2013 wp evidence-creation-through-knowledge-integration_ho

1 | Centre for Development Innovation, Wageningen UR
Working Paper
Creating robust evidence
through Knowledge
Integration
Wenny Ho
The Centre for Development Innovation (CDI),
Wageningen UR focuses on the global challenges of
secure & healthy food, sustainable markets, adaptive
agriculture, ecosystem governance and conflict,
disaster & reconstruction. By linking research and
practice, we support the creation of new business
models, forms of governance and policy, and
partnerships for change that balance the needs of
people, planet and profit in a rapidly changing world.
In this Working Paper Creating robust evidence
through Knowledge Integration we share our thinking
and experiences to help facilitate innovation, create
capacities for change and broker knowledge. It is the
coming together of two strands of work undertaken by
the author in the recent past: how to bridge the
knowledge divides between policy-makers, scientists
and practitioners; and how to set up Monitoring and
Evaluation systems in organisations pressed to fulfil
a multitude of demands. The first strand relates to the
question of how to co-create the knowledge needed
to address thorny or wicked issues in multi-
stakeholder settings (Ho, 2011). The second strand
relates to the quest of organisations to remain faithful
to their values and vision while responding to the
increasingly imperious calls for evidence of the
effectiveness of their work.
This paper outlines a theory-based process of
knowledge integration in which robust evidence is a
natural outcome of purposefully sequenced multi-
stakeholder interactions.
1. Challenging times, challenging issues
As experienced by many, the world is growing more
complex. More and more, the important questions to be
solved by societies cannot be answered by thinking in
clear-cut good or bad responses. The territory is
increasingly unfamiliar and the outcomes of all but the
simplest interventions are impossible to predict.
Addressing, intervening in, let alone managing of
societal change, are becoming processes with a high
degree of complexity. Outlines of the challenges are not
clear, can change in yet unknown ways, and furthermore
influence each other in visible and invisible ways, within
timeframes that may still be unknown. Nevertheless,
while the outlines of challenges or solutions may not be
clear, there are approaches that already past the pilot
phase that can efficiently deliver a probable solution.
One of these approaches is Knowledge Integration.1
Simply said, Knowledge Integration (KI) is the process
of creating a common model or representation of a
subject by bringing together knowledge coming from
different perspectives and parties. It does so by focusing
on synthesizing the understanding of that subject from
different perspectives.
Increasing the variety of perspectives has proven to
deliver more sound knowledge. This is equally true for
profit-oriented businesses as well as for governmental or
civil society organisations looking for other horizons.
Hence, by bringing together knowledge from different
sources, KI can produce data and information that can
withstand scrutiny by different stakeholders. Scrutiny
not merely of the data and information, but also of the
1
Knowledge Integration is preferred to Knowledge Co-Creation
as the latter relates more to developments in the business sector
such as value co-creation, user-driven innovation, co-production
with consumers. See also: Ho, 2011.

2 | Creating robust evidence through Knowledge Integration
way they have been produced.2
This is essential with
societal challenges where besides soundness and
accuracy, credibility of the knowledge produced is a
fundamental asset.
So, what are the principles and approaches of
Knowledge Integration, and how can KI be linked to
other elusive processes, such as organisational learning,
for it to contribute to the creation of more robust
evidence? This note deals with these two questions by
outlining first some key principles for knowledge
integration (section 2), before describing the steps and
loops of a general KI process.
KI may still be unfamiliar, but the methods and
instruments used in KI approaches have passed through
various cycles of try-out, albeit not always in similar
combinations or in the same sector.
Various handbooks are available that provide guidance
to the usage of methods and tools. Examples are conflict
resolution tools, meta-reviews or participatory appraisal
methods. Using these endows a KI process with tried
and tested stepping stones. It allows one to concentrate
on the more unfamiliar aspects of KI, often lumped
under ‘fingerspitzengefühl’ or gut feeling. The unpacking
of these aspects is the topic of section 3 of this note: the
logic of how to lay out those stepping stones in order to
create more robust evidence in a complex situation.
2. Some key principles for a Knowledge
Integration process
Key principles
Combining as a must-have skill
Increasing diversity of perspectives and insights is one
proven way to strengthen validity and credibility of
knowledge produced. However, the art of combining
does not end with synthesising the different perspectives
on a certain subject. Key to successfully dealing with
complex interventions and situations is setting up
purposeful combinations of other variables3
notably
methods, disciplines, intervention levels and
stakeholders along with the wisdom of knowing when to
adjust a particular mixture.
2
This is sometimes called socially robust knowledge.
3
This is repeatedly mentioned in the literature, for example,
Regeer and Bunders, 2009; and the 3M approach
(http://www.grpi.org/mdo.php). The former also provides an
overview of different methods. Both sources provide some
detailed guidance how to go about this.
Crossing boundaries
With the increasing complexity of issues at hand and the
resulting need for knowledge of different kinds and from
different sources, multi-party working practices such as
collaboration, alliance-building, networking, and
interactive platforms, gain importance. With these, ways
to bridge differences, manage tensions and connect the
different parties involved take centre stage. Boundary
work refers to activities designed to do just that: smooth
the way in inter-organisational dynamics. A strategy
meeting bringing different stakeholders together to
discuss what elements should be part of a shared
strategy is one example of a boundary activity.
Boundary actors are organisations or individuals with
the skills and language to connect different parties.
A facilitator trained in managing differences (e.g. in
discipline, culture, interests, power bases) between
stakeholders, can act as a boundary actor. Boundary
objects are objects that are accepted as valid and
legitimate across the involved parties. Examples are a
jointly created protocol or a map. All three, boundary
work, actors and objects, help to make boundaries
between stakeholders less definite, more fluid, and
easier to cross and bridge. They therefore contribute
greatly to the success of knowledge integration
processes. KI processes by definition involve multiple
actors. In consequence, with KI there is an explicit need
to invest in boundary work, actors and objects, and be
mindful at every stage of how to use and combine them.
One important skill has to accompany this emphasis on
crossing and bridging boundaries: the skill to discern,
among others, the relative weight of differences and
their changing nature.
Deal explicitly with power relations
Knowledge is closely related to power relations (Hivos,
2012). The link between power and knowledge
determines not only whose knowledge is in the market,
but also whose knowledge silently goes missing.
Awareness of power issues is therefore important for
knowledge integration as, apart from anything else, they
may negatively influence the robustness of knowledge
produced (see section 3). Strategies may be needed
from the onset to deal with the often large (and often
hidden) asymmetries of power felt by stakeholders. In
particular, marginalised groups may need support to
become involved (co-)managers of impending change
processes. This entails going beyond so-called
participatory activities where, for example, participants
are requested to gather or provide information.
Experiences have shown that linking knowledge
interventions with empowerment strategies to ‘level the
playing field’ leads to greater chances of success, as
approaches use a richer knowledge base.

