Reflexive layers of influence ( RLI):
A model of social influence, vehicle purchase behavior, and pro- societal values
Jonn Axsen a* and Kenneth S. Kurani a
Working Paper
July 30 th , 2010
Affiliations:
a Institute of Transportation Studies, University of California at Davis, 2028 Academic
Surge, One Shields Avenue, Davis, CA, 95616, U. S. A.
* Corresponding author. Tel.: 1 530 574 2150, Fax: 1 530 752 6572, E- mail address:
jaxsen@ ucdavis. edu
i
ABSTRACT
Understanding consumer purchase behavior will facilitate the successful deployment of new
vehicle technologies that offer societal benefits— such as plug- in hybrid vehicle ( PHEVs). To
move beyond the rational actor model and similarly simplified behavioral approaches, this paper
proposes an integrative, theoretically rich alternative: reflexive layers of influence ( RLI). RLI is
a framework that accounts for the role of social influence in an actor’s ( or car buyer’s)
development of pro- societal values and purchase behavior. Informed by a multidisciplinary
literature review and empirical observation, RLI represents three layers that underlie the actor’s
behavior; social influence is characterized by different processes at each layer. The bottom layer
is the actor’s functional awareness of the vehicle, which can be influenced by the diffusion of
simple information— such as the vehicle’s existence and basic purpose— from other actors or
information sources. Next is the actor’s assessment of the vehicle based on perceived private and
societal benefits. Assessment can be socially influenced through processes of translation, where
the actor forms interpretations based on self- concept and group membership, and negotiates
these interpretations through social interactions. The third layer is the actor’s self- concept— and
associated values and lifestyle practices. Through reflexivity, self- concept can serve to frame the
actor’s assessment, and can also be reinforced or altered according to their assessment, behavior,
social interactions and perceived lifestyle practices of other actors. This paper applies the RLI
framework to participants in a PHEV demonstration project as well as policy considerations.
1
1. INTRODUCTION
This paper proposes a conceptual framework to understand the role of social influence in
consumer purchase behavior and, in particular, the development of pro- societal values. Such
questions of how and under what conditions consumers buy new products ( or “ innovations”) are
central to the successful deployment of new vehicle technologies with societal benefits. To
explore such processes, transportation researchers and policymakers rely on behavioral models—
sets of assumptions about what consumers do and why. Jackson ( 2005) outlines five broad
categories of consumer behavior model relating to pro- societal consumption:
1. Expectancy- value models, such as the rational choice model, ( e. g. Train, 1980),
where consumers are assumed to be rational, deliberative and autonomous,
calculating costs and benefits from multiple options and selecting the one that
maximizes their net benefit.
2. Adjusted- expectancy- value models, such as the theory of reasoned action ( Ajzen and
Fishbein, 1980), which maintain the notion of cognitive deliberation, but also account
for the influence of subjective norms regarding a behavior.
3. Normative models, such as value- belief- norm theory ( Stern, et al., 1999),
representing the development of personal pro- environmental norms based on strong
altruistic or biospheric values.
4. Habit models, such as heuristics ( Tversky and Kahneman, 1974), which account for
the cognitive limitations of consumers.
5. Sociality models, such as symbolic- interactionism ( Blumer, 1969), where consumers
negotiate and create meaning for different products through social interactions.
Currently, transportation research and, in particular academic research into markets for
alternative fuel and electric- drive vehicles, is dominated by the expectancy and adjusted
expectancy choice models— particularly the rational choice model— despite the fact that
behavioral economists, psychologists and sociologist have long established that consumers do
not typically follow “ rational” decision processes, e. g. ( Thaler and Sunstein, 2003; Tversky and
Kahneman, 1974). Jackson ( 2005) states that “ a useful model has to account for: motivations,
attitudes and values; contextual or situational factors; social influence; personal capabilities; and
habits.” Jackson’s review suggests an accretion of multiple layers of understanding consumer
behavior; an integrative approach may yield important insights.
The proposed framework— reflexive layers of influence ( RLI)— incorporates some such depth,
with particular focus on improving representations of one behavioral determinant that
transportation researchers are only beginning to explore: social ( or interpersonal) influence ( e. g.
Carrasco, et al., 2008). In other words, RLI represents how a consumer’s perceptions, values and
actions are influenced by interactions with people in their social network. RLI is based on
literature review and empirical observations of interpersonal interactions within a study of social
networks
The next two sections summarize the literature review and empirical study, respectively. Section
4 details the RLI framework, including definitions and drivers of social influence. Section 5
illustrates the application of RLI to two case studies ( narratives) constructed from interviews
2
with participants in a plug- in hybrid electric vehicle ( PHEV) demonstration project, and Section
6 considers implications for policy. The paper then discusses the usefulness of RLI, and proposes
hypotheses for future exploration.
