embracing complexity: a transdisciplinary conceptual...

World Development 126 (2020) 104703

Contents lists available at ScienceDirect

World Development

journal homepage: www.elsevier .com/locate /wor lddev

Embracing complexity: A transdisciplinary conceptual framework forunderstanding behavior change in the context of development-focusedinterventions

https://doi.org/10.1016/j.worlddev.2019.1047030305-750X/� 2019 The Authors. Published by Elsevier Ltd.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.E-mail addresses: [email protected] (F. Lambe), [email protected] (Y. Ran),

[email protected] (M. Jürisoo), [email protected] (S. Holmlid), [email protected] (C. Muhoza), [email protected] (O. Johnson), [email protected] (M. Osborne).

Fiona Lambe a,⇑, Ylva Ran a, Marie Jürisoo a, Stefan Holmlid b, Cassilde Muhoza c, Oliver Johnson a,Matthew Osborne a

a Stockholm Environment Institute, SwedenbDepartment of Information and Computer Sciences, Linköping University, Swedenc Stockholm Environment Institute, Kenya

a r t i c l e i n f o

Article history:Accepted 28 September 2019Available online 15 October 2019

Keywords:Behavior changeDevelopment intervention designService designComplex adaptive systems

a b s t r a c t

Many interventions that aim to improve the livelihoods of vulnerable people in low-income settings failbecause the behavior of the people intended to benefit is not well understood and /or not reflected in thedesign of interventions. Methods for understanding and situating human behavior in the context ofdevelopment interventions tend to emphasize experimental approaches to objectively isolate key driversof behavior. However, such methods often do not account for the importance of contextual factors andthe wider system. In this paper we propose a conceptual framework to support intervention design thatlinks behavioral insights with service design, a branch of the creative field of design. To develop theframework, we use three case studies conducted in Kenya and Zambia focusing on the uptake of newtechnologies and services by individuals and households. We demonstrate how the framework can beuseful for mapping individuals’ experiences of a new technology or service and, based on this, identifykey parameters to support lasting behavior change. The framework reflects how behavior change takesplace in the context of complex social-ecological systems – that change over time, and in which a diverserange of actors operate at different levels – with the aim of supporting the design and delivery of morerobust development-oriented interventions.� 2019 The Authors. Published by Elsevier Ltd. This is anopenaccess article under the CCBY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction and aim tions must deal with inherently ‘‘wicked problems” that are by nat-

The success or failure of interventions that aim to changebehavior hinge on people thinking, deciding and acting in a certainway. Thus, for interventions to work, it is critically important thatthey are designed in accordance with how people actually think,decide and act (Datta & Mullainathan, 2014). This is no less truefor the design of programmes or policies aiming to change behav-ior in low income countries. Behavioral science has providedapproaches and methods for understanding human behavior,many of which have proven useful for the design and delivery ofinterventions aimed at low-income populations.

Given that much development research involves the study ofcomplex, adaptive systems, we assume that development interven-

ure ‘‘difficult or impossible to solve because of incomplete,contradictory, and changing requirements that are often difficultto recognize” (Rittel & Webber, 1973). Wicked problems cannotbe solved in a traditional linear fashion, because the problem def-inition evolves as new possible solutions are considered and/orimplemented (Rittel & Webber, 1973).

There is increasing acceptance that interventions that acknowl-edge individuals’ decision-making processes and the implicit trade-offs required of individuals are likely to be more successful(Banerjee, Duflo, Glennerster, & Kothari, 2010). To date, researchto understand individual behavior in the context of developmentinterventions has tended to focus on the use of experimental meth-ods to identify where behavioral insights can be usefully applied toimprove the effect of an intervention. Behavioral insights have beenparticularly successful for understanding one-off decision-makingat one point in time, e.g. farmers purchasing fertilizer (Duflo,Kremer, & Robinson, 2011) or families deciding to bring their chil-dren to the clinic for vaccination (Banerjee et al., 2010). There is agrowing body of work that focuses on understanding ongoing,

http://crossmark.crossref.org/dialog/?doi=10.1016/j.worlddev.2019.104703&domain=pdfhttp://creativecommons.org/licenses/by-nc-nd/4.0/https://doi.org/10.1016/j.worlddev.2019.104703http://creativecommons.org/licenses/by-nc-nd/4.0/mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://doi.org/10.1016/j.worlddev.2019.104703http://www.sciencedirect.com/science/journal/0305750Xhttp://www.elsevier.com/locate/worlddev

2 F. Lambe et al. /World Development 126 (2020) 104703

repeated behaviors and habit formation, for example, studies havelooked at the effect of incentives on long term habit formationaround hand washing in West Bengal, India (Hussam, Rabbani,Reggiani, & Rigol, 2016), interventions to reduce household waterconsumption in the long term in Costa Rica (Datta et al., 2015)and incentives to reduce daytime drinking amongst informal work-ers in India (Schilbach, 2019). Some studies have focused on shifts ina set of behaviors within a specific environment, for example,energy saving behaviours in the workplace (Klege, Visser, Datta, &Darling, 2018). There are fewer examples of behavioral insightsapplied to understand behavior in complex change processes,where a set of different behaviors need to change within individu-als, orwhere different actors need to shiftmultiple behaviors simul-taneously (e.g. farmers adopting a package of agricultural inputs, orhouseholds adopting clean cookstoves). To effectively apply behav-ioral insights, it is necessary to know precisely where andwhen in aprocess of changing behavior that specific behavioral determinantscome into play, as well as the relative importance of various behav-ioral determinants in the decision-making landscape and the roleand influence of other actors on the behavior change process

Research on resilience and social-ecological systems hasattempted to overcome the challenge of explaining behavior incomplex change processes, in particular the human dimensionsof social-ecological dilemmas (Fabinyi, Evans, & Foale, 2014). How-ever, the research tends to focus on social units, rather than thecomplex interplay between individuals and the social units.Fabinyi et al. (2014) highlight the failure of social-ecological sys-tems research to acknowledge the complexity and social diversityof studied systems.

In this paper we propose a conceptual framework that aims tointegrate insights from behavioural science and complex adaptivesystem dynamics using service design – a qualitative approach tounderstanding people in their wider context, and their needs,motivations and behaviours – with the intention of co-creatingimproved interventions that better meet their needs (Edvardsson,Kristensson, Magnusson, & Sundström, 2012; Patrício, Gustafsson,& Fisk, 2018; Pfannstiel & Rasche, 2017).

Following Imenda (2014) we define ‘‘conceptual framework” asa synthesis of concepts and perspectives drawn from manysources, which provides an integrated way of looking at a problem.The purpose of our conceptual framework is to supportdevelopment-oriented academics, practitioners, and other profes-sionals to understand the behavior change(s) that are required byindividuals, over time, to achieve sustained uptake of a new tech-nology or a change in practice. The framework has been developedand refined through a series of case studies and through consulta-tions with development professionals from a variety of fields dur-ing a four-year research program1. When combined, it is hoped thatthe framework and the supporting empirical material presented herewill demonstrate how integrating insights from service design andbehavioral science, against a backdrop of social-ecological systemstheory, can support more robust intervention design.

Section 2 sets out the theoretical background. Section 3 pro-vides a generic description of the methodological framework. InSection 4 three case studies are used to describe the frameworkand illustrate its iterative development. Finally, we discuss theoverall contribution of the framework and present suggestionsfor its future development and application.

2. Theoretical background

The development interventions in focus operate at the intersec-tion of environment and development, within complex adaptive

1 https://www.sei.org/projects-and-tools/projects/sei-initiative-behaviour-choice/.

systems, and deal with interlinkages between a multitude of actorsand scales. In our conceptual framework, we use social-ecologicalsystems theory as the theoretical backdrop needed for capturingthe multi-level system dynamics influencing individual behaviorand decision making. The logic of the framework is informed byservice design, a user centered approach to understanding complexsystems and by behavioral insights, namely a model of behavioraldesign developed by Datta and Mullainathan (2014) and the Beha-viour Change Wheel (Michie, van Stralen, & West, 2011).

2.1. Social-ecological systems theory

To study interventions that aim to address wicked problems inlow income settings, it is important to consider the complexity ofthe systems under study. Our conceptual framework relies onsocial-ecological systems thinking, which assumes that social andecological dynamics interact as a complex adaptive system (Folkeet al., 2010; Levin et al., 2013) in which the macroscopic propertiesof a system emerge from an interaction among its components, andthe interactions themselves can feed back and impact on subse-quent development. Thus, social-ecological systems theory viewshumans, or actors, as part of the complex adaptive system(Berkes, 2008).

In these types of systems: actors interact, often in unstructuredand unpredictable ways, which leads to the emergence of cross-scale patterns and feedback loops, influencing interactionsbetween actors (Levin et al., 2013). Thus, the components of a sys-tem change as a result of the interplay between the inherentlyadaptive actors and the developing properties of the whole(Lansing, 2003; Levin, 1999). To add complexity, the macroscopicproperties of a system develop from actions at a local scale, in turnfeeding back to influence the behavior, options and choices ofactors, diffusely and, over the long-term (Levin et al., 2013).

Identifying opportunities for creating new feedbacks, orstrengthening desirable feedbacks, requires an understanding ofthe drivers of behavior and decision-making at the local level,and how these relate to the wider social-ecological systems withinwhich households operate. To address these complexities, social-ecological systems analysis focuses on the social group in orderto influence behavior and feedbacks (Fabinyi et al., 2014). Althoughstudies of social-ecological systems and complex adaptive systemsrecognize the importance of studying actors within a system, aswell as the system itself, they have been critiqued for homogeniz-ing social complexity by assuming that people’s interests, expecta-tions and experiences are the same (Fabinyi et al., 2014), and fordownplaying experience-based behavior and the importance ofcultural context and meaning (Cote & Nightingale, 2012).