Let the path emerge
Knowledge integration approaches are complex
interventions: they do not offer a protocol, yet, as
mentioned earlier, require responding to emerging
situations by combining, among others, methods,
intervention levels and stakeholders. These combinations
are moreover not fixed over time, but will need to
change during the process. Additionally, KI approaches
are used in complex situations, where multiple actors are
to address an issue which relates to more than one
discipline. In such cases, complexity theory highlights the
importance of following an emergent approach in design
and implementation. Simply stated, instead of
attempting to impose and follow through a pre-designed
course of action, an emergent approach adjusts the
course of action as the path to a goal gradually reveals
itself. Beware: this does not imply a ‘do as you like’ or a
‘laissez faire’ approach. An emergent approach is
characterised by, after a light scan, probing first, next
making sense of the effects, and then, based on that,
responding to emerging patterns by adjusting or zooming
in and probing further. Undertaking these steps
purposefully distinguishes it from a laissez-faire
approach. Taking these steps in a transparent and
systematic way increases the robustness of both the
process and the products to which it leads. How to
undertake this tightrope walk of phasing a KI process
without falling into the trap of prescribing next steps is
the topic of the following section (section 3).
3. Knowledge Integration for evidence:
the art of creating one out of two helixes
As mentioned in section 1, an emergent KI process
consists of purposefully combining and recombining
known and unknown elements. It is important to
underline that despite these unknown elements, even
an emergent KI approach can follow a deliberate
sequencing. In fact, this deliberate sequencing helps
unpacking and sorting out the different components of
the unknown. Contrary to a laissez-faire approach, here
sequencing takes place in an intelligent, intentional
way. To achieve this, two helixes that draw from
different theoretical streams are merged together to
design and monitor phases of KI processes:
1. The organisational learning helix: Single-, and
double-loop learning processes;
2. The organisational change helix: Freeze-rebalance-
unfreeze.
The word ‘helix’ is used here to describe a winding
process that progresses through steps that may feel
alike, but that do not lead to a same returning point.
This distinguishes it from a cycle or circle. A helix is
hence made up by steps that are sequenced together
into non-overlapping windings. The next paragraphs
first deal with the steps in the organisational learning
helix and with those in the organisational change helix,
before merging the two helixes together to create
windings that form one purposefully sequenced and
theory-rooted KI process.
The organisational learning helix: single- and
double-loop learning processes
The single- and double-loop learning model4
had and
still has a great impact on the thinking on learning. In
this model two types of learning are distinguished.
In single-loop learning, individuals, groups or
organizations modify their actions according to the
difference between expected and obtained outputs.
Questions that typify single-loop learning are: ‘Have we
done what we planned to do, and did that go well?’
Single-loop learning relates to the level of results or
outputs based on doing things right.
Sometimes however, this may result in doing the wrong
things right. In double-loop learning this is addressed
by questioning, analysing and modifying the values,
assumptions and policies that led to the actions in the
first place. It relates to the much deeper questions of
‘Is it okay what we planned to do, and why?’, based on
doing the right things. Double-loop learning is therefore
4
This working paper focuses on the terms, first coined by
Argyris and Schön (1978), of first- and second-loop learning.
Conceptualisations of triple-loop learning are not included here.
Practical explanations of first- and second-loop learning can be
found on many websites, including
https://sites.google.com/site/reflection4learning/double-loop-
learning