2. THREE RESEARCH PERSPECTIVES ON SOCIAL INFLUENCE
Across behavioral disciplines, there are several different perspectives and theories of social
influence. This section summarizes three such perspectives, each labeled according to the
assumed process of influence: diffusion, translation, and reflexivity. These perspectives are
discussed in more detail and applied by Axsen and Kurani ( 2010).
First, is the most common approach to the adoption of new technologies: diffusion of
innovations ( DOI) ( Rogers, 2003). Diffusion is “ the process in which an innovation is
communicated through certain channels over time among the members of a social system”
( Rogers, 2003). The adoption process is assumed to driven by the communication of information,
flowing from innovators ( members of the first group to adopt an idea) and early adopters
( members of the second) to individuals in later categories ( early majority, later majority and
laggards). DOI has been criticized for many drawbacks, including a lack of emphasis on
underlying motivations to adopt ( Rogers, 2003) including an over- reliance on the tautology of
“ innovativeness” as a personality trait ( Hirschman, 1980) and inability to account for dynamics
in social and technology systems ( Blaut, 1987).
Second, translation treats new technologies as dynamic, socially- constructed artifacts ( Bruun and
Hukkinen, 2003). At first, a newly introduced artifact has a high degree of interpretive
flexibility; different social groups may have differing interpretations of its meaning and content
which influences further technological development ( Pinch and Bijker, 1984). Eventually, a state
of closure or stabilization is reached as the interpretations of various social groups converge
( Bruun and Hukkinen, 2003) or align ( Callon, 1991; Hannemyr, 2003). As a new technology is
introduced and developed, the social groups themselves may be redefined and transformed, such
as through grouping around a certain cause or idea ( Kline and Pinch, 1996), and the entire social
system itself may shift in response to the technology in question ( Law, 1992). Adoption of the
artifact is driven across groups by this process of translation; new ideas and objects are
negotiated and altered as a result of context and interactions among social groups ( Pentland and
Feldman, 2007).
Third, reflexivity is drawn from Giddens’ ( 1991) structuration approach to the relationship
between self development and social structures in an uncertain, modern social world lacking the
roles and expected behaviors enforced by tradition. In this modern world, individuals must
actively seek out and define their self identity, taking on “ a reflexive project” ( Giddens, 1991). In
this paper, reflexivity is taken to be the dynamic, continuous, self- aware process of defining and
expressing oneself. An individual’s behavior is guided by efforts to establish a sense of order,
direction, and development for their self- identity. As part of this project, individuals seek a
lifestyle as a package of practices that are associated with their particular trajectory, such as
fashion, eating, or any other “ means of symbolic display” ( Giddens, 1991). Thus, the purchase of
a pro- societal vehicle may not just be driven by a motivation for advantageous functional
3
benefits, but may also be one component, or trial, of a more fundamental expression of, or shift
towards, an environmental or socially- conscious lifestyle.
3. EMPIRICAL FINDINGS FROM PARTICIPANT ASSESSMENT OF PHEVS
To explore the above three research perspectives, and to inform the construction of the new
conceptual framework presented in this paper ( RLI) a multi- method, qualitative research project
was designed and implemented to collect empirical data on social influence. The sample was a
subset of 10 households participating in a PHEV demonstration project conducted in the
Sacramento, California region.
Research instruments included a series of semi- structured interviews. Each of the 10 “ primary”
households constructed a map of their social network, then recruited several of their friends or
family ( or other alters— members of their social network) to take part as “ secondary”
participants. Next, the primary household substituted a PHEV for their current vehicle for a
multi- week trial. They recorded any social interactions that pertained to the PHEV in a social
episode diary, and described these interactions in subsequent interviews. In the final interview,
the primary household assessed the PHEV technology, then ranked how influential each social
interaction was in regards to their PHEV assessment. Researchers then integrated this and other
collected data to construct a narrative for each primary household and secondary participant—
that is, a chronological, causal flow of experiences and events. The full methodology is
explained in Axsen ( 2010).
Application of the three perspectives to the elicited narratives suggest that diffusion, translation
and reflexivity provide different, but complementary concepts to help understand the social
dynamics and negotiations observed in this empirical study ( Axsen and Kurani, 2010). Results
also indicate that participating households were more likely to develop and value pro- societal
interpretations of the PHEV when three factors were present:
i) the household easily formed a basic understanding of functional aspects of PHEV
technology, e. g. that it can plug in and offset gasoline use with electricity use;
ii) the household discovered some degree of supportive pro- societal values within their
social network; and
iii) the household’s self- concept and associated lifestyle practices were in a state of
liminality, characterized by ambiguity, openness and indeterminacy ( Turner, 1969).
