Paper presented at the 33" International Conference of the System Dynamics Society, July 19-23,
2015, Cambridge, MA
Learning as conceptual change during community based group
interventions. A case study with smallholder farmers in Zambia.
Gerid M. Hager’’, Birgit Kopainsky’, Progress H. Nyanga”
1system Dynamics Group, Department of Geography, University of Bergen, Postbox 7800, 5020
Bergen, Norway
* E-mail: geridhager@gmail.com
? Geography and Environmental Studies Department, School of Natural Sciences, University of
Zambia, P.O. Box 32379, Lusaka, Zambia, 10101
Abstract
Food and farming systems are both complex and dynamic in nature. While smallholder farmers
hold great practice-based knowledge, rapidly changing and ever more challenging framework
conditions require them to constantly learn and adapt to change. Understanding knowledge
change is important to improve the design of shared learning experiences, especially as they are
one prerequisite for the effective implementation of adaptive policies. This paper studies
knowledge change within smallholder farmers in Zambia as a result of systems thinking group
interventions. We report on the design, implementation and preliminary analysis of two group
workshops with ten smallholder farmers each and follow-up interviews with some of the
participants. Our results so far suggest that the use of system maps and collaboratively unfolding
them seem crucial components in the process of knowledge change. Thanks to the system map,
participants display a fairly high degree of convergence on systemic issues and policy
considerations during post-workshop interviews. Their explanations of challenges and possible
options to overcome them are not limited to the available visual components of the
collaboratively developed system map. Instead, participants extend the discussion of options
beyond the system map while remaining able to reason in a concise, systemic way.
“The lesson is very much important in someone’s
life and is better than these other gifts.”
Participant after receiving a
group picture as a token of
appreciation for taking time
for a follow-up interview.
Introduction
Food and farming systems are both complex and dynamic in nature. They require stakeholders to
constantly learn and adapt to change, especially in the face of expected long-term climate change
and situations causing short- to medium-term food insecurities, such as local weather, storage
insufficiency, singular crop dependency or economic precariousness (Ericksen 2008, Folke et al.
2005). While smallholder farmers hold great practice-based knowledge about the farming and
food system, it is believed that both the integration of different local knowledge bases and the
integration of local and science-based knowledge increases adaptive capacity of those dependent
on and regularly interacting with the system (Berkes et al. 2003, Mackinson & N¢@ttestad 1998).
However, to learn is not only to add new knowledge but the transformation and creation of
understanding during a learning process. Conceptual change theories explore possible learning
mechanisms, such as mental model building, knowledge integration or socially situated tool
appropriation (Mayer, 2002). The field of System Dynamics and system thinking has a long
tradition in facilitating learning about complex systems, through the use of models (Sterman 2001,
Lane 1992, Vennix 1996). Mediated modeling or community based group model building uses the
methodology to build local knowledge capacity and improve decision making in dynamic settings
(Metcalf 2010, Hovmand, 2014). While there has been extensive research on learning outcomes
such as performance or differences between prior and posterior understanding, little research has
been done to understand how exactly conceptual change takes place in the course of System
Dynamics and system thinking learning interventions (Kopainsky & Saldarriaga, 2012). However, to
better understand the unfolding of conceptual change during a system thinking intervention can
potentially give valuable insight to future design and expected deliverables of such interventions
as well as increase the ability of practitioners to meaningfully embed them in various contexts.
This paper reports on the research into how knowledge change within smallholder farmers
corresponds to the use of qualitative system models and systems thinking group interventions as
part of a broader attempt to equip them with effective and sustainable short-and long-term
strategies to increase their food security as well as ensure their economic stability. The study aims
at uncovering mechanisms that promote conceptual change, particularly the integration of diverse
local as well as science-based knowledge during the course of a facilitated systems thinking group
intervention to inform future intervention planning and design in the community learning field.
In order to reach the objectives the following research questions need to be answered:
A1. What does farmer’s knowledge look like after a facilitated systems thinking group
intervention?
B1. What mechanisms can be identified that facilitate or inhibit knowledge change?
B2. How does farmer’s knowledge and understanding of the farming and food system develop
in the course of a facilitated systems thinking group intervention?