To illustrate a behavior change process, we draw on conceptsfrom resilience and ecology research (Holling, Schindler, Walker,& Roughgarden, 1995) of an ecosystem shift between ‘stablestates’, driven by a shift in state variables that alters the landscapeand causes the system to move into a new state (Beisner, Haydon,& Cuddington, 2003). Applying this concept to explain a behaviorchange process, the ball (Fig. 1) represents an actor, or an actortype, that, due to changes in state variables, (e.g. behavioral driversand behavior change techniques) transitions from one behavior, toanother, constituting a new stable state.

2.2. Behavioral insights for low-income settings

‘Behavioral insights’ is the collective term for empiricallygrounded knowledge based on cognitive psychology, behavioralsciences and the social sciences about how people behave andmake choices. Behavioral insights are applied to better understand,and predict, human decision-making (Anderson & Stamoulis,2006; Team, 2017). Insights from behavioral research tell us that

https://www.sei.org/projects-and-tools/projects/sei-initiative-behaviour-choice/

Fig. 1. Behavior change understood as a shift from one stable state to another.

F. Lambe et al. /World Development 126 (2020) 104703 3

individuals typically make decisions based only in part on eco-nomic rationales, acting to the best of their knowledge and influ-enced by norms or emotional responses (Kahneman, 2013).Several underpinning principles have been shown to be importantfor explaining decision-making and choice. These include thinkingautomatically (Kahneman, 2013), the use of mental models, andthinking socially (World Bank. (2014), 2014). The application ofbehavioral insights in the field of public policy has gained signifi-cant traction over the past few years, both nationally and in inter-national organizations. A study by Lourenço, Almeida, andTroussard (2016), commissioned by the EU, identified over 200examples in 32 countries of public policy initiatives related tobehavioral perspectives.

There is a growing body of literature on the influence of behav-ioral insights in the context of development interventions in low-income settings, e.g. for programs focusing on agriculture (Dufloet al., 2011; Liu & Huang, 2013; Verschoor, D’Exelle, & Perez-Viana, 2016), improving the quality of education (Benhassine,Devoto, Duflo, Dupas, & Pouliquen, 2015), encouraging individualsaving (Karlan, McConnell, Mullainathan, & Zinman, 2016), provid-ing access to electricity (Lee, Miguel, & Wolfram, 2016) andimproving health outcomes (Hallsworth, Snijders, Burd, Prestt,Judah, & Huf, 2016). Although widening the geographical coverageof studies, they have been criticized for being too narrowly focusedand difficult to generalize beyond specific cases, and thus difficultto scale up (Datta & Mullainathan, 2014; Tantia, 2017).

Datta and Mullainathan (2014) propose an approach for apply-ing behavioral insights to the design of development interventionsand suggest three stages in the process where behavioral insightscan be influential: in defining the problem, in diagnosing the prob-lem, and in designing the intervention. Datta and Mullainathan’smodel of behavioral design has been further developed to includethe stage ‘‘scale” and possible iterative loops between the firstthree stages (Tantia, 2017). See Fig. 2.

The authors of the behavioral design framework recognize theopportunity to closely link behavioral insights and interventiondesign and highlight the need to embed innovation in the processby designing interventions with an iterative experimentation pro-cess (Datta & Mullainathan, 2014). This enables researchers andpractitioners to identify unintended consequences, generate bettersolutions and diagnoses, and develop diagnostic techniques rele-

vant for other contexts. An iterative experimental approachrequires being willing to develop an intervention without neces-sarily isolating the causal effect of a single cognitive process orpathway, but rather focusing on a set of interconnected designinnovations. A key strength of such an approach lies in the thor-ough testing that takes place at each point in the process allowingfor mistakes, for example, misdiagnosed problems to be correctedalong the way.

Behavioral design has been applied extensively to identify andaddress reasons why public programs are not performing asexpected, the most advanced study being the Behavioral Interven-tions to Advance Self Sufficiency (BIAS) project which tested theeffect of behavioral nudges in 15 randomized control trials in eightdifferent locations in the United States (Richburg-Hayes, Anzelone,Dechausay, & Landers, 2017). The study found that the nudgesapplied had a statistically significant impact on at least one pri-mary outcome of interest, leading the authors to conclude thatbehavioural interventions hold promise as a tool for deliveringeffective public programmes (Richburg-Hayes et al., 2017). Thebehavioral design approach used in the BIAS project is similar tothe United Kingdom Medical Research Council’s framework fordeveloping and evaluating complex interventions to tackle healthproblems, (Craig et al., 2008) in that the design and implementa-tion process is iterative.

A number of models have been developed that aim to explainbehavior change related to the longer-term uptake of developmentinterventions. Notable are Mosler’s RANAS framework for behaviorchange in the water and sanitation sector (Mosler, 2012), the Beha-viour Centered Design Framework (Aunger & Curtis, 2016), USAID’sDesigning for Behaviour Change Framework (USAID, 2017a) andMichie et al. (2011) Behaviour Change Wheel, a synthesis of 19frameworks of behaviour change, drawn from a wide range ofdomains. Although each of these frameworks provides a usefulguide for practitioners seeking to develop behavior-based inter-ventions in low income settings, they do not fully address the com-plexity of implementing interventions in terms of how behaviouraldeterminants change over time, the motivations of different typesof users of a service or system or the variety of actors involvedbeyond the individual and household scale, all of which have con-sequences for how sustainable and scalable an intervention is overthe long term.

Fig. 2. The stages of the behavioral design process adapted from (Tantia, 2017).


2.3. Service design

Service design has a legacy in design and service research(Patrício et al., 2018) and emerged around the turn of the millen-nium (Segelström, 2013; Wetter Edman, Göteborgs universitet, &Konstnärliga fakultetskansliet, 2011). As a design practice, servicedesign is a creative, human-centered and iterative approach to ser-vice innovation (Wetter-Edman et al., 2014), gaining ground as asystematic method for creating systems and services that are use-ful, efficient, effective and desirable to the user (Penin, 2017;Stickdorn & Schneider, 2012). Service design is a qualitativeapproach to understanding people’s needs, wider context, motiva-tions and behaviors, which aims to co-create improved services orsystems that better meet their needs (Edvardsson et al., 2012;Manzini, 2015; Pfannstiel & Rasche, 2017). Service design may beseen both as a set of tools and techniques as well as an approachto service innovation and shows promise as a methodology toaddress challenges within the public sector (Malmberg, 2017)and for addressing wicked problems in social systems (Banathy,1996; Jones, 2014).

A central tenet in service design research is the principle of co-creation, where actors in service systems engage in a creative pro-cess to define problems and explore solutions. In recent years, ser-vice design has been increasingly applied in low-income settings toimprove public services to better meet the needs of users and deli-ver positive social impact through so-called design labs, or publicpolicy labs (Bason, 2017; Escobar, 2017).

3. Generic description of the methodological framework

Our conceptual framework draws on the Behaviour ChangeWheel framework, developed by Michie et al. (2011). In this frame-

Fig. 3. Moving from one state of behavior

work, capability, opportunity, and motivation interact to generatebehaviour in a system known as COM-B (Michie et al., 2011). Thesecomponents can be linked to more fine-grained behavior-changetechniques (BCTs), which are active components of an interventiondesigned to change behavior (Michie et al., 2013). Michie et al.have systematically generated and applied collections or ‘‘tax-onomies” of BCTs and from there, developed a ‘‘cross behaviour”taxonomy which includes 93 distinct BCTs (Michie et al., 2015).To enhance usability and accuracy of the taxonomy, the identifiedBCTs were organized into 16 groups (Michie et al., 2015).

Michie et al. (2011) demonstrate the connection between capa-bility, opportunity and motivation, key behavioral mechanismsand BCTs, and the interventions and policies that could be intro-duced to target or change these mechanisms. However, as Michieet al. highlight, ‘‘there is a general recognition that context is keyto the effective design and implementation of interventions, butit remains under-theorized and under-investigated”. And althoughthe Behavioral Change Wheel links behavioral mechanisms todecision-making, it does not account for the need to coordinateand sequence BCTs within a process of developing and implement-ing an intervention. Failure to coordinate BCTs within a changeprocess can result in a ‘‘scattershot” approach to interventiondesign whereby knowledge about behavioral drivers is applied atthe wrong point in the process, where synergies between BCTsare missed or, where BCTs come into conflict with one another.This is shown in Fig. 3 where BCTs are depicted as scattered puzzlepieces.

By using service design, it is possible to identify, at the individ-ual level, behavioral drivers, their underlying mechanisms, BCTsand behavioral archetypes, and to relate these to a specific changeprocess in a coordinated way. This is illustrated in Fig. 4 where thepuzzle pieces, representing BCTs, are joined up. The service design

to another without coordinating BCTs.

Fig. 4. Moving from one state of behavior to another with a coordinated approach.


approach strengthens Datta and Mullainathan’s behavioral designprocess by demonstrating how behavioral change mechanismsand BCTs could be sequenced over time to support sustainedbehavior change in the context of a development intervention.

Our conceptual framework for intervention design (depicted inFig. 5) follows six consecutive stages. Whereas the behaviouraldiagnosis and design model recommends the iteration between

Fig. 5. Conceptual framework for beha

the first three stages (define, diagnose and design) (Barrows,Dabney, Hayes, & Rosenberg, 2018), our framework suggests thatiterations should be made between each stage.

Stage one – Problem co-definition – is a formal and thorougheffort to gather evidence to support the initial underlying assump-tions of an intervention. This is done in close collaboration withkey actors and stakeholders, including the intervention funder(s),

viour-based intervention design.


target beneficiaries and sector experts. This stage can also involve areview of the existing literature on the context of the intervention.