4. AN INTEGRATIVE PERSPECTIVE: REFLEXIVE LAYERS OF INFLUENCE ( RLI)
The RLI framework was developed to integrate the research perspectives and observed
behavioral patterns summarized above. This section explains RLI according to its
conceptualization of: the new technology in question, the social system boundaries, the timing of
purchase, and the processes of social influence that drive behavior, e. g. purchase. Where
appropriate, terminology and concepts are borrowed and adapted from the three perspectives
4
discussed above— though in doing so, each term is carefully redefined for the present application
in RLI. In some cases, new concepts are suggested based on the empirical observations.
The RLI framework is intended to be an integration of concepts that appropriately represents
consumer behavior, is parsimonious, and is useful for future application. This discussion frames
this model as generally as possible— allowing application to a variety of transportation ( and non-transportation)
technologies and behaviors with potential pro- societal benefits. However, here
focus is placed on the case of pro- societal vehicles like PHEVs.
4.1 The technology as a dynamic, socially- defined artifact
RLI describes a new pro- societal vehicle as an artifact ( as borrowed from the translation
perspective). This term highlights the culturally- embedded, socially- defined nature of
technology. In RLI, a new artifact is defined as an object that different actors may associate with
one or more ideas that they perceived as novel, including private and pro- societal benefits. Over
time the new artifact may be physically altered and its private and pro- societal benefits may be
reinterpreted by actors within the relevant social system.
4.2 System boundaries: Dynamics in the relevant social system
In RLI, the relevant social system includes all individual actors and social groups that can
influence potential adopters’ interpretations of the new artifact, including previous adopters,
observers, manufacturers, media, interest groups and governments. The term actor is a scalable
unit of analysis— it can be an individual, household, group of friends, company, or any other
social group that is relevant to the artifact in question and may at times act as single unit. The
term social group distinguishes a group of actors from an individual actor.
The presence of, and relationships between, relevant actors can shift over time— sometimes as a
result of the new artifact itself. For instance, the purchase and use of the artifact by an adopter in
one social group may stimulate conversation and interest from actors in a more socially distant
group. Through such interactions, new social ties could develop between the two social groups,
and the originally weak social ties strengthened. Thus, over the course of the adoption process,
the relevant social groups may be physically altered ( through the addition or removal of
members), or reinterpreted by an actor ( like the new artifact itself). For instance, a group of
actors may form around a new idea pertaining to PHEVs, such as oil independence, and their
negotiated interpretations of PHEVs may change as more group members adopt PHEVs and
interact.
4.3 The timing of adoption: Aligning context and actor interest
Determining if and when an actor adopts a new artifact involves the interaction of several
factors, which within RLI are divided into contextual and actor factors. Contextual factors are
external to the actor, and can include the actor’s:
• disposable income;
• life stage, e. g. student, career, young family, or retired; and
5
• cycle of stock turnover, i. e. when a new artifact is needed or when an incumbent artifact
needs to be replaced or displaced.
Contextual factors also include characteristics of the market, such as:
• price of artifact;
• availability of the artifact, including variety of models and features, dealership location,
supply constraints and waiting lists;
• government policy relating to the artifact, e. g. tax rebates, special privileges, codes and
standards; and
• marketing and advertising, or more generally, messaging efforts related to the artifact.
The importance of contextual factors can be illustrated for a hypothetical actor considering an
HEV. The timing of HEV purchase, if it occurs at all, could be influenced by many factors other
than the actor’s motives or interests. At a certain stage in their life, they might not buy an HEV
due to: low disposable income; a mismatch between lifestyle needs, e. g. recreation or a sporty
look, and the available model; or a lack of need for a vehicle in general ( stock turnover). Thus,
given the complex interactions among contextual factors even in this simplistic example, RLI
seeks to avoid the mistake of attributing the timing of new vehicle adoption to actor factors
alone.
Of course, actor characteristics also matter. Under RLI, there are two main antecedents to an
actor’s internal motivations and intention to purchase the artifact: awareness and positive
assessment. First, the actor must have awareness knowledge, i. e. knowledge that the artifact
exists and is available for sale, and have some basic functional understanding of what it does,
e. g. that a PHEV plugs in and/ or can offer increased fuel economy. Such basic, functional
understanding is required before a stable assessment can be formed.
Second, the actor must somehow form a positive assessment of the artifact— perceiving private
and/ or pro- societal interpretations that they feel are, on net, desirable from their own perspective
and fitting in with their self- concept. An assessment is the actor’s overall judgment of the
artifact, divided into two components: private assessment ( how the artifact might benefit the
actor), and societal assessment ( how the artifact might benefit other actors or society). The
actor’s overall assessment may be based only on one component, or some combination of both.
An assessment is informed by one or more interpretation, which is a more specific judgment
relating to a particular functional or societal benefits ( or dis- benefits) of the artifact— such as
perceived fuel savings or reduced greenhouse gas emissions.