Research Design
Analytical Framework and Definitions of Concepts
Figure 1 shows the analytical framework for the study including research (left column) that was
conducted by Spicer (2015). The research design is based on considerations in multiple fields,
amongst which are: the learning sciences, particularly conceptual change theories (Limon &
Mason 2002; Parnafes et al. 2008; Greeno & van de Sande 2007), facilitated modeling, particularly
group model building and community based system dynamics (Vennix 1996; Franco & Montibeller
2010; Hovmand 2014), and systems thinking, particularly qualitative system dynamics (Lane 1995;
Wolstenholme 1999). The fields of facilitated modeling and systems thinking inform the stage of
intervention design. They provide methods for group modeling (incl. diagrams, sign systems and
procedural conventions) as well as content and definitions for knowledge analysis, such as
dynamic system concepts or systemic reasoning levels. The learning sciences offer procedures and
frameworks for the study of shared learning processes.
| BI. What change mechanisms can _: Al. What does their knowledge
| be identified? | look like after the intervention?
| B2. How does knowledge develop in
the course of the intervention? i
System Model 1 il Systems Thinking Group System Model 2
(Input) ' Intervention ! (Output)
‘ Model '
: Interaction} +
‘ v
Intuitive and
Experiential Knowledge Conceptual Change Integrated Knowledge
Figure 1: Research questions and analytical framework of conceptual knowledge change in the course of a systems
thinking group intervention using qualitative system models
System models as used in this framework serve different purposes. On the one hand, they are
defined as conceptual artifacts and interpretive maps of understanding inductively derived from
interview data (Kim & Andersen 2012). On the other hand, they are seen as transitional or
boundary objects (Franco 2013; Hovmand 2014), or so-called “things to think with” (Roepstorff
2008) jointly created during a systems thinking intervention offering specific affordances to the
learners. The notion of boundary objects has been used and supported in the field of participatory
modeling and group model building to capture the innate purpose and function of diagrams in
shared learning experiences (Scott et al. 2014; Black & Andersen 2012; Paroutis et al. 2015).
Diagrams qualify as boundary objects if they help a group develop or use shared language
(syntactic boundary), shared content (semantic boundary), and a shared vision for action
(pragmatic boundary). Well-known characteristics or affordances of system diagrams are their
tangibility, mutability and associability. They come in one tangible form that can be perceived by
all participants, they can be changed on the spot and they visualize relationships and
dependencies amongst aspects relevant to the participants (Black & Andersen 2012; Paroutis et al.
2015). Conceptual change is broadly understood as a process of learning and knowledge
development (Mayer 2002). We use the terms conceptual change, conceptual development and
knowledge development interchangeably. More specifically, conceptual change is framed here in
accordance with Greeno and van de Sande's (2007) integrated view of situated and distributed
cognition. It is seen as changing and achieving coherent states of understanding through
participation and interaction in groups.
Methodology
Given the qualitative nature of the research questions, the small sample and the fairly natural
context of the study, the research strategy follows a qualitative approach and is carried out as a
mixed-methods case study. The types of knowledge gained are explanatory (B1) and descriptive
(A1, B2). The data was collected in form of audio/videotaped (participant) observation as well as
semi-structured interviews.
Knowledge analysis (Parnafes et al., 2008) is directed at research questions A1, B1 and B2.
Knowledge analysis attempts to identify and track changes in categories/units of knowledge held
by the participants and to investigate knowledge integration during the intervention as well as
during post-intervention interviews.
Knowledge analysis is complemented by interaction and conversation analysis used in the group
learning and group model building literature as well as the soft OR field associated with facilitated
modeling (cf. Canary and Seibold 2010, diSessa 2007, Jordan and Henderson 1995, Franco and
Rouwette 2011) to fully exploit possible insights to research questions B1 and B2. Two dimensions
of interaction are interesting to consider:
1. The interaction between group members during the intervention, as well as
2. The interaction of group members with the model.
Case Study
Case study region
Zambia, officially the Republic of Zambia, is a landlocked country in Sub-Saharan Africa. It lies
south of the Democratic Republic of Congo and Tanzania, north of Mozambique, Zimbabwe,
Botswana and Namibia and shares boarders with Angola to the west and Malawi to the east.
According to the Center for International Futures, its population in 2015 is estimated at 15 million
people with a very high annual growth rate of 2.77 % and a projected doubling of its population
from about 15 million in 2015 to 30 million in 2045. With an area of roughly 750,000 sq. km it
remains having low population density (compare with Turkey: it has a similar size but a population
of 80 million). In 2012 nearly two thirds of the country area were covered by forest and 30% were
used for agriculture (FAO, Food and Agriculture Organization of the United Nations). Zambia also
faces major challenges. Almost 30% of its urban and 80% of its rural population live below the
national poverty line of 1.25$/day PPP (Millennium Development Goals Indicators, United Nations
Statistics Division). The Global Food Security Index places it amongst the 10 most food insecure
countries globally with little less than half of its population qualifying as undernourished. Zambia
scores absolute lows on food affordability, quality and safety (105 out of 109 countries) and low
on food availability (83 out of 109).