Stage two – experience-based problem diagnosis –verifies theproblem identified in stage one from the perspective of the benefi-ciary (e.g. individual, household, farmer) to identify the underlyingcauses. User journey mapping is conducted and key behavioral dri-vers at the individual level are identified in different phases of thejourney (see supplementary material for a detailed account of userjourney mapping). This stage also merges iterative interventiondesign with the actual journey of the target actors through anintervention, allowing pivotal ‘behavioral moments’ to be identi-fied, highlighting, for example, where an intervention could bederailed due to a disconnect with the users’ needs or motivations,or the emergence of previously undetected opportunities to sup-port the intervention. Unlike behavioral mapping which pinpointsdiscrete behavioral action points in a process and seeks to under-stand ‘‘sub-optimal behavior” (Barrows et al., 2018; Richburg-Hayes et al., 2017), the objective of user-centered mapping is tounderstand the user’s behavior and decision making in relationto the wider system, beyond a given intervention process.

Stage three – System mapping – involves creating a detailed mapof the entire system, including the socio-economic, ecological,structural and institutional dimensions of the setting, using thetarget beneficiary’s experience or ‘journey’ as the starting point.Though similar to process mapping whereby steps in a pro-grammes’ process are analyzed from the perspective of the pro-gramme clients (users) and staff (Richburg-Hayes et al., 2017),actor mapping seeks to widen the scope and includes the roles ofall key actors and processes in the intervention, as well as actorsnot directly engaged in the intervention but who have influenceover or interests in it. The resulting system map from stage twois ‘‘verified” with stakeholders in a workshop setting to ensureaccuracy. See supplementary material for a detailed descriptionof system mapping.

Stage four – rapid prototyping – uses the insights gathered in theprevious three stages to develop rapid prototypes (quick sketches)of an intervention which are then piloted in a subset of the targetpopulation to verify the insights gathered so far and to reducedesign flaws in the intervention. These rapid prototypes areinformed by the BCTs and behavioral mechanisms, in turn basedon the behavioral drivers identified in stage two. Rapid prototypingcan be used in parallel with experience-based problem diagnosis, togenerate rough ideas about the design of the intervention whichcan be quickly tested during interviews with users.

Stage five – design and testing – tests a fully designed interven-tion in a subset of the population and makes changes based on theresults.

Stage six – upscaling – scales up the intervention beyond the ini-tial population, possibly in a new location. This stage entailsreturning to stage 1 and following the cycle again in order to verifyassumptions about a new location and new actors. The second

Table 1Stages of the framework applied in each case study and key contributions of each case st

Case study Stages of the framework app

Kenya and Zambia cookstoves (2015–2016) Stage two experience-based pexperience-based problem diaexperience-based problem diaexperience-based problem diaRapid prototypingexperience-based problem dia

Kenya mango farmers (2017) experience-based problem diaSystem mappingRapid prototyping

Zambia mini/grids (2017) experience-based problem dia

stage should be straightforward because it builds on previouslycollected and verified data. However, care is taken to identify crit-ical elements that might need to be re-designed for the interven-tion to be transferrable to a new location.

Ideally, all the above steps would be followed early in the pro-cess of designing an intervention. However, as illustrated by thethree case studies described below, the conceptual frameworkcan also be applied at any stage in the intervention process to iden-tify and correct flaws in design.

4. Application of the conceptual framework

The conceptual framework was developed iteratively through aseries of case studies on behavior change in relation to the uptakeof new technologies. We present findings from three case studies:uptake of clean cookstoves in peri-urban Nairobi, Kenya, and urbanLusaka, Zambia; uptake of off-grid electricity services by house-holds in Zambia; and uptake of pre- and post-harvest technologiesamong mango farmers in Kenya. The case studies follow the devel-opment of the interventions with a focus on improving an existingintervention, rather than informing the design of a new one. Eachcase study has contributed differently to the development of theconceptual framework, as Table 1 illustrates.

4.1. Identifying user archetypes and opportunities for behavior changetechniques: Case study on adoption of advanced biomass cookstoves inKenya and Zambia

Approximately 80% of households in sub-Saharan Africa do nothave access to clean energy for cooking (International EnergyAgency. (2018), 2018). Although advanced cookstoves have beenpromoted for decades by governments, NGOs and the private sec-tor in different parts of the world, the level of adoption still falls farshort of what is needed to achieve substantial benefits (Barnes,2014). We used case studies from Kenya and Zambia to examinewhat drives households to adopt clean stoves for most or all oftheir cooking needs. The study aimed to better understand the dri-vers of behavior related to adoption of clean cookstoves by house-holds in Kiambu, Kenya and Lusaka, Zambia. In each case westudied cookstove users’ experience of purchasing and using anadvanced biomass cookstove. The main research question askedwhat support is needed, and at which point in the actor journeyis it needed, to achieve lasting behavior change? Results and con-clusions from this case study are further described in Jürisoo,Lambe, and Osborne (2018).

4.1.1. MethodsThe primary methods to collect data were open-ended inter-

views using trigger material, rough pen-and-paper sketches usedto discuss prototypes of possible changes to the intervention, anduser journey mapping. For selecting interviewees, our main crite-

udy to the framework.

lied Key contribution to framework development

roblem diagnosis User journey mappinggnosis Identifying BCTs at different phases in the journeygnosis Sequencing of BCTsgnosis Identifying archetypes

Targeted solutions for identified archetypesgnosis Linking archetypes to BCTs

gnosis User journey mappingSystem map/value chain mapTargeted solutions for specific actors

gnosis User journey mapping, over time


rion was cookstove users who had purchased an advanced cook-stove. In both locations, we made use of existing partnerships tofacilitate interviewee selection and access to households that hadpurchased advanced cookstoves. In Kenya we conducted 19 inter-views and in Lusaka we conducted 17.

We gathered data to map a composite user journey, broken intophases of ‘‘before”, ‘‘during” and ‘‘after” using the stove. The ‘‘be-fore” phase refers to the stages of hearing about the stove anddeciding to purchase it; the ‘‘during” phase refers to the periodof starting to use the stove and establishing a new cooking prac-tice; the ‘‘after” phase refers to the period when the user startslooking for a new technology to replace and/or complement thestove. The user journey phases were developed during the Kenyanstudy and further tested and validated in the Zambian case.

4.1.2. Key results and discussionThe composite user journey is presented in Fig. 6. The red dots

illustrate key components of the behavioral change process understudy: becoming aware of advanced stoves, buying a stove, andmaking it the household’s main or only cooking device. The bluedots represent points during the user journey where, if conditionsare unfavorable, the opportunity to induce a change in cookingpractices can be lost. These can also be viewed as points in thedecision-making landscape where a specific type of support is vitalfor achieving a long-lasting change in behavior.

A key finding of the case studies was that for the cookstoveinterventions to be successful (i.e. advanced cookstove s areadopted by households), BCTs, need to be identified and carefullysequenced throughout the user journey. There is a tendency forcookstove programs to focus their efforts on the ‘‘before” phaseof the user journey, on the provision of marketing and technicalinformation to support potential users to acquire a new cookstove.We found, in both studies, that acquiring the new stove is only thefirst step. What the user experiences when getting started with thenew technology, in the ‘‘during” phase, and how well the technol-ogy meets their expectations, is critical. See supplementary mate-rial for a table summarizing the BCTs identified in the ‘‘during” and‘‘after” phases of the user journey.

Fig. 6. Composite actor journey map for impro

Thus, the identified BCT grouping ‘continuous social support’, tohelp users overcome technical problems, remind householdsabout how to use and maintain the stove, and build confidenceusing the new technology, is imperative at the start of the ‘‘dur-ing” phase. However, this BCT grouping will be less useful if it isnot available to the actors early on; they may already have expe-rienced disappointment that the intervention is not fulfilling theirexpectations.

In addition, the study identified that the BCT groupings ‘changein physical environment’ (i.e. increasing fuel availability) ‘rewardand threat’ (i.e. financial incentives) and ‘shaping knowledge’ (i.e.information access) are also crucial at different stages along thejourney. For more detail see supplementary material and (Jürisooet al., 2018).

Beyond sequencing of BCTs, the user journey mapping alsohelped in identifying three distinct types of cookstove user, looselydefined by their main motivation for purchasing the stove. We alsoobserved that each type requires specific support at differentpoints in the adoption process. Those who were motivated to pur-chase a stove by saving money tend to take several weeks beforethe value of the stove is realized. These users seem aware of thefact that change takes time and are willing to continue to use thestove even where problems were encountered early on. In termsof support, these users need accurate technical information onstove use, how to optimize use so as not to waste fuel, and howto avoid accidents with the stove.

The user group motivated by convenience needs a relativelyimmediate improvement for the value of the stove to be realized,otherwise they tend to become disillusioned. This group requirescontinuous support from the start, ideally from a trusted source.Users attracted to the aesthetic appeal of the stove reported pur-chasing the stove to increase their social status or to be perceivedas modern and aspirational. We found that for this group of usersthe ‘‘newness” tended to decrease over time and the immediaterewards in terms of less smoke and fuel saved do not necessarilymotivate long-term use. Compared to other types of users, thisgroup did not appear to need as much support in the early phaseas those motivated by convenience or by saving money.

ved cookstove users in Kenya and Zambia.