At a given point in time, an actor’s interpretations relating to the artifact may be completely non-existent,
flexible, or stable. Interpretations are non- existent if the actor is unaware of the
artifact’s existence. Once aware, the actor typically starts to form some interpretations of what
benefits ( or dis- benefits) the artifact might offer. Interpretations remain flexible when the actor
has little information about and experience with the artifact— they may be unsure of what the
artifact offers, or uncertain if promised benefits will be delivered or what costs may be incurred.
The actor’s interpretations may become stable with increased information, experience or a
convergence of interpretations among relevant social groups. If an actor forms a stable, positive
6
assessment of the artifact, subject to the contextual factors noted above they may form a
purchase intention, and perhaps eventually purchase it.
The primary driver of the actor’s assessment is their self- concept relating to their lifestyle
practices and trajectory. Thus, assessment is not likely to follow the rational, deliberative process
represented by the rational choice model— unless, but perhaps not even if, the actor
conceptualizes himself as a rational, deliberative person. The actor’s self- concept and lifestyle
shape what aspects of the artifact they perceive to be important, thus shaping their interpretations
and assessment of the artifact. Their overall assessment is positive if the artifact and its
interpretations are perceived to align with their self- concept and perceived lifestyle trajectory—
even if the interpretations are based on inaccurate or ill- formed information. At the same time,
interpretation and assessment processes reflexively influence the actor’s self- concept and
lifestyle trajectory. In short, the actor’s self- concept is itself dynamic and open to influence.
To identify contexts in which an actor’s self- concept is more subject to change, RLI employs the
notion of lifestyle liminality ( Turner, 1969). An actor is in a luminal state if their self- concept is
ambiguous, in a state of transition, and/ or open to change. Such an actor is not firmly committed
to a well- defined set of lifestyle practices. Liminality may be associated with numerous factors,
such as: the actor’s life stage; their commitments and responsibilities; their access to resources
such as time and money; the degree of routinization of their lifestyle practices, e. g. work time
and location; and access to a diversity of values and lifestyle practices through interactions with
a diverse social network. An actor’s self concept is relatively stable if they are committed to a
well- defined set of lifestyle practices. Liminality can vary across actors and over time for a
particular actor.
4.4 Driving adoption: Layers of social influence
RLI describes assessment and adoption according to four basic layers. Building up from the
bottom, those layers are functional awareness, artifact assessment, self- concept, and behavior,
e. g. purchase. It is the visible outcome that captures the attention of most behavior researchers,
but a focus only on this “ surface” neglects the important processes that occur below. The
ordering of these layers is depicted according to a tendency observed among primary
participants: the stabilization of lower level ( or foundational) processes tends to be required
before higher layer processes can take place. For example, a stable assessment can only take
place after the actor has first become aware of the artifact and second, translated functional
information to form their private or societal interpretations. The actor starts to reflexively relate
the artifact to their self- concept as they start to form a stable assessment. Finally, purchase
( which was not observable in the present study) tends to occur once the actor somehow
establishes that the artifact fits with their self- concept. Despite this ordering, interactions among
layers can occur in multiple directions ( not just bottom to top or vice versa) and at multiple
points in time.
Processes of social influence tend to follow different patterns at each layer of RLI. Diffusion,
translation and reflexivity offer complementary explanations, and RLI borrows concepts from
each perspective and redefines each process. Figure 1 conceptualizes how social interactions
regarding the different attributes of an innovation may fall onto a continuum representing the
7
complexity of interactions. Some types of information are simple and static and their flow is
better explained as diffusion, such as awareness knowledge. Awareness knowledge flows in one
direction ( the sender does again become aware of the artifact) and is not significantly altered as it
is passed among individuals. Diffusion also applies to simple functional information about the
artifact, such as what it does. RLI defines processes of diffusion as occurring when such simple
information regarding the artifact diffuses uni- directionally within and between social groups;
the information changes little with each communication. Diffusion can also occur at the
assessment layer, such as when the actor simply assimilates interpretations from another actors’
assessment.
Diffusion
Translation
— Who am I? ( self- concept)
— Who does it benefit other than me? ( societal)
— How does it benefit me? ( private)
— What does it do? ( functional info)
— Does it exist? ( awareness)
Types
of
ideas:
Types of
social
influence:
Reflexivity
Simple
interactions
Complex
interactions
FIGURE 1 The continuum of interaction complexity
Figure 2 depicts processes of social influence as they relate to the four layers identified above.
The flow of information relating to the actor’s interpretations and assessment is characterized as
translation— the interaction influences the actor’s assessment of the artifact within the actor’s
unique context of experience and self- concept. The content and presentation of the information
and how it is perceived can change substantially with each communication, and information and
interpretations may be translated in multiple directions during a social interaction. Further,
interactions with relevant social groups can influence a given layer through direct translation, i. e.
the content of the interaction directly addresses the assessment layer, or through indirect
translation, e. g. the interaction content is diffused to the awareness layer, but the actor translates
this functional information into their unique assessment.