Historically, Zambia’s economy and most of its export volume is dependent on the copper
industry. However, as of 2012, 56% of its total labor force worked in agriculture, with an average
of 80% in rural areas (2012 Labour Force Survey, Central Statistical Office Zambia). The national’s
crop number one is maize, followed by
groundnuts, cotton, soybeans and
mixed beans, sunflower, sweet
potatoes, wheat, millet, rice, sorghum
and tobacco. Smallholders account for
the vast majority of farms, cropped
area, maize production, and fertilizer
use. Agriculture and agricultural
practices face multiple challenges,
affecting small- and medium-scale
farmers most. Accessibility to
functioning market and transportation
infrastructure is lacking. Frequently, Figure 2: Chibombo District in Central Province, Zambia.
farmers cannot afford or access inputs (Source: http://en.wikipedia.org/wiki/Districts_of_Zambia)
such as seeds, fertilizers, or herbicides at a reasonable price and at the right time. Especially small-
scale farmers have no options for or knowledge of proper financial management and services, let
alone access to long-term investments. At the same time, small-scale farming practices tied to
staple crop production are often unbalanced and result in soil erosion and a loss in soil fertility,
which destabilizes productivity in the long run (Neubert et al. 2011). Furthermore, external
drivers, such as climate change and economic shocks, are posing increasingly significant challenges
to the agricultural sector. The region has seen a rise in mean temperature over the last four
decades. Furthermore, rainfall patterns have changed significantly since the late 1980s (de Wit
2006; Fumpa-Makano 2011).
In this paper, we present data from two workshops and corresponding follow-up interviews in
Chisombo district. Chibombo district is located in Central Province between the country’s capital
Lusaka and Kabwe (Figure 2). It is about 90 km to the north of Lusaka and is largely a farming
district where about 90% of the district population depend on agriculture for their livelihoods. The
district recieves rainfall between 800 and 1200 mm per year. Climatic conditions make it suitable
for production of maize, cotton, sunflower, cowpeas, beans, groundnuts, paprika, soya beans, and
tobacco etc. The agricultural sector is characterized by a mixture of commercial and small-scale
farmers. Over seventy-five (75%) of crop production in the district is done by small scale farmers
and commercial agriculture is largely concentrated in the south of the district. Our case study
villages, Chitumbo and Mpikwa, are located in the north of the district where small-scale
agriculture is dominant.
Participants
Workshop participants were recruited from two villages in Chibombo district, namely Chitumbo
and Mpikwa village. An initial listing of all the farmers in the two villages provided an overview of
their respective socio-economic characteristics such as household size, farm size, animal stock,
education level, asset endowment, input use, planting time and planting method. Based on these
characteristics, ten participants from each village were selected for the workshops. Participants
were household heads, five of them female and five of them male in each workshop. Additional
selection criteria included the representation of diverse socio-economic household characteristics
and the willingness of participants to attend the workshop without payment. At the end of each
workshop, we distributed some basic household items as a token of appreciation. However,
farmers’ motivation to attend was interest rather than the promise of any payment.
Intervention Design
Pre-workshop considerations
The starting point for preparing the two
workshops was a basic causal loop diagram
(CLD) that represented the main feedback
loops driving household food security. The
available food
CLD is based on in-depth interviews with 20
small-scale farmers in four regions in
Zambia. Four of the workshop participants
from worshop 1 had taken part in those
interviews. The interviews focused on
produced food
for further information). Figure 3: CLD as a workshop starting point
small-scale farmers’ dynamic decision
making in the context of food security and
livelihoods (see section on data collection
Figure 3 shows the base CLD used for the workshop design. It links the key stock variables through
two major reinforcing feedback loops. Available financial assets (“cash”), the amount of food
produced in a year (“produced food”) and the actual food security situation (“available food”) are
at the heart of farmers’ concerns. The more cash a household has available, the more inputs
(fertilizer, seed, herbicides) they can purchase. More inputs allow farmers to cultivate more land
and thus increase the amount of food they produce. Some of this food is sold directly (cash crops)
and some is only sold if there is any surplus between what is produced and what is required to
feed the household (represented in the variable “available food”). The more a household can sell,
the more income they generate which adds to the stock of cash. A similar reinforcing feedback
loop links livestock to cash (through sales of animal products or animals) and back to livestock
(through purchases of additional livestock). Note that Figure 3 only describes the reinforcing loops
that either drive the accumulation of food and financial assets or lead to their continuous
depletion, depending on whether a household manages to exceed a critical resource threshold (cf.
Stephens et al., 2012). The figure does not show the manifold balancing feedback loops that
regulate the accumulation of the stocks and the operation of the loops. Instead, it focuses on the
main structure relevant for explaining the notion of vicious and virtuous cycles and the existence
of thresholds that determine loop direction. For similar reasons of clarity and simplicity, we did
not indicate the polarity of the relationships in the diagram. For the processes displayed in Figure
3 this was not crucial.
The workshop design resembled a basic form of group model building (Vennix 1996, Hovmand
2014) using qualitative system models. Figure 6 illustrates the different phases of the participatory
process. The workshops lasted between 2.5 and 3.5 hours each.