4.2. User mapping to understand the wider context: case study onsolar PV mini-grids for household electricity provision in rural Zambia

More than half of the population of sub-Saharan Africa – 590million people – do not have access to electricity (InternationalEnergy Agency, 2018). Renewable energy mini-grids are expectedto play a major role in the pursuit of universal access to modernenergy services, particularly in areas where grid extension is tech-nically or financially unviable (IRENA, 2013; Szabó, Bódis, Huld, &Moner-Girona, 2011). Out of the roughly 315 million rural Africansthat the IEA envisions will gain electricity access by 2040, about45% would be served by mini-grids (International Energy Agency,2014). However, little is known about the socioeconomic determi-nants in Africa of uptake of electricity from renewable energymini-grid systems. This case study explored the behavioral andsocio-cultural factors that support and constrain the adoption ofelectricity services provided by a solar mini-grid project in ruralZambia.

The case study location is Mpanta, Zambia. Mpanta is a ruralcommunity of 2673 people, situated on the shores of Lake Bang-weulu in Luapula Province in northern Zambia (RuralElectrification Authority. (2016), 2016). In November 2013, a60 kW solar mini grid in Mpanta was commissioned by the RuralElectrification Authority to provide essential electricity servicesfor lighting (including street lighting) and light load appliances(such as televisions, radios, fridges and mobile phones) to 450users comprising of households, a school and staff houses, a ruralhealth center, harbor facilities, small businesses and churches.

After commissioning of the mini-grid in 2013, users were ini-tially connected for free. Each user was required to pay a monthlyfixed tariff based upon their user category (i.e. residential, com-mercial and social services) and, if a residential user, the numberof rooms in their house. Following this initial free connection per-iod, a 50 ZMW (5 USD) connection fee and 15 ZMW (1.5 USD) wir-ing fee were introduced. Meanwhile, to encourage communityparticipation and ownership of the project, mini-grid operation,plant maintenance and revenue collection was handed over to alocal Multi-Purpose Cooperative Society (hereafter KafitaCooperative).

4.2.1. MethodsUser journey mapping was used to map and explore users’

needs, expectations and experiences ‘‘before”, ‘‘during” and ‘‘after”connecting to the solar mini-grid. The user journey mapping wasconducted based on 28 semi-structured interviews with usersand non-users of the solar mini-grid services. Intervieweesincluded 21 households (12 still connected, 4 disconnected and 5never connected), 5 businesses (4 still connected, 1 disconnected)and 2 institutions (both still connected). Data saturation wasreached after 28 interviews, thus determining the sample size.

For data analysis, user and non-user interview responses werecoded in a spreadsheet based on user category (households, busi-nesses and institutions) and connection status (connected, discon-nected and not connected). Responses were then analyzed togenerate insights on the varied emotional and physical experiencesassociated with connecting to the mini-grid and using its servicesand the different contextual factors shaping adoption or non-adoption of mini-grid electricity services. For more detailed resultsand conclusions see (Muhoza & Johnson, 2018).

4.2.2. Key results and discussionKey results from the Zambia case relate to the importance of

embedding local context and the needs and motivations of theusers of services in the design of an intervention. Mapping theexperiences of individual users over time highlighted inconsisten-

cies in the delivery of the intervention which can result in disap-pointment and reduced overall effect of the scheme.

The user journey map in Fig. 7 visualizes the experience of usersand non-users before, during and after connecting to the Mpantasolar mini-grid. These three stages in the user journey correspondto: becoming aware of the mini-grid service, getting connected toit, and continued or discontinued use of electricity. Fig. 5 highlightstouch points associated with the various phases in the userjourney.

The case study found that information about the interventionwas not coherently provided to all users, in the same way. Forexample, some were informed that the electricity provided wouldbe free of charge while others were aware of the actual costs. Somewere told that they would be able to use the electricity for cookingand other uses that would require heavy loads, even though thesystemwas not designed for heavy loads. Others had a clear under-standing of the capacity of the system.

The experience of becoming connected to the mini-grid alsovaried greatly depending on when households joined the scheme.Those who participated in the scheme early on, and benefited fromfree connections, were first visited by an agent from the Rural Elec-trification Agency who brought the application form to their homefor completion, and then by an engineer who would install the nec-essary wiring to connect the household to the distribution networkand provide free lightbulbs. Those who joined the scheme later hadto visit the Kafita Cooperative offices in person and apply and paythe connection fees, which were often prohibitive for low-incomehouseholds.

Many users (47%) were disconnected because they defaulted ontheir payments. The user journey mapping provided insightfulinformation about why so many were disconnected. The commu-nity relies on small-scale fishing as the main source of income,but during December to March every year a ban is imposed toallow fish stocks to replenish. During this time, household incomestend to decrease, leaving many users unable to afford the fixed tar-iff. Thus, if a similar mapping had been conducted while designingthe business model for the mini-grid, economic incentives would,preferably, have been directed to cover costs during this period,and attract and retain users.

4.3. Situating individual behavior within the wider system: case studyon technologies for reducing post-harvest losses in small scale mangoharvesting in Kenya

Agriculture is the most important provider of livelihoods inKenya, with more than 75% of the population depending on thesector for food and income (USAID, 2017b). Mango is an importantfood and cash crop, with a six-fold production increase between2000 and 2014. However, more than 25% of the crop is currentlylost during and after harvesting due to pests, inadequate on-farmstorage and a lack of direct access to markets among small-scalefarmers to sell their produce (Financial Sector Deepening (FSD),2015). To reduce losses, development interventions have beenintroduced to small-scale farmers to reduce losses and improveincomes by producing higher-quality fruits, to process the man-goes (into more durable and/or higher-value products), and toimprove fruit storage. However, uptake of technologies is generallylow, particularly among more marginalized groups, includingfemale-led households.

The study focused on two sites: Tana River County in easternKenya, and Meru County in the center of the country. With 76.9%of the population living below the poverty line, Tana River Countyis among the poorest in Kenya (CRA, 2011). Approximately 40% ofhouseholds in Tana River County are engaged in small-scale farm-ing (MOPHS & IMC, 2010). The poverty rate in Meru County is28.3% which is well below the national poverty rate of 47.2%

Fig. 7. Actor journey map for users and non-users of electricity from Mpanta mini grid.


(CRA, 2011). High-input, rain-fed agriculture complemented byirrigation is the main source of livelihood in the county, contribut-ing about 80% to the average household income (MoALF, 2016).The study intended to investigate how to improve the develop-ment and implementation of technologies, aiming to reduce lossesamong smallholder mango farmers.

4.3.1. MethodsIn this case study, the intervention aimed to help small-scale

farmers to reduce loss of their mangoes by marketing improvedharvest and post-harvest technologies. The primary actors wereassumed to be smallholder farmers, and technologies were alreadydeveloped and introduced to farmers at different locations in inKenya. However, we applied user journey mapping to help identifynot only what could be improved in terms of technology design,but also the underlying factors behind the low uptake of technolo-gies. In addition, we conducted participatory observations, open-ended interviews with a range of stakeholders and two field work-shops, in total 206 interactions with stakeholders in the two loca-tions. Periodically, we presented our evolving analysis to farmersand other stakeholders for their feedback. We also mapped themango value chain within which the new technologies and ser-vices would be provided, to understand the roles of, and relation-ships between, different actors and the links between them atmultiple scales. These insights were consolidated in a systemmap and a corresponding narrative for mango farmers in Holaand Meru.

4.3.2. Key results and discussionThe case study demonstrated the need to account for the inter-

ests and incentives of a wide system of actors when designing anintervention. In terms of reducing losses in Kenyan mango produc-

tion, we found that the underlying problem that the interventionsought to address had been misdiagnosed, and as a consequence,interventions were not developed and introduced to the rightactors. As illustrated in Fig. 6, the user journey of a farmer couldfollow a number of different scenarios: the farmer could indeedbe harvesting the fruits, and thus be the target beneficiary of inter-ventions aiming to reduce pre-harvest and post-harvest losses.However, many farmers did not harvest the fruits but either hiredharvesters, or sold fruits to a broker, who used their own har-vesters. In these cases, the technologies were introduced to thewrong group of actors. The systems mapping identified an arrayof actors, interacting in various ways. Besides farmers and end-buyers (such as retailers or mango-processing companies) therewere three other key actors (or key roles) in the value chain: har-vesters, brokers and farmer organizations (see Fig. 8 below). With-out mapping the entire system, the importance of these otheractors and their connection to the farmers would not have beenidentified.

The pre-harvest and post-harvest interventions had been devel-oped with the objective of enhancing the quality of the produceand thus improving farmers’ incomes. However, the study foundthat farmers did not have access to the market, or buyers, thatwould give them increased return for better quality fruits. Further-more, since harvesting was carried out by hired labour, reduceddamage to fruit during the harvest was out of the farmers’ control,which meant that technologies to reduce such losses were of littleuse to the farmers.

5. General discussion

In this section we discuss the conceptual framework consider-ing the case study findings, with a focus on overarching parameters

Fig. 8. Actor scenarios along the mango value chain.


that should be considered when designing robust and imple-mentable interventions in a low-income context.

5.1. Behaviour change occurs (and is reinforced) within separate butinterconnected phases of the actor journey

Depending on the type of behavior change required, usersengage with varying intensity in different parts of the user journey.For adoption of an intervention, such as a cookstove, a significantchange in behavior is required in the sense that it involves activeengagement in several of the phases of the user journey. The casestudy on adoption of improved cookstoves in Kenya and Zambiademonstrates that developing a new habit with a new technologyrequires a user to stay motivated over a long period of time. Inaddition, users need to learn how to use a new technology earlyin the process, which requires changing several behaviors at once,e.g. cooking food more quickly, not leaving the stove unattended ornot using the traditional stove, and maintaining the behaviorchange until new habits are formed.

With high-effort behavior change, such as technology adoption,supportive actions may be needed over a long period, as the changein behavior is not immediate, nor a one-time action. Breakingdown the experience of adopting a new technology into con-stituent phases provides a simple way to identify when, in thejourney, the user needs support to develop a new habit and whattype of support is needed.