Social influence at the self- concept layer constitutes reflexivity— assessment, intention to adopt
and purchase are driven by the actor’s project of the self, and the actor reflexively derives and
develops interpretations of how different practices fit into their self- concept and how the self-concept
itself might change. Direct reflexivity occurs when the content of an interaction
addresses the actor’s self- concept, such as two car buyers talking about whether they are
motivated by financial savings ( private values) or saving the environment ( societal values).
Indirect reflexivity is when an actor’s self- concept is influenced by content addressing other
layers, such as if a car buyers’ friend’s comment that PHEVs are a great way to reduce GHGs
8
( societal assessment) stimulates the car buyer to consider becoming a person that is more
engaged in pro- environmental practices ( self- concept). Reflexivity is by definition dynamic,
multi- directional and iterative, where self- concept may both influence and be influenced by the
actor’s artifact assessment and purchase behavior. For instance, if the actor identifies with an
environmentalist self- concept, they may purchase an artifact they associate with helping the
environment; by purchasing the artifact, they reinforce their self- concept of being an
environmentalist.
Diffusion
Translation
Reflexivity
Who
am I?
Benefits
to me/
others?
Artifact
assessment
Self-concept
Layers: Actor
What is it?
Awareness
information
Adopt or
not.
Behavioral
outcome Processes
of influence:
Social
interactions
Social
interactions
Contextual
factors
FIGURE 2 The basic processes of social influence
The RLI framework accounts for the three empirical discoveries related to a household’s
consideration of pro- societal values within the PHEV demonstration, as identified in Section 3:
1. Basic functional understanding is required before a more sophisticated private and/ or
societal assessment can be made ( thus, awareness is layer 1).
2. Pro- societal support within the actor’s social network can occur through translation at
the assessment and self- concept layers ( layers 2 and 3).
3. An actor with a primarily private self- concept must be in a state of lifestyle liminality
in order to consider developing a pro- societal self- concept ( layer 4).
In RLI depictions below ( Figures 3 and 4), liminality is representeted at a given state of time by
a dotted oval, and stability is represented by a solid oval. Similarly, a dotted oval at the private or
societal assessment layers indicates the assessment is flexible, and stability is represented by a
solid oval.
9
5. APPLYING THE RLI FRAMEWORK
To illustrate the use of the RLI framework in representing processes of social influence and the
development of consumer perceptions and values, this section applies RLI to two of the social
networks observed in the PHEV demonstration project ( Axsen, 2010): the Noels’ as a household
with a stable, private self- concept; and the Forts as a household that shifts their values in favor of
a pro- societal self- concept over the course of their PHEV trial. ( All names are pseudonyms).
5.1 The Noels: Stable, private self- concept
A simplification of Rupert and Amy Noels’ PHEV experience is depicted in Figure 3. In the
beginning of their PHEV trial ( t = 1), the Noels learned about the PHEV’s existence and its basic
functions from researchers. Upon learning that the PHEV could reduce gas use, the Noels’
initially ( indirectly) translated this information into their private assessment of saving money,
which was implicitly linked to their fairly stable self- concept of being a family- oriented
household. By the end of their PHEV trial ( t = 2), the Noels’ private assessment had been
influenced through several social interactions. When Amy Noel’s Uncle stated that the PHEV
could help reduce trips to the gas station, she assimilated his interpretation into her own private
assessment ( a direct translation). Also, by driving their PHEV with their three children as
passengers, the Noels learned that although such a vehicle was fun for the kids, the Prius was too
small for their family— a PHEV for them would need to be larger. These private interpretations
are indirectly reflexively linked with ( and in this case reinforced) the Noels’ self- concept of
being a family- oriented household.
None of the Noels’ friends, family or other alters referred to societal aspects of the PHEV, and
this layer remained irrelevant to the Noels throughout their PHEV trial. From their assessment,
the Noels form an intention to purchase such a vehicle subject to conditions that align with their
family- oriented self- concept. Because this study did not observe PHEV adoption, it is unknown
if the Noels will turn their pro- PHEV intention into purchase at some point in the future.
Ultimately, such a purchase would only occur with an alignment between the conditions of their
behavioral intention ( which may still change over time), and the contextual factors such as model
availability, purchase price, and their household income.
The Noels diffused information about the PHEV to some friends, family and other alters,
including Rupert Noels’ manager, John, who served as secondary participant. From Rupert, John
learned that PHEVs exist and could reduce gas spending, and, contrary to his initial perceptions,
that electric- drive vehicles could have powerful acceleration. John indirectly translated this
information to inform his private assessment of the PHEV, though his already positive societal
assessment of electric- drive remained unchanged. His private and societal assessments are
reflexively linked with his stable self- concept of being a smart consumer, as well as his intention
to purchase a PHEV if it offers fuel economy he deems to be satisfactory.