Workshop phase one
The diagram as shown in Figure 3
primarily served as a mental guiding
aid for us as workshop facilitators. We
started by subsequently putting all
seven variables on the table and
discussed them without yet
connecting them. Instead, we used
visual and tangible objectes to make
sure that participants agreed on
variable names and what these
variables represented. Figure 4 shows
the diagram from the first workshop
with the variable names written ona __ Figure 4: Placing objects (seeds, fertilizer, salt, coins etc.) and pictures
to develop a shared understanding of meanings of variables.
piece of paper, complemented with
pictures and tangible objects supporting the meaning of the variable as well as a participant
actively placing a bag of harvested seeds on the available food variable.
Workshop phase two
After introducing the variables, we gradually linked them until the two main feedback loops were
complete.
Workshop phase three
For illustrating the mechanism of the two loops, we used glasses for each of the stock variables.
Participants gradually and consequitively filled them with water to their desired levels as we
collaboratively unfolded the operation of the reinforcing processes. Figure 5 shows the diagram
with pictures, objects and glasses for four of the five stock variables.
Workshop phase four
Subsequently, we focused on the discussion of stresses on the system (processes that “drain the
glasses”) and available options to shift loop direction towards a virtuous cycle and to increase
accumulation in the respective resource stocks.
Figure 5: Representation of stocks and levels in the system model with pictures, objects and water glasses
Figure 6: Different stages of the participatory model building process
Post-workshop interviews
The interviews were conducted one week after the workshops, asking open questions about (1)
structural components of the system of interest (resembling variables, causal structures, loops),
(2) policy options for averting unfavorable stock levels as well as their dynamic implications, (3)
the use of objects, pictures and visual diagram properties, such as arrows and loops, (4) potentially
unresolved disagreements with what had been discussed during the workshop, (5) moments of
surprise during the workshop, (6) potentially “new insights” gained as well as key lessons learnt,
and (7) participant’s impression on the participatory method in contrast to more commonly used
methods in theoretical rural farmer education, such as lecturing. The interviews were supported
by prints of pictures of the final workshop diagrams (Figure 8 and 9 respectively) as the goal was to
better understand if the participants could use the diagrams in a meaningful way, and not to
plainly test whether they had memorized anything from the workshop.
Data Collection
The primary sources of the research are videos of the group workshops conducted by two of the
authors (BK, PN) as well as transcriptions of the videos, translated from Tonga to English (cross-
checked by PN). A further primary source are semi-structured post-workshop interviews, also in
the form of videos and translated transcripts.
A base CLD of farmer’s understanding of their food and farming system (Figure 3) is used as a
secondary data source. It functions as a starting point for the group intervention as well as a
reference point for subsequent analyses of conceptual change. Spicer (2015) inductively
developed the CLD using primary data from in debth interviews (Saldarriaga et al., 2014). For the
purpose of the workshops, the diagram was simplified as much as possible.
Data Analysis
Part 1 — Analyzing mediated conversational action (research questions B1 and B2)
(i) To analyze the workshop transcript we looked for coding schemes that had been used in similar
contexts and considered similar theoretical implications. We compared different existing
approaches and coding schemes for analyzing conversational interaction in collaborative learning
settings (Roschelle 1992; Hmelo-Silver 2003; van Boxtel et al. 2000; Fischer et al. 2002; Arvaja et
al. 2007; Mercer 2004; Kumpulainen & Mutanen 1999). We decided to use Canary and Seibold's
(2010) Conversational Argument Coding Scheme (CACS), which allows for a close look into verbal
communicative interaction between people. It focuses on and is limited to the communicative
function of people’s utterances in their attempt to find agreement and “reach accord with the
minds and behaviors” (ibid, p.12) of others. The unit of analysis is the vocalized thought or speech
turn, most often an independent utterance indicating an idea or distinct thought. If an idea or
support for it changes, a new thought turn is present. The scheme distinguishes between starting
points of an argumentative structure, developing points, convergence markers, prompters and
delimiters. Apart from being a coding scheme for conversational interaction, it also offers a
systematic frame for the interpretation of different types of argument structures. The entire
workshop transcript was coded using the CACS following the CACS manual (Canary, 1992).
(ii) Inspired by Paroutis et al. (2015) we take an inductive approach to analyze tool use and tool-
mediated interaction. Accordingly, we code the entire session based on how participants use and
interact upon the central diagram.
(iii) Furthermore, we draw on the idea of diagram affordances as applied in the operational
research and System Dynamics field (Franco 2013; Paroutis et al. 2015; Black & Andersen 2012)
and insights on the role of visual representations in group learning situations (Suthers 2005). The
aim is to expand on the potential role of system diagrams used in community-based system
thinking workshops.
