes. London: SAGE.
and policy. Journal of Cleaner Production, 16(Supplement 1), 1–6. Smith, A., & Raven, R. P. J. M. (2012). What is protective space? Reconsidering niches in
Mylan, J., Morris, C., Beech, E., & Geels, F. W. (2019). Rage against the regime: Niche- transitions to sustainability. Research Policy, 41(6), 1025–1036.
regime interactions in the societal embedding of plant-based milk. Environmental Sovacool, B. K. (2017). Experts, theories, and electric mobility transitions: Toward an
Innovation and Societal Transitions, 31, 233–247. integrated conceptual framework for the adoption of electric vehicles. Energy
Nelson, R. R., & Winter, S. G. (1982). An evolutionary theory of economic change. Research & Social Science, 27, 78–95.
Cambridge, MA: Harvard University Press. Sovacool, B. K., & Axsen, J. (2018). Functional, symbolic and societal frames for auto-
Noel, L. D., Carrone, A. P., Jensen, A. F., Zarazua de Rubens, G., Kester, J., & Sovacool, B. mobility: Implications for sustainability transitions. Transportation Research Part A:
K. (2019). Willingness to pay for electric vehicles and vehicle-to-grid applications: A Policy Practice, 730–746.
Nordic choice experiment. Energy Economics, 78, 525–534. Sovacool, B., & Hess, D. J. (2017). Ordering theories: Typologies and conceptual frame-
Nykvist, B., & Nilsson, M. (2015). The EV paradox—A multilevel study of why Stockholm works for sociotechnical change. Social Studies of Science, 47(5), 703–750.
is not a leader in electric vehicles. Environmental Innovation and Societal Transitions, Sovacool, B. K., Kester, J., Noel, L., & Zarazua de Rubens, G. (2018). The demographics of
14, 26–44. decarbonizing transport: The influence of gender, education, occupation, age, and
Ortiz de Guinea, A. O., & Webster, J. (2017). Combining variance and process in in- household size on electric mobility preferences in the nordic region. Global
formation systems research: Hybrid approaches. Information and Organization, 27(3), Environmental Change, 52, 86–100.
144–162. Sovacool, B. K., Noel, L., & Orsato, R. J. (2017). Stretching, embeddedness, and scripts in
Peres, R., Muller, E., & Mahajan, V. (2010). Innovation diffusion and new product growth a sociotechnical transition: Explaining the failure of electric mobility at better place
models: A critical review and research directions. International Journal of Research in (2007–2013). Technological Forecasting and Social Change, 123, 24–34.
Marketing, 27(2), 91–106. Straub, E. (2009). Understanding technology adoption: Theory and future directions for
Raven, R., Kern, F., Verhees, B., & Smith, A. (2016). Niche construction and empower- informal learning. Review of Educational Research, 79, 625–649.
ment through socio-political work. A meta-analysis of six low-carbon technology Tesla Inc. (2018). Annual report 2018. Available online at: https://www.sec.gov/
cases. Environmental Innovation and Societal Transitions, 18, 164–180. Archives/edgar/data/1318605/000156459019003165/tsla-10k_20181231.htm (ac-
Rip, A., & Kemp, R. (1998). Technological change. In: S. Rayner, & E. L. Malone (Eds.), cessed on 2 December 2019).
Human choice and climate change (Vol. 2, pp. 327–399). Columbus, Ohio: Battelle Tyfield, D., Ely, A., & Geall, A. (2015). Low carbon innovation in China: From overlooked
Press. opportunities and challenges to transitions in power relations and practices.
Roberts, C. (2017). Discursive destabilisation of socio-technical regimes: Negative Sustainable Development, 23(4), 206–216.
storylines and the decline of the American railroads. Energy Research and Social Vargo, S. L., Koskela-Huotari, K., Baron, S., Edvardsson, B., Reynoso, J., & Colurcio, M.
Science, 31, 86–99. (2017). A systems perspective on markets–Toward a research agenda. Journal of
Roberts, C., & Geels, F. W. (2019). Conditions for politically accelerated transitions: Business Research, 79, 260–268.
Historical institutionalism, the multi-level perspective, and two historical case studies Venkatesh, V., Thong, J. Y. L., & Xu, X. (2012). Consumer acceptance and use of in-
in transport and agriculture. Technological Forecasting and Social Change, 140, formation technology: Extending the unified theory of acceptance and use of tech-
221–240. nology. MIS Quarterly, 36, 157–178.
Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: Free Press. Venkatesh, V., Thong, J. Y. L., & Xu, X. (2016). Unified theory of acceptance and use of
Rosenbloom, D., Berton, H., & Meadowcroft, J. (2016). Framing the sun: A discursive technology: A synthesis and the road ahead. Journal of the Association for Information
approach to understanding multi-dimensional interactions within socio-technical Systems, 17(5), 328–376.
transitions through the case of solar electricity in Ontario, Canada. Research Policy, Weber, R. (2012). Evaluating and developing theories in the information systems dis-
45(6), 1275–1290. cipline. Journal of the Association for Information Systems, 13(1), 2–30.
Schot, J., & Geels, F. W. (2008). Strategic niche management and sustainable innovation Whitmarsh, L. (2012). How useful is the Multi-Level Perspective for transport and sus-
journeys: Theory, findings, research agenda, and policy. Technology Analysis & tainability research? Journal of Transport Geography, 24, 483–487.
Strategic Management, 20, 537–554. Williams, R., & Edge, D. (1996). The social shaping of technology. Research Policy, 25(6),
Schot, J., Kanger, L., & Verbong, G. (2016). The roles of users in shaping transitions to 865–899.
new energy systems. Nature Energy. Yin, R. K. (2017). Case study research: Design and methods (6th ed.). Sage Publications.

541

You might also like