5.2 The Forts: Opening to a pro- societal self- concept
Figure 4 depicts the Fort’s PHEV experience. After learning about the PHEV from researchers,
the Forts were initially unsure if such a car could save them money, would be perceived by
10
others as a “ weenie car” and was safe enough for them to drive regularly ( flexible private
interpretations). They were also unsure if the PHEV had environmental benefits ( flexible societal
interpretations) that would link to their recent experimentation with environmental lifestyle
practices ( lifestyle liminality), such as reducing home energy and water use. By the end of their
trial, after driving the PHEV as a family and receiving pro- societal support in their social
network, the Forts reduced their private concerns of safety and image and solidified their
commitment to environmental values.
The Forts also diffused functional and assessment information to alters like their friend Lindsey.
Lindsey learned functional information about the PHEV from Selena Fort, such as how it
achieved impressive fuel economy and was easy to recharge. Lindsey linked these benefits to her
stable self- concept of being a family- oriented single parent who was concerned about saving
money and making smart financial decisions. In this social interaction, influence was
unidirectional; Lindsey did not exert social influence on the Forts’ assessment of the PHEV ( as
reported by the Forts).
These examples illuminate the important, though typically indirect, roles that diffusion,
translation and reflexivity can play in PHEV assessment, self- concept, purchase intention and
ultimately, though not observable in this empirical study, purchase behavior. The RLI framework
accounts for heterogeneity among instances of social influence: information and interactions can
range from simple to highly complex, the flow can be one- way or multidirectional, and the
process can directly or indirectly impact awareness, functional understanding, assessment, self-concept,
behavioral intention and purchase behavior. In these examples, direct influence is
observed at layers of awareness, functional understanding, private and societal assessment and
self- concept.
PHEVs
exist;
use less
gas
We value
family
Artifact
assessment
Self-concept
Behavioral
intention
Layers:
The Noels
Private:
saves $
Awareness
information
Want family
size PHEV
We value
family
The Noels
Want
PHEV w/
100 mpg
I am a smart
consumer
John ( Coworker)
Time Period
Participant:
t = 1 t = 2
The Noels’ context:
Income: $ 80- 89k
Purchase: 2- 5 yrs
Home charge: yes
Life: young family
Technology: low
John’s context:
Income: >$ 125k
Purchase: 1- 2 yrs
Home charge: yes
Life: empty nest
Technology: mod
Diffusion
Translation
Reflexivity
Stable
state
Liminal/
flexible
Influ-enced
Social
interaction
Interview
w/ PHEV
researcher
Interview
w/ PHEV
researchers
Talking
with Uncle
Driving the
kids
Private:
saves $;
 gas visits;
size
PHEVs
exist;
use less
gas
Societal:
 pollution
Private:
$; power
PHEVs
exist;
use less
gas
Legend:
FIGURE 3 The Noels’ RLI framework
273
We are smart,
fun, enviro?
Artifact
assessment
Self-concept
Behavioral
intention
Layers:
The Forts
Awareness
information
Want
HEV or PHEV
(+ 45mpg)
We are smart,
fun, enviro!
Want small
SUV PHEV
I am a
family person
Lindsay ( Family friend)
Time Period
Participant:
t = 1 t = 2
The Forts
Societal:
Pro- enviro?
Private:
weenie?
Linday’s context:
Income: $ 60- 69k
Purchase: 2- 5 years
Home recharge: yes
Life stage: single
parent
Technology: low
Fort’s context:
Income: $ 100- 124k
Purchase: not sure
Home recharge: yes
Life stage: family
Technology: High
Interview
w/ PHEV
researchers
Diffusion
Translation
Reflexivity
Stable
state
Liminal/
flexible
Influ-enced
Social
interaction
Legend:
PHEVs
exist;
use less
gas
Interview
w/ PHEV
researchers
PHEVs
exist;
use less
gas
Societal:
Enviro!
Private:
Driving and safe
talking
w/ family
Seeing Prius
drivers
wave
Private:
Saves $;
easy
charge
PHEVs
exist;
use less
gas
FIGURE 4 The Forts’ RLI framework
277
6. POLICY IMPLICATIONS
RLI can be applied beyond the scope of the PHEV demonstration project, such as policy
analysis. The rational choice model of behavior suggests only two levers for policymakers to
influence consumer behavior: changing cost ( via financial incentives or disincentives) and
providing functional information about the product or behavior. In contrast, this study observed
social influence change households’ values and expressions of possible future behaviors. RLI
views the government as an actor, and implemented policies are a form of social interaction
between the government and car buyers. Careful consideration of how different policies and
types of information and experiences influence car buyers can help policymakers to better design
policy, predict its effects, and measure its impacts. In particular, policymakers might consider the
differences between the processes of diffusion, translation and reflexivity.