Part 2 — Analyzing conceptual change towards shared systems understanding (reseach questions
Al and B2)
(i) Given the assumption that we are able to phenomenologically capture some of those processes,
we coarsely identify episodes in the workshops, within which conceptual change may have take
place. The identification is based on descriptions of conceptual change as synthesizing prior
knowledge with new experiences, repairing conflicting conceptions through replacement and
organizing pieces of knowledge and socially forming agreement upon the use of tools (Mayer,
2002) as well as the notion of convergence as a potential indicator in collaborative settings
(Roschelle, 1992). Then, we qualitatively analyze those episodes and describe the unfolding of
knowledge change across types and qualities of knowledge (de Jong and Ferguson-Hessler, 1996)
in relation to Hopper and Stave’s (2008) Taxonomy of System Thinking Characteristics.
(ii) We triangulate the results by drawing on interview data collected one week after the
workshops. Given that the participants had no prior training of causal mapping or explicit socio-
ecological systems thinking, if collaborative conceptual change has taken place to some degree
and it was of convergent form, we would at least expect to see (1) a person’s ability to handle the
diagram and its logic to recollect and form causal chain and loop arguments as well as make
dynamic inferences relating to the contents discussed during the workshop, (2) evaluations of the
tools used resembling an understanding of their intended purposes and (3) a fair amount of
overlapping responses from participants. As regards (1) and (3), a compiled system map is created
inductively based on the information shared in the interviews displaying matching parts between
individuals as well as further individual contributions. The map is complemented by short
sequences and analyses of literal explanations from participants to assess the nature of their
argumentation. (2) is achieved by collecting, comparing and analyzing participant’s responses to
evaluate the tools and concepts used during the workshop.
Results
They results from the detailed analysis suggest that knowledge integration of local based
knowledge with science-based systemic knowledge can be facilitated through community based
system thinking workshops using system models.
There are five distinct and constitutive system concepts required for the farmers in this case study
to develop an intermediate level of system understanding that equips them with systemic
reasoning capacity and some basic form of systemic conceptual agency:
- The concept of variables as independent ontological entities, disentangled from
their daily and procedural use
- The conceptual understanding of causal relations between those entities (incl.
notion of their “nature”)
- The conception of a closed loop
- The concept of stock levels, and
- The concept of reinforcement.
To develop the above concepts, participants engage in a collaborative effort to establish
shared understanding. In order to do so, they change qualitative knowledge aspects and
transition in-between knowledge types. Also, two knowledge domains are at play at the same
time and overlap: knowledge of the farming and food domain, and knowledge of the systems
thinking domain. Finally, the further the process of system mapping, the less intense becomes
the conceptual learning effort but the more complex and fluid becomes knowledge
integration. In other words, while in the beginning participants are busy developing
understanding on concepts from a systems perspective, in the end they are busy integrating
them with other forms of knowledge in more complex tasks such as strategy development and
evaluation.
The role of using system models and the process of collaboratively and interactively unfolding
the model are crutial components in the process of conceptual change. With the system map
at hand, participants display a fairly high degree of convergence on systemic issues and policy
considerations during the post-workshop interviews. They also converge on their use of
language, their ability to meaningfully draw connections between crucial variables and identify
loops, which is in stark contrast to Spicer (2015), who discusses the challenges of coding the
multi-faceted uses of language and meanings in their in-debth interviews. Interestingly, the
explanations given are not limited to the available visual components of the diagram (Figure 8
and 9 respectively). In fact, participants extend the discussion of issues beyond the visible
while remaining able to reason in a concise, systemic way. Figure 7 shows the synthezised
representation of verbal expressions of system features of three interview transcripts, which
together with the qualitative analysis of individual’s statements serves as a partial answer to
research question A1.
Literal excerpt from one of the post-workshop interviews:
“From cash | come here on inputs because when you have cash you can buy inputs, after buying
inputs | come to land but before | come to land first it will rain and after it has rained | get the inputs
and then plant. When | finish | come here on food produced, this same food produced | will reserve
some for food and the surplus i will sell to have cash again. Then from this cash after selling of the
surplus food | come and buy some livestock. The good part of the livestock is that they provide us with
the manure such as chicken manure or any other animal manure which we use to marinating our soil
fertility since we talked about the ways of maintains the soil.”
The causal chain described is extensive and it includes the major production loop, as well as the
cash-livestock-manure loop, which feeds back into the production loop through soil fertility.
What’s more, it is far from being a mere recitation going around the visually available loops in the
diagram. This becomes clear in the explicit portrayal and discussion of parts within the system
where multiple factors interconnect (the separation of food produced into surplus food for sale
and food for consumption and the dependence of cultivation of land on rain). It is also visible in
the value statement of appreciation of livestock (“The good part of the livestock is that...”) as
providing positive influence on another part of the system, namely soil fertility and its function to
maintain productivity of the land.