As proposed in this study, key behavioral drivers, BCTs andbehavioral mechanisms must be identified and sequenced intothe decision-making landscape, for interventions to be successfuland scalable. Based on our case study findings, we suggest thatdesigning interventions requires active engagement on the partof the implementer or service provider to iteratively develop theservice or intervention in collaboration with the users, particularlywhen the intervention in question seeks to introduce a new tech-nology or displace an old one.

In the cookstove and mini-grid cases, we found that the extentto which user expectations set in the ‘‘before” phase of an interven-tion are met in the ‘‘during” phase is central to success. In the cook-stove case, individuals and households were provided withinformation about the functioning of advanced cookstoves byway of marketing and promotion – often related to the high effi-ciency of the stoves and the potential to save significantly on fuelpurchase. Indeed, the potential to save fuel was the most com-monly cited reason for purchasing an advanced cookstove. Where

users encountered problems getting started with the stoves, andfuel savings were not quickly realized, the result was often disap-pointment and, in some cases, reduced or discontinued use of thenew stove.

In the mini-grid case there is a clear pattern of unmet expecta-tions among service users. Following connection to the mini-grid,households reported being disappointed that they were unable touse the electricity for cooking or productive uses and for someusers the connection fee was higher than expected.

Although awareness raising and promotion of new services andtechnologies is necessary, implementers and service providersneed to strike a careful balance between communicating the ben-efits of the new service or technology and ensuring a clear under-standing on the part of their users of the costs and limitations. Thecareful management of expectations in the ‘‘before” phase, basedon a clear understanding of the factors motivating users to adopta new technology or service, is a prerequisite for adoption and sus-tained use of the technologies and services in the ‘‘during” phase.

In the case of Kenyan mango farmers, the technologies them-selves were not necessarily malfunctioning; the problem was thatthey did not target the challenges or bottlenecks that users faced,nor were the behavioral components that needed to changesequenced in the decision-making landscape of the system. Thestudy showed that to achieve the objectives of the intervention,farmers must be provided with the appropriate incentives. Farmersneed assurance that adopting a technology to improve the qualityof their produce would generate a higher income, otherwise thetraditional harvesting methods, which are less costly both in termsof time and capital, will remain more attractive.

5.2. Identifying where BCTs or groups of BCTs could be applied

The Kenya and Zambia cases identified behavioral drivers ofimproved cookstove uptake, and the BCTs that could support theprocess of behavior change. In a recent review, the BCT ‘‘shapingknowledge” was identified as the active ingredient in 85% of cook-stove interventions globally and ‘‘social support” in 64% (Goodwinet al., 2015). Thesefindings suggest thatdesigners of improvedcook-stove interventions have confidence in the effect of these BCTs. TheBCTs ‘‘Shaping knowledge” and ‘‘social support” are often appliedtogetherwith ‘‘reward and threat” (most often in the formof a finan-cial incentive) to encourage uptake of improved cookstoves.

However, our conceptual framework extends the work ofGoodwin et al. (2015) by not only identifying key BCTs but also


illustratingwhere in the behavior change process they are most rel-evant. For example, in the case of cookstove adoption, our studiesidentify that ‘‘shaping knowledge” (information and demonstra-tion), ‘‘reward and threat” (in the form of financial incentives)and the ‘‘impact of peers” (friends and social groups) are importantBCTs in the ‘‘before” stage of cookstove adoption, while ‘‘social sup-port” (user support to overcome problems using the stove) and ‘‘re-ward and threat” (ongoing financial incentives, e.g. subsidized fuel)are more important in the ‘‘during” phase.

5.3. The importance of user archetypes

The Kenya and Zambia case studies demonstrate that the con-ceptual framework can be useful for identifying categories of users,and the type of support that they require in different stages inintervention. For example, we identified three main cookstove userarchetypes, defined by their key motivation for purchasing andusing an improved cookstove: those who were motivated by savingmoney, those who sought convenience and those who appreciatedthe aesthetic appeal of the new cookstove. The user journeys foreach archetype revealed positive and negative experiences in dif-ferent phases of the actor journey and, thus, very different needsin their journey toward adopting the cookstoves.

The mango case study demonstrates the importance of under-standing the roles, motivations and incentives of all key actors inthe system when designing an intervention aimed at improvingsystem outcomes for individual actors. It also highlights the needto work closely with the target users of a technology in the earlystages of intervention design, both in co-identifying the problemand co-designing the intervention.

5.4. Connecting interventions to the wider social-ecological context

Users in our case studies are not behaving and making decisionsin isolation; rather they are embedded in multi-level systems withother actors and ongoing processes. All three case studies highlightthe importance of situating an intervention or change processwithin a broader societal context or social system. The mini gridscase study in Zambia makes shows that it is important to acknowl-edge that development interventions operate within a social-ecological system. Despite economic incentives (e.g. low fixed tar-iffs) users were unable to afford the cost of the electricity service.The business model had failed to recognize that the interventionwas introduced in a community highly dependent on seasonalincomes. This could easily have been acknowledged and designedfor, if actors had been consulted early in the design process andimplementation plan. In order to improve actors’ ability to payon a regular basis, there is a need to diversify actors’ sources ofincome. In the mango case study, interventions were introducedthat sought to change behavior and decisions in relation to agricul-tural management. However, the targeted value-chain was notthoroughly mapped to identify interlinked actor scenarios. Pre-and post-harvest technologies were introduced to farmers who inmany cases were not involved in pre and post-harvesting of man-goes. In addition, farmers were assumed to be primarily mangofarmers, yet mangoes were seldom their primary source of income.If the development and delivery of technologies had been an iter-ative and collaborative process involving all relevant stakeholders,it is likely that relevant actor groups and the most appropriateBCTs for targeting them would have been identified early on.

5.5. Limitations and future research

The conceptual framework has been applied in only three cases,all located in either Kenya or Zambia, which may limit the gener-alizability of the findings to other contexts. In addition, the frame-

work has so far only been tested in cases focusing on the uptake ofnew technologies, and not yet applied in cases where the focus ischange of practice. However, the case studies focused on technolo-gies that are relevant far beyond the geographic scope of the casestudies, and on communities of smallholder farmers and house-holds that share characteristics with millions of people in sub-Saharan Africa and South and Southeast Asia. Furthermore, for allcase studies described here, the ‘‘real world” interventions in focuswere already in the early implementation phase at the time weconducted fieldwork, which meant that we did not have the oppor-tunity to study other phases, for example ‘‘upscaling”. Thus, ourinsights about the usefulness of the conceptual framework arebased on the study of a limited number of stages of the interven-tion development process and the approach would benefit fromfurther development and testing.

There is a growing number of studies that apply behaviouraldiagnosis and design approaches to intervention design, and somestudies have conducted rigorous testing of the behavioural designapproach (e.g. (Richburg-Hayes et al., 2017). In addition, there arestudies looking at sustained change in behaviour, for example(Allcott & Rogers, 2014; Ashraf, Bandiera, & Jack, 2014; Hussamet al., 2016). Our proposed framework seeks to contribute to thefield by addressing behavior change in complex systems wheremultiple behaviors need to change at different points in time dur-ing an intervention process, and where actions may be needed by arange of actors at different parts of the system.

6. Conclusions

Behavioral science-based approaches to designing and testingdevelopment interventions have come a long way in terms of iden-tifying key cognitive processes and behavioral levers to triggerbehavior change in low income settings. However, theseapproaches do not fully account for the complexity and interde-pendence within social-ecological systems, or for the fact thatchange processes, once triggered, play out over time and are expe-rienced differently by different people or archetypes in a system.

The conceptual framework proposed in this paper seeks tomerge the methodological approach of service design with behav-ioral insights to better address complexity in social-ecological sys-tems. Service design offers both a process for carefully findingsolutions, and a methodology for basing such solutions on a knowl-edge base that is as widely and inclusively informed as possible.The user journey component of the framework allows us to visual-ize the experiences and perceptions of users of a given technologyor service throughout a change process and ensures that importantbehavioral drivers and social processes are captured at every phaseof the journey. The systems mapping component situates the livedexperiences of users within complex social-ecological systems andhighlights connections between users and other potentially impor-tant actors and processes at different levels of society.

In the case studies presented here we examined the factorsaffecting a sustained shift in behaviors over time. These are situa-tions where a change in behavior on the part of individuals andhouseholds is required every day, and where behavior may beinfluenced by multiple factors operating at different levels – cogni-tive, psychological, social and structural – and where feedbackloops may occur. We show how the proposed framework can beused to pinpoint when in a in a temporal continuum a behaviorchange technique or group of behavior change techniques andbehavioral determinant is relevant, at what point in the changeprocess they matter, and based on this how those that steer inter-ventions can intervene to support lasting behavior change. As such,the framework could help development practitioners and donorsto plan and allocate constrained funding to focus on phase process


that is likely to need more attention and resources. Our aimbeyond this article is to is to apply the framework in additional‘‘real world” case studies, cases that focus on changing practicesand to update and refine the framework accordingly. We are seek-ing opportunities for applying the framework in all phases of inter-vention design, from Problem co-definition to scaling and replicating.

Declaration of Competing Interest

This work was supported by the Swedish International Develop-ment Agency (Sida), Stockholm, Sweden and the Rockefeller Foun-dation, New York, U.S.A. There is no specific grant numberassociated with the funding. Neither funding body was involvedin the study design, data collection, analysis and interpretation,report writing or decision to submit the article for publication.