Several policy examples are depicted in Figure 5. A publicity campaign can attempt to
intentionally diffuse, or disseminate, information about PHEVs in efforts to achieve societal
goals, such as awareness of the technology and functional information about what it does.
Policymakers might also disseminate this information through labeling standards, or energy
information websites. Successful policy- driven diffusion may help to establish the awareness and
functional understanding layers that are necessary for an actor to further assess and value the
technology. However, diffusion alone does not necessarily impact the actor’s assessment, self-concept,
or purchase behavior.
Translation describes a more sophisticated form of social influence where the actor develops a
more refined and stable understanding of the PHEV, how it might benefit them personally, if it
might benefit society, and ( through reflexivity) if they should care if it benefits society. What
policies might be coordinated with a simple publicity campaign to help actors and social groups
translate information? Product labeling serves as one type of translation— where policymakers
frame the PHEV according to particular benefits, such as cost savings ( a private benefit) or GHG
emissions ( a societal benefit). Other policies may also be indirectly ( or unintentionally)
translated by actors. While a subsidy directly affects the price of a PHEV ( a contextual factor in
RLI), it may also help diffuse awareness about the technology, and also may be translated
through considerations of why the subsidy is being offered, e. g. that PHEVs are good for society,
PHEVs are ineffective technologies that need government help, or the government is wasting tax
dollars. Further, a government mandate, such as the Zero- Emissions Vehicle ( ZEV) mandate, can
also reflexively contribute to social discourse about what kind of vehicles consumers should
desire or at least what attributes of vehicles should be pertinent to consumers’ self concepts, e. g.,
whether consumers should value private and societal benefits ( Brown, 2001). In short,
policymakers need to consider the variety of impacts of a given policy, including the differing
processes of social influence, and RLI provides a useful framework to do so.
Change context
Diffusion
Translation
Reflexivity
Artifact
assessment
Self-concept
Layers:
Awareness
information
Behavioral
outcome
Contextual
factors
Legend:
Awareness,
functions
Who am
I?
Private/
societal
benefits?
Car Buyer
Adopt or
not.
 price
 availability
Policy
Subsidy
Mandate
GHG
labeling
Energy
labeling
Publicity
campaign
FIGURE 5 Policy levers for social influence
7. DISCUSSION AND HYPOTHESES
To an extent, the RLI framework accounts for each of the components Jackson ( 2005) highlights
as integral for a useful model, including motivations, value, context and social influence. An
actor’s motivations stem from their self- concept— their desire to construct a meaningful narrative
of their life experiences that represents who they are and where they are going. This self- concept
is reflexively linked to how the actor frames their attitudes towards the purchase and use of a
new artifact, as well as other potential lifestyle practices they may or may not adopt. Included is
how the actor considers and prioritizes private and societal benefits of the artifacts. Social
influence is represented as processes that help actors to learn about the artifact and its functions,
assess how it benefits them and society, test different lifestyle practices, and develop their own
self- concepts. Assessment and intention are implemented into action only when aligned with
contextual factors, such as market availability, pricing and government support, and personal
capabilities, such as household income and access to the artifact. Finally, habit is represented as a
tendency towards the status quo— the actor tends to engage in practices that align with their self-concept.
If self- concept is stable, practices established as well- aligned with this concept may
become relatively automatic. However, when in a state of lifestyle liminality, the actor may
break old habits and consciously experiment with new practices as trials of a new lifestyle.
The primary contribution of the RLI framework is the orientation and structuring of important
concepts from different research perspectives on social influence into a single framework. Here
we have applied RLI to qualitative research in a descriptive manner. RLI could also be used to
guide quantitative research, and could feasibly be helpful in predictive applications. For instance,
the hypotheses articulated below could be tested with statistical models in a large- scale,
representative sample of car buyers. Hypotheses flowing from RLI may include:
1. Actors’ self- concept can change, including their valuation of private versus pro- societal
benefits.
2. Actors with a primarily private self- concept are more likely to value the pro- societal
benefits of a new artifact if the actor:
a. has a basic functional understanding of the artifact,
b. is in a state of lifestyle liminality, and/ or
c. finds support for pro- societal values among their relevant social system.
3. Processes of direct translation are more likely to influence adoption behavior than
processes of diffusion and/ or indirect translation.
4. In a given social interaction, direct translation is more likely to occur for the actor than
diffusion or indirect translation if:
a. the alter is socially close,
b. the alter has previous experience with the artifact,
c. the interaction takes place over multiple social episodes, and/ or
d. information exchange and/ or influence are multidirectional.
5. Direct reflexivity occurs less frequently than other forms of social influence, but when it
does occur, it is more likely to influence the actors’ vehicle assessment and purchase
behavior than other processes of social influence.