Sales" sie Decas
Prevention
Buming Land \VZ
a ". i ——
a ee or tere
ms (Seeds,
Fertilizer, Oxen,
Food a |
Food Surplus senate)
h Rain/Water
| Land owned/rented
Desired Food Abusive Sales
ial Farming
Lae Crop se
Available Food
ee a
Proper Storage
Avoiding SS Se land)
Carelessness / Debt Payback ~~
G ing Produced Py ae
— * Trees
Debt Avoidance
Figure 7: Map of participant’s causal chain descriptions shared in three interviews following workshop 2
More than anything, Figure 7 provides a visual representation of the participants’ ability to actively
use the diagram at hand to verbally recreate and share understanding of a multiloop system ina
causally consistent manner, actively responding to and therfore framing it according to the
interviewer's questions and focus on potentialities.
Furthermore, we are able to approximate research questions B1 and B2. The use of tangible
objects was important in two ways. First, the pictures and objects representing variable names
urged participants to agree on the terminology as well as the meaning of variables before they
proceeded to building the system model and discussing shocks and adaptation options. This
discussion also helped illustrating the notion of causality. For example, some of the participants
repeatedly placed the picture and also the physical object representing inputs on the land variable
and explained that inputs allowed them to cultivate more land. The disentanglement of the
procedural aspect of “inputs go on the land” and categorical aspects (land differs from inputs) as
well as their relationship enabled a brief discussion on the differentiation between cause and
effect.
Figure 8: Final system model resulting from workshop 1
Second, the water glass analogy was powerful for illustrating the behavior a reinforcing feedback
loop gives rise to. In addition, it provided a crucial anchoring point for the subsequent discussion
of stresses to the system and available options to shift loop direction towards a virtuous cycle and
to increase accumulation in the respective resource stocks. The water glasses provided an intuitive
way of seeing stresses to the system as processes that drain the glasses and available options as
possibilities to (re)fill the glasses. The terms “draining the glasses” and “filling the glasses”
occurred repeatedly throughout the workshops. Participants intuitively integrated them to
illustrate their comments and evaluations, without further probing by the facilitators.
Figure 9: Final system model resulting from workshop 2
Discussion and conclusions
The purpose of this paper is to report on the study of knowledge change in smallholder farmers in
Zambia who were exposed to a group systems thinking intervention. In the course of
approximately three hours, participants went through the entire process of building and analyzing
a conceptual system dynamics model. Insights into the way in which their knowledge changed and
the factors that facilitated or inhibited this process are important in two regards.
On the one hand, they provide a basis for the design of effective strategies that increase farmers’
adaptive capacity in the long term. The complexity and diversity of farming systems and the range
of uncertainty they face are such that there are no universal practices and technologies for
achieving and maintaining food security and other food system outcomes (e.g. Janssen &
Anderies, 2013). It is therefore crucial that smallholder farmers understand the direct as well as
indirect impacts of available practices and technologies and that they are able to adjust them to
their specific circumstances.
On the other hand, the insights from the workshops in Zambia have important methodological
implications. The conditions for community based systems thinking interventions in rural Zambia
with smallholder farmers differ to a fair extent from contexts within which group model building
has traditionally been employed and conducted (cf. Vennix 1996, Rouwette and Vennix, 2006).
Even participatory modeling in social-ecological contexts primarily include stakeholders with
institutional affiliations rather than groups often marginalized and unrecognized in formal decision
making contexts (Metcalf et al. 2010, Beall and Zeali 2008, Suwarno et al. 2009). Also, the
community based process faces what might be perceived as particular limitations, especially when
embarking on from a system dynamics perspective. Apart from a few, the participants have rather
low levels of education (basic levels of reading, writing and arithmetic) or no formal education at
all. Furthermore, the materials meaningfully available for such interventions lack computers,
presentation screens or projectors, which makes quantitative modeling and simulation rather
impossible. This research is therefore one of the comparatively few attempts to expand the field
to such new target groups and contexts. And the results as described are promising.
On a more anecdotal note, at the end of the post-workshop interview, one of the participants was
very motivated to have us return in some time so he could share what they had achieved by
adopting some of the discussed options based on insights from the workshop. However anecdotal,
it makes clear a sense of empowerment and commitment to action in the participant. The
question remains, and is a vast field for future research, whether and how that commitment can
be sustained and transformed into meaningful action. It also remains unclear whether some rather
simple heuristic of “trying to keep the cups full”
increase stock levels — and the focus on potentialities in the virtuous reinfocing process without
explicitely discussing the complete structure in use can partially guide informed decision making
and behavior change in the long run.
— to monitor and employ options to keep or
Finally, and relating back the insights to the practicing field of systems thinking and system
dynamics we offer new variations of methodological tools for system thinking and group model
building for community-based system dynamcis, which can also potentially result in new group
model building scripts for Scriptapedia (Hovmand et al. 2011). Amongst them are scripts for
variable clarification through the use of tangibles and pictures as well as definition of desired
states and illustrating dynamics using glass cups and liquid.