Acknowledgements

We wish to thank all the individuals and households in Kenyaand Zambia who took time to speak with us and share their experi-ences as service and technology users in the three case studies. Weare also grateful to our partners at SNV, the Netherlands Develop-ment Organisation; Emerging Cooking Solutions and Vitalite fortheir invaluable input on the field work design and emerging con-ceptual model. Finally, we wish to thank our service design col-leagues at Expedition Mondial, Sweden for their supportconducting the case studies, and their thoughtful input and creativeideas for integrating service design into the conceptual model.

Appendix A. Supplementary data

Supplementary data to this article can be found online athttps://doi.org/10.1016/j.worlddev.2019.104703.

References

Allcott, H., & Rogers, T. (2014). The short-run and long-run effects of behavioralinterventions: experimental evidence from energy conservation. AmericanEconomic Review, 104(10), 3003–3037. https://doi.org/10.1257/aer.104.10.3003.

Anderson, C. L., & Stamoulis, K. (2006). Applying behavioural economics tointernational development policy. Helsinki: United Nations University. Worldinstitute for development economics research (UNU-WIDER).

Ashraf, N., Bandiera, O., & Jack, B. K. (2014). No margin, no mission? A fieldexperiment on incentives for public service delivery. Journal of Public Economics,120, 1–17. https://doi.org/10.1016/j.jpubeco.2014.06.014.

Aunger, R., & Curtis, V. (2016). Behaviour Centred Design: Towards an appliedscience of behaviour change. Health Psychology Review, 10(4), 425–446. https://doi.org/10.1080/17437199.2016.1219673.

Banathy, B. H. (1996). Designing social systems in a changing world. Retrieved fromhttp://link.springer.com/openurl?genre=book&isbn=978-1-4757-9983-5.

Banerjee, A. V., Duflo, E., Glennerster, R., & Kothari, D. (2010). Improvingimmunisation coverage in rural India: clustered randomised controlledevaluation of immunisation campaigns with and without incentives c2220–c2220. BMJ, 340(may17 1).

Barnes, B. R. (2014). Behavioural change, indoor air pollution and child respiratoryhealth in developing countries: A review. Int J Environ Res Public Health, 11(5),4607–4618.

Barrows, A., Dabney, N., Hayes, J., & Rosenberg, R. (2018). Behavioral design teams. Amodel for integrating behavioral design in city government. Ideas.

Bason, C. (2017). Leading public design: Discovering human-centred governance.Bristol Chicago, IL: Policy Press.

Beisner, B., Haydon, D., & Cuddington, K. (2003). Alternative stable states in ecology.Frontiers in Ecology and the Environment, 1(7), 376–382. https://doi.org/10.1890/1540-9295(2003) 001[0376:ASSIE]2.0.CO;2.

Benhassine, N., Devoto, F., Duflo, E., Dupas, P., & Pouliquen, V. (2015). Turning ashove into a nudge? A ‘‘Labeled Cash Transfer” for education. American EconomicJournal: Economic Policy, 7(3), 86–125. https://doi.org/10.1257/pol.20130225.

Berkes, F. (Ed.). (2008). Navigating social-ecological systems: building resilience forcomplexity and change (Digitally print. version). Cambridge: Cambridge Univ.Press.

Cote,M., &Nightingale, A. J. (2012). Resilience thinkingmeets social theory: Situatingsocial change in socio-ecological systems (SES) research. Progress in HumanGeography, 36(4), 475–489. https://doi.org/10.1177/0309132511425708.

CRA (2011). Kenya county factsheets. Retrieved from Commission on RevenueAllocation website: https://child.org/sites/default/files/Kenya-County-Factsheet.pdf.

Craig, P., Dieppe, P., Macintyre, S., Michie, S., Nazareth, I., & Petticrew, M. (2008).Developing and evaluating complex interventions: the new Medical ResearchCouncil guidance. BMJ, a1655.

Datta, S., Miranda, J. J., Zoratto, L., Calvo-Gonzalez, O., Darling, M., & Lorenzana, K.(2015). A behavioural approach to water conservation. Evidence from CostaRica. World Bank Policy Research Working Paper, 1 (WPS7283). Retrieved fromhttp://documents.worldbank.org/curated/en/809801468001190306/pdf/WPS7283.pdf.

Datta, S., & Mullainathan, S. (2014). Behavioral design: A new approach todevelopment policy. Review of Income and Wealth, 60(1), 7–35. https://doi.org/10.1111/roiw.12093.

Duflo, E., Kremer, M., & Robinson, J. (2011). Nudging farmers to use fertilizer:Theory and experimental evidence from Kenya. American Economic Review, 101(6), 2350–2390. https://doi.org/10.1257/aer.101.6.2350.

Edvardsson, B., Kristensson, P., Magnusson, P., & Sundström, E. (2012). Customerintegration within service development—A review of methods and an analysisof insitu and exsitu contributions. Technovation, 32(7–8), 419–429. https://doi.org/10.1016/j.technovation.2011.04.006.

Escobar, A. (2017). Response: Design for/by [and from] the ‘global South’. DesignPhilosophy Papers, 15(1), 39–49. https://doi.org/10.1080/14487136.2017.1301016.

Fabinyi, M., Evans, L., & Foale, S. J. (2014). Social-ecological systems, social diversity,and power: insights from anthropology and political ecology. Ecology andSociety, 19(4).

Financial Sector Deepening (FSD). (2015). Opportunities for financing the mangovalue chain: A case study of Lower Eastern Kenya. Retrieved from FinancialSector Deepening (FSD) website: http://s3-eu-central-1.amazonaws.com/fsd-circle/wp-content/uploads/2015/08/30093918/15-06-29-Mango-value-chain-report.pdf.

Folke, C., Carpenter, S. R., Walker, B., Scheffer, M., Chapin, P., & Rockström, J. (2010).Resilience thinking: Integrating resilience, adaptability and transformabilityRetrieved from. Ecology and Society, 15(4).

Goodwin, N. J., O’Farrell, S. E., Jagoe, K., Rouse, J., Roma, E., Biran, A., & Finkelstein, E.A. (2015). Use of behavior change techniques in clean cooking interventions: Areview of the evidence and scorecard of effectiveness. Journal of HealthCommunication, 20(suppl 1), 43–54. https://doi.org/10.1080/10810730.2014.1002958.

Hallsworth, M., Snijders, V., Burd, H., Prestt, J., Judah, G., Huf, S., et al. (2016).Applying behavioural insights: simple ways to improve health outcomes,Report of the WISH Behavioral Insights Forum 2016. Retrieved from WISH,Qatar Foundation and The Behavioural Insights Team website: http://38r8om2xjhhl25mw24492dir.wpengine.netdna-cdn.com/wp-content/uploads/2016/11/WISH-2016_Behavioral_Insights_Report.pdf.

Holling, C. S., Schindler, D. W., Walker, B. W., & Roughgarden, J. (1995). Biodiversityin the functioning of ecosystems: an ecological synthesis. In C. Perrings, K.-G.Maler, C. Folke, C. S. Holling, & B.-O. Jansson (Eds.), Biodiversity loss (pp. 44–83).https://doi.org/10.1017/CBO9781139174329.005.

Hussam, R., Rabbani, A., Reggiani, G., & Rigol, N. (2016). Habit Formation andRational Addiction: A Field Experiment in Handwashing. Harvard UniversityWorking Paper, (18–030). Retrieved from http://www.hbs.edu/faculty/Publication%20Files/18-030_63e232aa-d361-4673-aa2a-9f8dab3dccc2.pdf.

Imenda, S. (2014). Is there a conceptual difference between theoretical andconceptual frameworks? Journal of Social Sciences, 38(2), 185–195. https://doi.org/10.1080/09718923.2014.11893249.

International Energy Agency. (2014). Africa energy outlook: A focus on energyprospects in sub-Saharan Africa. Retrieved from International Energy Agencywebsite: https://www.iea.org/publications/freepublications/publication/WEO2014_AfricaEnergyOutlook.pdf.

International Energy Agency. (2018). World Energy Outlook 2018. https://doi.org/10.1787/weo-2018-en.

IRENA. (2013). Africa’s renewable future: The path to sustainable growth. Retrievedfrom International Renewable Energy Agency website: http://www.irena.org/DocumentDownloads/Publications/Africa_renewable_future.pdf.

Jones, P. H. (2014). Systemic design principles for complex social systems. In G. S.Metcalf (Ed.). Social systems and design (Vol. 1, pp. 91–128). . https://doi.org/10.1007/978-4-431-54478-4_4.

Jürisoo, M., Lambe, F., & Osborne, M. (2018). Beyond buying: The application ofservice design methodology to understand adoption of clean cookstoves inKenya and Zambia. Energy Research & Social Science, 39, 164–176. https://doi.org/10.1016/j.erss.2017.11.023.

Kahneman, D. (2013). Thinking, fast and slow (1st pbk. Ed.). New York: Farrar, Strausand Giroux.

Karlan, D., McConnell, M., Mullainathan, S., & Zinman, J. (2016). Getting to the top ofmind: How reminders increase saving.Management Science, 62(12), 3393–3411.https://doi.org/10.1287/mnsc.2015.2296.

Klege, R., Visser, M., Datta, S., & Darling, M. (2018). The power of nudging: Usingfeedback, competition and responsibility assignment to save electricity in anon-residential setting (ERSA Working Paper No. 763). Retrieved fromEconomic Research South Africa website: https://econrsa.org/system/files/publications/working_papers/working_paper_763.pdf.