RLI is equally valid as a framework to guide the design of qualitative research projects and the
construction of participant narratives. In fact, the authors hope that RLI will inspire
transportation behavior researchers to more actively consider the strengths of qualitative research
approaches and instruments to observe the complexities of consumer behavior.
8. CONCLUSION
To create an integrative behavioral model to represent processes of social influence and vehicle
purchase behavior, this paper draws from a literature review and empirical observations of
PHEV demonstration participants. The result— the reflexive layers of influence ( RLI)
framework— proves useful for describing illustrative examples, and considering the design and
implementation of transportation energy policy relating to consumer behavior. Future research
should further test this model through application to alternative contexts, and can also use the
insights and hypotheses gleaned from this framework to guide both quantitative and qualitative
research design. Policymakers may also want to heed insights that RLI holds for policy design,
particularly to effectively leverage processes of social influence in efforts to align consumer
behavior with societal goals.
ACKNOWLEDGEMENTS
The authors would like to thank the California Air Resources Board, the Social Sciences and
Humanities Research Council of Canada, and the 40 individuals that took part in this study.
REFERENCES
Ajzen, I. and Fishbein, M., 1980. Understanding Attitudes and Predicting Social Behaviour.
Prentice- Hall Inc., Englewood Cliffs, NJ
Axsen, J., 2010. Interpersonal Influence within Car Buyers' Social Networks: Observing
Consumer Assessment of Plug- in Hybrid Electric Vehicles ( PHEVs) and the Spread of Pro-
Societal Values. University of California, Davis, Institute of Transportation Studies, Dissertation.
Axsen, J. and Kurani, K. S., 2010. Interpersonal influence within car buyers' social networks:
Five perspectives on plug- in hybrid vehicle demonstration participants. University of California,
Davis, Institute of Transportation Studies, Working Paper.
Blaut, J. M., 1987. Diffusionism - A uniformitarian critique. Annals of the Association of
American Geographers 77( 1), 30- 47.
Blumer, H., 1969. Symbolic Interactionism -- Perspectives and Method. University of California
Press, Berkeley, CA.
Brown, M. B., 2001. The civic shaping of technology: California's electric vehicle program.
Science Technology & Human Values 26( 1), 56- 81.
Bruun, H. and Hukkinen, A., 2003. Crossing boundaries: An integrative framework for studying
technological change. Social Studies of Science 33( 1), 95- 116.
Callon, M., 1991. Techno- economic networks and irreversibility. In: J. Law ( Eds.), A Sociology
of Monsters: Essays on Power, Technology and Domination, Routledge: London, 132- 161.
Carrasco, J. A., Hogan, B., Wellman, B. and Miller, E. J., 2008. Collecting social network data to
study social activity- travel behavior: an egocentric approach. Environment and Planning B:
Planning and Design 35( 6), 961- 980.
Giddens, A., 1991. Modernity and Self- Identity: Self and Society in the Late Modern Age.
Stanford University Press, Stanford, CA.
Hannemyr, G., 2003. The Internet as hyperbole: A critical examination of adoption rates.
Information Society 19( 2), 111- 121.
Hirschman, E. C., 1980. Innovativeness, novelty seeking, and consumer creativity Journal of
Consumer Research 7( 3), 283- 295.
Jackson, T., 2005. Motivating Sustainable Consumption: A Review of Evidence on Consumer
Behaviour and Behavioural Change. Sustainable Development Research Network, University of
Surrey.
Kline, R. and Pinch, T., 1996. Users as agents of technological change: The social construction
of the automobile in the rural United States. Technology and Culture 37( 4), 763- 795.
Law, J., 1992. Notes on the theory of the actor network: Ordering, strategy, and heterogeneity
Systems Practice 5( 4), 379- 393.
Pentland, B. T. and Feldman, M. S., 2007. Narrative networks: Patterns of technology and
organization. Organization Science 18( 5), 781- 795.
Pinch, T. J. and Bijker, W. E., 1984. The social construction of facts and artifacts - Or how the
sociology of science and the sociology of technology might benefit each other. Social Studies of
Science 14( 3), 399- 441.
Rogers, E., 2003. Diffusion of Innovations ( 5 th Ed.). Free Press, New York.
Stern, P., Dietz, T., Abel, T., Guagnano, G. and Kalof, L., 1999. A value- belief norm theory of
support for social movements: The case of environmental concern. Human Ecology Review 6,
81- 97.
Thaler, R. H. and Sunstein, C. R., 2003. Libertarian Paternalism. The American Economic
Review 93( 2), 175- 179.
Train, K., 1980. The potential market for non- gasoline- powered automobiles. Transportation
Research Part A: Policy and Practice 14A, 405- 414.
Turner, V., 1969. The Ritual Process: Structure and Anti- Structure. Aldine Publishing, Chicago,
Il.
Tversky, A. and Kahneman, D., 1974. Judgement under uncertainty: Heuristics and biases.
Science 185, 1124- 1131.