Acknowledgements
Data collection was funded by the Peder Sather Center through the project “Knowledge analysis in
coupled social-ecological systems. A pilot study in smallholder farmer communities in Zambia”.
One of the authors (BK) is supported by the Norwegian Research Council through the project
“Simulation based tools for linking knowledge with action to improve and maintain food security
in Africa” (contract number 217931/F10). The views and conclusions expressed in this paper are
those of the authors alone and do not necessarily reflect the views of the funding agencies. We
are extremely grateful to Jacynta Spicer for her generous help in the data collection and to James
Siamulilo and Benson Kacha for fieldwork assistance.
References
Arvaja, M. et al., 2007. Combining individual and group-level perspectives for studying
collaborative knowledge construction in context. Learning and Instruction, 17: 448-459.
Beall, A. & Zeali, L., 2008. Participatory modeling of endangered wildlife systems: Simulating the
sage-grouse and land use in Central Washington. Ecological Economics: 24-33
Berkes, F., Colding, J., & Folke, C. 2000. Rediscovery of Traditional Ecological Knowledge as
Adaptive Management. Ecological Applications, 10: 1251-1262.
Black, L.J. & Andersen, D.F., 2012. Using visual representations as boundary objects to resolve
conflict in collaborative model-building approaches. Systems Research and Behavioral Science, 29:
194-208.
Canary, D.J., 1992. Manual for coding conversational arguments. Unpublished manuscript,
Department of Speech Communication, The Pennsylvania State University, University Park.
Canary, D. J., & Seibold D. R. 2010. Origins and Development of the Conversational Argument
Coding Scheme. Communication Methods and Measures, 4: 7-26.
de Jong, T. & Ferguson-Hessler, M.G.M., 1996. Types and qualities of knowledge. Educational
Psychologist, 31(2): 105-113.
diSessa, A. A. 2007. An interactional analysis of clinical interviewing. Cognition and Instruction, 25:
523 - 565.
de Wit, M., 2006. Climate Change and African Agriculture, Policy Note No. 27, CEEPA, University of
Pretoria.
Ericksen, P. J. 2008. Conceptualizing food systems for global environmental change research.
Global Environmental Change, 18: 234-245.
Fischer, F. et al., 2002. Fostering collaborative knowledge construction with visualization tools.
Learning and Instruction, 12: 213-232.
Folke, C., Hahn, T., Olsson, P., & Norberg, J. 2005. Adaptive governance of social-ecological
systems. Annual Review of Environment and Resources, 30: 441-473.
Franco, L. A. 2013. Rethinking Soft OR Interventions: Models as Boundary Objects. European
Journal of Operational Research, 231: 720-33.
Franco, L.A. & Montibeller, G., 2010. Facilitated modelling in operational research. European
Journal of Operational Research, 205(3): 489-500.
Franco, L. A. & Rouwette, E. A. J. A. 2011. Decision development in facilitated modelling
workshops. European Journal of Operational Research, 212: 164-178.
Fumpa-Makano, R., 2011. Forests and Climate Change, Integrating Climate Change Issues into
National Forest Programmes and Policy Frameworks, Food and Agriculture Organization of the
United Nations (FAO), Lusaka
Greeno, J.G. & van de Sande, C., 2007. Perspectival Understanding of Conceptions and Conceptual
Growth in Interaction. Educational Psychologist, 42(1): 9-23.
Hmelo-Silver, C.E., 2003. Analyzing collaborative knowledge construction: Multiple methods for
integrated understanding. Computers and Education, 41: 397-420.
Hooper, M. & Stave, K. 2008. Assessing the effectiveness of systems thinking interventions in the
classroom. Proceedings of the 26th International Conference of the System Dynamics Society.
Hovmand, P. S. 2014. Community Based System Dynamics. New York, NY: Springer New York.
Hovmand, P., Rouwette, E. A. J. A., Andersen, D., Richardson, G. P., Calhoun, A., Rux, K., et al.
2011. Scriptapedia: a handbook of scripts for developing structured group model building sessions.
Paper presented at the International System Dynamics Conference, Washington, DC.
Janssen, M. A., & Anderies, J. M. 2013. A multi-method approach to study robustness of social—
ecological systems: the case of small-scale irrigation systems. Journal of Institutional Economics,
9(4): 427-447.
Jordan, B., & Henderson, A. 1995. Interaction Analysis: Foundations and Practice. Journal of the
Learning Sciences, 4: 39-103.
Kim, H. & Andersen, D.F., 2012. Building confidence in causal maps generated from purposive text
data: mapping transcripts of the Federal Reserve. System Dynamics Review, 28(4): 311-328.
Kopainsky, B., & Saldarriaga, M. 2012. Assessing understanding and learning about dynamic
systems. Paper presented at the 30" International Conference of the System Dynamics Society, St.