Lansing, J. S. (2003). Complex adaptive systems. Annual Review of Anthropology, 32(1), 183–204. https://doi.org/10.1146/annurev.anthro.32.061002.093440.

https://doi.org/10.1016/j.worlddev.2019.104703https://doi.org/10.1257/aer.104.10.3003http://refhub.elsevier.com/S0305-750X(19)30351-1/h0010http://refhub.elsevier.com/S0305-750X(19)30351-1/h0010http://refhub.elsevier.com/S0305-750X(19)30351-1/h0010https://doi.org/10.1016/j.jpubeco.2014.06.014https://doi.org/10.1080/17437199.2016.1219673https://doi.org/10.1080/17437199.2016.1219673http://Retrieved+from+http://link.springer.com/openurl%3fgenre%3dbook%26isbn%3d978-1-4757-9983-5http://Retrieved+from+http://link.springer.com/openurl%3fgenre%3dbook%26isbn%3d978-1-4757-9983-5http://refhub.elsevier.com/S0305-750X(19)30351-1/h0030http://refhub.elsevier.com/S0305-750X(19)30351-1/h0030http://refhub.elsevier.com/S0305-750X(19)30351-1/h0030http://refhub.elsevier.com/S0305-750X(19)30351-1/h0030http://refhub.elsevier.com/S0305-750X(19)30351-1/h0035http://refhub.elsevier.com/S0305-750X(19)30351-1/h0035http://refhub.elsevier.com/S0305-750X(19)30351-1/h0035http://refhub.elsevier.com/S0305-750X(19)30351-1/h0040http://refhub.elsevier.com/S0305-750X(19)30351-1/h0040http://refhub.elsevier.com/S0305-750X(19)30351-1/h0045http://refhub.elsevier.com/S0305-750X(19)30351-1/h0045https://doi.org/10.1890/1540-9295(2003)001[0376:ASSIE]2.0.CO;2https://doi.org/10.1890/1540-9295(2003)001[0376:ASSIE]2.0.CO;2https://doi.org/10.1257/pol.20130225http://refhub.elsevier.com/S0305-750X(19)30351-1/h0060http://refhub.elsevier.com/S0305-750X(19)30351-1/h0060http://refhub.elsevier.com/S0305-750X(19)30351-1/h0060https://doi.org/10.1177/0309132511425708https://child.org/sites/default/files/Kenya-County-Factsheet.pdfhttps://child.org/sites/default/files/Kenya-County-Factsheet.pdfhttp://refhub.elsevier.com/S0305-750X(19)30351-1/h0075http://refhub.elsevier.com/S0305-750X(19)30351-1/h0075http://refhub.elsevier.com/S0305-750X(19)30351-1/h0075http://refhub.elsevier.com/S0305-750X(19)30351-1/h0080http://refhub.elsevier.com/S0305-750X(19)30351-1/h0080http://refhub.elsevier.com/S0305-750X(19)30351-1/h0080http://refhub.elsevier.com/S0305-750X(19)30351-1/h0080http://refhub.elsevier.com/S0305-750X(19)30351-1/h0080https://doi.org/10.1111/roiw.12093https://doi.org/10.1111/roiw.12093https://doi.org/10.1257/aer.101.6.2350https://doi.org/10.1016/j.technovation.2011.04.006https://doi.org/10.1016/j.technovation.2011.04.006https://doi.org/10.1080/14487136.2017.1301016https://doi.org/10.1080/14487136.2017.1301016http://refhub.elsevier.com/S0305-750X(19)30351-1/h0105http://refhub.elsevier.com/S0305-750X(19)30351-1/h0105http://refhub.elsevier.com/S0305-750X(19)30351-1/h0105http://s3-eu-central-1.amazonaws.com/fsd-circle/wp-content/uploads/2015/08/30093918/15-06-29-Mango-value-chain-report.pdfhttp://s3-eu-central-1.amazonaws.com/fsd-circle/wp-content/uploads/2015/08/30093918/15-06-29-Mango-value-chain-report.pdfhttp://s3-eu-central-1.amazonaws.com/fsd-circle/wp-content/uploads/2015/08/30093918/15-06-29-Mango-value-chain-report.pdfhttp://refhub.elsevier.com/S0305-750X(19)30351-1/h0115http://refhub.elsevier.com/S0305-750X(19)30351-1/h0115http://refhub.elsevier.com/S0305-750X(19)30351-1/h0115https://doi.org/10.1080/10810730.2014.1002958https://doi.org/10.1080/10810730.2014.1002958http://38r8om2xjhhl25mw24492dir.wpengine.netdna-cdn.com/wp-content/uploads/2016/11/WISH-2016_Behavioral_Insights_Report.pdfhttp://38r8om2xjhhl25mw24492dir.wpengine.netdna-cdn.com/wp-content/uploads/2016/11/WISH-2016_Behavioral_Insights_Report.pdfhttp://38r8om2xjhhl25mw24492dir.wpengine.netdna-cdn.com/wp-content/uploads/2016/11/WISH-2016_Behavioral_Insights_Report.pdfhttps://doi.org/10.1017/CBO9781139174329.005http://www.hbs.edu/faculty/Publication%2520Files/18-030_63e232aa-d361-4673-aa2a-9f8dab3dccc2.pdfhttp://www.hbs.edu/faculty/Publication%2520Files/18-030_63e232aa-d361-4673-aa2a-9f8dab3dccc2.pdfhttps://doi.org/10.1080/09718923.2014.11893249https://doi.org/10.1080/09718923.2014.11893249https://www.iea.org/publications/freepublications/publication/WEO2014_AfricaEnergyOutlook.pdfhttps://www.iea.org/publications/freepublications/publication/WEO2014_AfricaEnergyOutlook.pdfhttps://doi.org/10.1787/weo-2018-enhttps://doi.org/10.1787/weo-2018-enhttp://www.irena.org/DocumentDownloads/Publications/Africa_renewable_future.pdfhttp://www.irena.org/DocumentDownloads/Publications/Africa_renewable_future.pdfhttps://doi.org/10.1007/978-4-431-54478-4_4https://doi.org/10.1007/978-4-431-54478-4_4https://doi.org/10.1016/j.erss.2017.11.023https://doi.org/10.1016/j.erss.2017.11.023http://refhub.elsevier.com/S0305-750X(19)30351-1/h0170http://refhub.elsevier.com/S0305-750X(19)30351-1/h0170https://doi.org/10.1287/mnsc.2015.2296https://econrsa.org/system/files/publications/working_papers/working_paper_763.pdfhttps://econrsa.org/system/files/publications/working_papers/working_paper_763.pdfhttps://doi.org/10.1146/annurev.anthro.32.061002.093440


Lee, K., Miguel, E., & Wolfram, C. (2016). Appliance ownership and aspirationsamong electric grid and home solar households in rural Kenya. AmericanEconomic Review, 106(5), 89–94. https://doi.org/10.1257/aer.p20161097.

Levin, S., Xepapadeas, T., Crépin, A.-S., Norberg, J., de Zeeuw, A., Folke, C., ... Walker,B. (2013). Social-ecological systems as complex adaptive systems: Modelingand policy implications. Environment and Development Economics, 18(02),111–132. https://doi.org/10.1017/S1355770X12000460.

Levin, S. A. (1999). Fragile dominion: Complexity and the commons. Retrieved fromhttps://books.google.se/books?id=TpfuAAAAMAAJ.

Liu, E. M., & Huang, J. (2013). Risk preferences and pesticide use by cotton farmers inChina. Journal of Development Economics, 103, 202–215. https://doi.org/10.1016/j.jdeveco.2012.12.005.

Lourenço, J. S., Almeida, S. R., & Troussard, X. (2016). Behavioural insights applied topolicy: European Report 2016. (No. EUR277726). Retrieved from EuropeanUnion website: http://publications.jrc.ec.europa.eu/repository/bitstream/JRC100146/kjna27726enn_new.pdf.

Malmberg, L. (2017). Building design capability in the public sector. Retrieved fromhttp://public.eblib.com/choice/publicfullrecord.aspx?p=4810296.

Manzini, E. (2015). Design, when everybody designs: An introduction to design forsocial innovation. Cambridge, Massachusetts: The MIT Press.

Michie, S., Richardson, M., Johnston, M., Abraham, C., Francis, J., Hardeman, W., ...Wood, C. E. (2013). The behavior change technique taxonomy (v1) of 93hierarchically clustered techniques: building an international consensus for thereporting of behavior change interventions. Annals of Behavioral Medicine, 46(1),81–95. https://doi.org/10.1007/s12160-013-9486-6.

Michie, S., van Stralen, S., & West, R. (2011). The behaviour change wheel: A newmethod for characterising and designing behaviour change interventions.Implement Science, 6.

Michie, S., Wood, C. E., Johnston, M., Abraham, C., Francis, J. J., & Hardeman, W.(2015). Behaviour change techniques: The development and evaluation of ataxonomic method for reporting and describing behaviour change interventions(a suite of five studies involving consensus methods, randomised controlledtrials and analysis of qualitative data). Health Technology Assessment, 19(99),1–188. https://doi.org/10.3310/hta19990.

MoALF. (2016). Climate risk profile: Meru County (No. Kenya County Climate RiskProfile Series). Retrieved from International Center for Tropical Agriculture(CIAT) and Kenya Ministry of Agriculture, Livestock and Fisheries (MoALF)website: https://cgspace.cgiar.org/rest/bitstreams/119947/retrieve.

MOPHS, & IMC. (2010). Integrated health and nutrition survey, Greater Tana RiverDistrict. Retrieved from Ministry of Public Health and Sanitation (MOPHS) andInternational Medical Corps (IMC) website: https://www.humanitarianresponse.info/en/operations/kenya/assessment/integrated-health-and-nutrition-survey-greater-tana-river-district.

Mosler, H.-J. (2012). A systematic approach to behavior change interventions for thewater and sanitation sector in developing countries: A conceptual model, areview, and a guideline. International Journal of Environmental Health Research,22(5), 431–449. https://doi.org/10.1080/09603123.2011.650156.

Muhoza, C., & Johnson, O. (2018). Exploring household energy transitions in ruralZambia from the user perspective. Energy Policy, 121(C), 25–34.

Patrício, L., Gustafsson, A., & Fisk, R. (2018). Upframing service design andinnovation for research impact. Journal of Service Research, 21(1), 3–16.https://doi.org/10.1177/1094670517746780.

Penin, L. (2017). An introduction to service design. London; New York, NY:Bloomsbury Publishing.

Pfannstiel, M. A., & Rasche, C. (Eds.). (2017). Service business model innovation in thehealthcare and hospital management: Models, strategies, tools. Cham, Switzerland:Springer.

Richburg-Hayes, L., Anzelone, C., Dechausay, N., & Landers, P. (2017). Nudgingchange in human services: Final report of the behavioral interventions toadvance self-sufficiency (BIAS) project. (No. 2017–23). Retrieved from Office ofPlanning, Research and Evaluation, Administration for Children and Families, U.S Department of Health and Human Services website: https://www.mdrc.org/sites/default/files/2017_MDRC_BIAS_Final_Report_FR.pdf.

Rittel, H. W. J., & Webber, M. M. (1973). Dilemmas in a general theory of planning.Policy Sciences, 4(2), 155–169. https://doi.org/10.1007/BF01405730.

Rural Electrification Authority. (2016). Rural Electrification Authority, Mpanta SolarMini-grid data.

Schilbach, F. (2019). Alcohol and self-control: A field experiment in India. AmericanEconomic Review, 109(4), 1290–1322. https://doi.org/10.1257/aer.20170458.

Segelström, F. (2013). Stakeholder engagement for service design: how servicedesigners identify and communicate insights. Retrieved from http://public.eblib.com/choice/publicfullrecord.aspx?p=3328041.

Stickdorn, M., & Schneider, J. (2012). This is service design thinking: Basics, tools, cases.Hoboken, NJ: Wiley.

Szabó, S., Bódis, K., Huld, T., & Moner-Girona, M. (2011). Energy solutions in ruralAfrica: Mapping electrification costs of distributed solar and diesel generationversus grid extension. Environmental Research Letters, 6(3), 034002. https://doi.org/10.1088/1748-9326/6/3/034002.

Tantia, P. (2017). The new science of designing for humans. Stanford SocialInnovation Review (Spring, 2017, pp. 29–33).

The Behavioural Insights Team. (2017). The behavioural insights team: UpdateReport 2016-17. Retrieved from The Behavioural Insights Team website: http://38r8om2xjhhl25mw24492dir-wpengine.netdna-ssl.com/wp-content/uploads/2017/10/BIT_Update-16-17_E_pdf.

USAID. (2017a). Designing for behavior change for agriculture, natural resourcemanagement, health and nutrition. Retrieved from The Technical andOperational Performance Support (TOPS) Program, USAID website: https://www.fsnnetwork.org/sites/default/files/DBC_English.pdf.

USAID. (2017b). Kenya agriculture and food security. USAID factsheet. [Fact Sheet].Retrieved from USAID website: https://www.usaid.gov/sites/default/files/documents/1860/Kenya_OEG_Agriculture_and_Food_Security_Fact_Sheet_2017.pdf.

Verschoor, A., D’Exelle, B., & Perez-Viana, B. (2016). Lab and life: Does risky choicebehaviour observed in experiments reflect that in the real world? Journal ofEconomic Behavior & Organization, 128, 134–148. https://doi.org/10.1016/j.jebo.2016.05.009.

Wetter Edman, K., Göteborgs universitet, & Konstnärliga fakultetskansliet. (2011).Service design: a conceptualization of an emerging practice. ArtMonitor:Konstnärliga fakultetskansliet, Göteborgs universitet, Göteborg.

Wetter-Edman, K., Sangiorgi, D., Edvardsson, B., Holmlid, S., Grönroos, C., &Mattelmäki, T. (2014). Design for value co-creation: Exploring synergiesbetween design for service and service logic. Service Science, 6(2), 106–121.https://doi.org/10.1287/serv.2014.0068.

World Bank. (2014). World Development Report 2015: Mind, society, and Behavior.Retrieved from http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-0342-0.

https://doi.org/10.1257/aer.p20161097https://doi.org/10.1017/S1355770X12000460https://books.google.se/books%3fid%3dTpfuAAAAMAAJhttps://doi.org/10.1016/j.jdeveco.2012.12.005https://doi.org/10.1016/j.jdeveco.2012.12.005http://publications.jrc.ec.europa.eu/repository/bitstream/JRC100146/kjna27726enn_new.pdfhttp://publications.jrc.ec.europa.eu/repository/bitstream/JRC100146/kjna27726enn_new.pdfhttp://public.eblib.com/choice/publicfullrecord.aspx%3fp%3d4810296http://refhub.elsevier.com/S0305-750X(19)30351-1/h0220http://refhub.elsevier.com/S0305-750X(19)30351-1/h0220https://doi.org/10.1007/s12160-013-9486-6http://refhub.elsevier.com/S0305-750X(19)30351-1/h0230http://refhub.elsevier.com/S0305-750X(19)30351-1/h0230http://refhub.elsevier.com/S0305-750X(19)30351-1/h0230https://doi.org/10.3310/hta19990https://cgspace.cgiar.org/rest/bitstreams/119947/retrievehttps://www.humanitarianresponse.info/en/operations/kenya/assessment/integrated-health-and-nutrition-survey-greater-tana-river-districthttps://www.humanitarianresponse.info/en/operations/kenya/assessment/integrated-health-and-nutrition-survey-greater-tana-river-districthttps://www.humanitarianresponse.info/en/operations/kenya/assessment/integrated-health-and-nutrition-survey-greater-tana-river-districthttps://doi.org/10.1080/09603123.2011.650156http://refhub.elsevier.com/S0305-750X(19)30351-1/h9000http://refhub.elsevier.com/S0305-750X(19)30351-1/h9000https://doi.org/10.1177/1094670517746780http://refhub.elsevier.com/S0305-750X(19)30351-1/h0260http://refhub.elsevier.com/S0305-750X(19)30351-1/h0260http://refhub.elsevier.com/S0305-750X(19)30351-1/h0265http://refhub.elsevier.com/S0305-750X(19)30351-1/h0265http://refhub.elsevier.com/S0305-750X(19)30351-1/h0265https://www.mdrc.org/sites/default/files/2017_MDRC_BIAS_Final_Report_FR.pdfhttps://www.mdrc.org/sites/default/files/2017_MDRC_BIAS_Final_Report_FR.pdfhttps://doi.org/10.1007/BF01405730https://doi.org/10.1257/aer.20170458http://public.eblib.com/choice/publicfullrecord.aspx%3fp%3d3328041http://public.eblib.com/choice/publicfullrecord.aspx%3fp%3d3328041http://refhub.elsevier.com/S0305-750X(19)30351-1/h0295http://refhub.elsevier.com/S0305-750X(19)30351-1/h0295https://doi.org/10.1088/1748-9326/6/3/034002https://doi.org/10.1088/1748-9326/6/3/034002http://38r8om2xjhhl25mw24492dir-wpengine.netdna-ssl.com/wp-content/uploads/2017/10/BIT_Update-16-17_E_pdfhttp://38r8om2xjhhl25mw24492dir-wpengine.netdna-ssl.com/wp-content/uploads/2017/10/BIT_Update-16-17_E_pdfhttp://38r8om2xjhhl25mw24492dir-wpengine.netdna-ssl.com/wp-content/uploads/2017/10/BIT_Update-16-17_E_pdfhttps://www.fsnnetwork.org/sites/default/files/DBC_English.pdfhttps://www.fsnnetwork.org/sites/default/files/DBC_English.pdfhttps://www.usaid.gov/sites/default/files/documents/1860/Kenya_OEG_Agriculture_and_Food_Security_Fact_Sheet_2017.pdfhttps://www.usaid.gov/sites/default/files/documents/1860/Kenya_OEG_Agriculture_and_Food_Security_Fact_Sheet_2017.pdfhttps://www.usaid.gov/sites/default/files/documents/1860/Kenya_OEG_Agriculture_and_Food_Security_Fact_Sheet_2017.pdfhttps://doi.org/10.1016/j.jebo.2016.05.009https://doi.org/10.1016/j.jebo.2016.05.009https://doi.org/10.1287/serv.2014.0068http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-0342-0http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-0342-0

Embracing complexity: A transdisciplinary conceptual framework for understanding behavior change in the context of development-focused interventions1 Introduction and aim2 Theoretical background2.1 Social-ecological systems theory2.2 Behavioral insights for low-income settings2.3 Service design

3 Generic description of the methodological framework4 Application of the conceptual framework4.1 Identifying user archetypes and opportunities for behavior change techniques: Case study on adoption of advanced biomass cookstoves in Kenya and Zambia4.1.1 Methods4.1.2 Key results and discussion

4.2 User mapping to understand the wider context: case study on solar PV mini-grids for household electricity provision in rural Zambia4.2.1 Methods4.2.2 Key results and discussion

4.3 Situating individual behavior within the wider system: case study on technologies for reducing post-harvest losses in small scale mango harvesting in Kenya4.3.1 Methods4.3.2 Key results and discussion

5 General discussion5.1 Behaviour change occurs (and is reinforced) within separate but interconnected phases of the actor journey5.2 Identifying where BCTs or groups of BCTs could be applied5.3 The importance of user archetypes5.4 Connecting interventions to the wider social-ecological context5.5 Limitations and future research

6 ConclusionsDeclaration of Competing InterestAcknowledgementsAppendix A Supplementary dataReferences

embracing complexity: a transdisciplinary conceptual...

Documents