Gallen, Switzerland.
Kumpulainen, K. & Mutanen, M., 1999. The situated dynamics of peer group interaction: an
introduction to an analytic framework. Learning and Instruction, 9(5): 449-473.
Lane, D.C. 1992. Modelling As Learning: A consultancy methodology for enhancing learning in
management teams. European Journal of Operational Research, 59: 64-84.
Lane, D.C. 1995. Trying To Think Systematically About 'Systems Thinking’. Journal of the
Operational Research Society, 46: 1158-1162.
Limon, M., & Mason, L. (Eds.) 2002. Reconsidering Conceptual Change: Issues in Theory and
Practice. Dordrecht: Kluwer.
Mackinson, S. & N@ttestad, L. 1998. Combining local and scientific knowledge. Reviews in Fish
Biology and Fisheries, 8: 481-490.
Mayer, R. E. 2002. Understanding Conceptual Change: A Commentary. In Limon, M., & Mason, L.
(Eds.). Reconsidering Conceptual Change: Issues in Theory and Practice: 101-111. Dordrecht:
Kluwer.
Mercer, N., 2004. Sociocultural Discourse Analysis: Analysing Classroom Talk as a Social Mode of
Thinking. Journal of Applied Linguistics, 1: 137-168.
Metcalf, S. S. et al. 2010. Sharing the Floodplain: Mediated Modeling for Environmental
Management. Environmental Modelling & Software, 25: 1282-90.
Neubert, S., Komm, M., Krumsiek, A., Schulte, A., Tatge, N., & Zeppenfeld, L. (2011). Agricultural
development in a changing climate in Zambia. Increasing resilience to climate change and
economic shocks in crop production (Vol. 57). Bonn: Deutsches Institut fiir Entwicklungspolitik.
Parnafes, O., Hammer, D., Louca, L. et al. 2008. How to Study Learning Processes? Reflection on
Methods for Fine-Grain Data Analysis. Proceedings of the 8"" International conference for the
learning sciences 3: 306-313.
Paroutis, S., Franco, L.A. & Papadopoulos, T., 2015. Visual Interactions with Strategy Tools:
Producing Strategic Knowledge in Workshops. British Journal of Management, 26: S48-S66.
Roepstorff, A. 2008. Things to Think with: Words and Objects as Material Symbols. Philosophical
transactions of the Royal Society of London. Series B, Biological sciences, 363: 2049-2054.
Roschelle, J., 1992. Learning by Collaborating: Convergent Conceptual Change. Journal of the
Learning Sciences, 2(3): 235-276.
Rouwette, E. A.J. A., & Vennix, J. A. M. 2006. System dynamics and organizational interventions.
Systems Research and Behavioral Science, 23(4): 451-466.
Saldarriaga, M., Nyanga, P., & Kopainsky, B. 2014. Dynamic decision making in coupled social-
ecological systems. Smallholder farmers’ goals, resources and constraints in improving food
security and adapting to climate change in Zambia. Paper presented at the 32nd International
Conference of the System Dynamics Society, Delft, The Netherlands.
Scott, R.J., Cavana, R.Y. & Cameron, D., 2014. Mechanisms for Understanding Mental Model
Change in Group Model Building. Systems Research and Behavioral Science.
Spicer, J. (2015). Representation and Dynamic Implications of Mental Models of Food Systems.
Poster presented at the 33 International Conference of the System Dynamics Society, July 19-23,
2015, Cambridge, MA
Stephens, E. C., Nicholson, C. F., Brown, D. R., Parsons, D., Barrett, C. B., Lehmann, J.,... Riha, S.J.
(2012). Modeling the impact of natural resource-based poverty traps on food security in Kenya:
The Crops, Livestock and Soils in Smallholder Economic Systems (CLASSES) model. Food Security,
4(3), 423-439.
Sterman, J. D. 2001. System Dynamics Modeling: Tools For Learning In A Complex World. California
Management Review, 43: 8-25.
Suthers, D.D., 2005. Collaborative Knowledge Construction through Shared Representations.
Proceedings of the 38th Annual Hawaii International Conference on System Sciences.
Suwarno, A., et al., 2009. Participatory modelling to improve partnership schemes for future
Community-Based Forest Management in Sumbawa District, Indonesia. Environmental Modelling
& Software, 24: 1402-1410
van Boxtel, C., van der Linden, J. & Kanselaar, G., 2000. Collaborative learning tasks and the
elaboration of conceptual knowledge. Learning and Instruction, 10: 311-330.
Vennix, J. 1996. Group Model Building — Facilitating Team Learning Using System Dynamics.
Chichester: Wiley.
Wolstenholme, E.F., 1999. Qualitative vs quantitative modelling: the evolving balance. Journal of
the Operational Research Society, 50(4): 422-428.