Figure 1: a visual representation of single- and double-loop learning
(source: www.selfleadership.com, consulted 16th
June 2012)
learning about the validity of what underpins single-loop
learning. Single- and double-loop learning complement
each other.
A helix where the windings are composed of single- and
double-loop learning steps can be constructed as follows
(see also figure 1).
a. Single-loop learning:
i. A first step focuses on piloting: based on
previous experiences or ideas, we try something
out. A simple corresponding intervention theory5
is ‘if we do A, B will happen’.
ii. A second step then aims to identify and
incorporate lessons to strengthen theoretical
understandings: we look at what happened
followed by a reformulation or further
concretisation of the intervention theory based on
what we learnt from solving the problems
encountered. The refined intervention theory
becomes ‘if we do A and A1, then B will happen’
or ‘if we do A and avoid A2, B will happen’.
iii. A third step improves practice by incorporating
these lessons.
b. Double-loop learning: by looking up from our project,
we can question what we are doing. It is important to
become aware of the narrow ‘reality’ of that one
particular project. Questions related to a double-loop
intervention theory are ‘should we be doing A?’ or ‘Is
it just to aim for B?’ Meta-studies are an example of
‘looking up’. Comparative analyses of programmes
5
Note that the term ‘theory’ is here used as an idea or
explanation, that may or may not yet be (partially) proven, but
which takes into consideration earlier explanations and lessons
learnt so far. By making these thoughts or ideas explicit, a
theory can justify and guide action in a practical way.
across countries are other examples.6
They present
ways to identify generic influencing factors or
formulate generic principles. Comparing intervention
approaches at the meta-level can also lead to more
general frameworks which can then guide towards
best practice.
By knowing more about the nature of learning and
aim for an intentional interweaving of single- and
double-loop learning processes, a helix of
organisational learning can be created. The helix
closely connects learning with action, because it
alternately injects moments of identifying,
explaining and incorporating learning in the action
(practice) and of reflecting about the action (theory).
In this way, the organisational learning helix leads to a
strengthened and expanded theoretical and practice
base of organisations.
The organisational change helix: freeze-rebalance-
unfreeze sequences
The freeze-rebalance-unfreeze helix originates from the
literature on organisational change (Weick and Quinn,
1999). This second helix applies especially to complex
situations where change can rarely be planned
effectively. It builds around processes of joint sense-
making and interpretation of information that follow the
sequences of freezing-rebalancing-unfreezing. To freeze
is to make patterns visible, for example, by collectively
creating a story of change or undertaking a joint review
process. To rebalance is to reinterpret, re-label and re-
sequence the patterns, so that blockages are better
understood and overcome. To unfreeze is to resume the
process or action in ways that are now more mindful
because of the incorporation of lessons and insights
gained in the rebalancing7
.
In this way, the freeze-rebalance-unfreeze helix
connects learning with sense-making and change. These
are three important processes when working on not
well-definable issues in a changeable environment.
Connecting these processes in a systematic fashion can
make an intervention strategy more powerful. Moreover,
by making the sequences of particularly freezing and
6
This differs notably from the term broadening used in
Transition theory to indicate the repetition of an experiment in
different contexts (see, for example, Loorbach and van Raak,
2006). Here the purpose is to strengthen, for example, the
internal and external validity of evidence by roping in different
sources and audiences, comparing findings from different
contexts and sources.
7
David Kolb’s much used Action Learning model puts these
steps s in a context of action research. Kolb’s action learning,
however, is limited to single-loop learning.

rebalancing collective, the change process can be
consciously designed to stimulate communication
between stakeholders inside and outside an
organisation. Similarly, interactions between different
intervention levels can be encouraged, for example, by
creating multi-stakeholder platforms where local
communities can interact directly with national or
international groups. Both ways, between groups and
between levels, contribute to bringing knowledge from
diverse sources together and synthesising them to a
common understanding. In short, they are conducive to
knowledge integration.
In the next section, KI will be further embedded in the
key organisational processes of learning, sense-making
and changing by merging together the two helixes of
learning and changing.
Merging together the helixes to produce evidence
One simple definition of evidence is that it is something
that furnishes proof 8
.Four criteria are commonly used
for judging the robustness of evidence: relevance,
corroboration, veracity and validity (Rieper et al.,
2012: 2)9
. In a way, these criteria are relative; for
example, what is relevant depends on one’s viewpoint,
and how much is acceptable can be part of a multi-
stakeholder negotiation. In consequence, standards
8
http://www.merriam-
webster.com/dictionary/evidence?show=0&t=1369394430,
accessed 24th
May 2013.
9
Rieper et al. (2012: 2):
- Relevance: in relation to a given assertion
- Sufficiency: in sense of corroboration with other instances of
the same kind of evidence or other kinds of evidence
- Veracity: the process of gathering evidence has been free
from distortion and as far as possible uncontaminated by
vested interests
- Validity: internal validity refers to how true inferences are
regarding cause-effect or causal relationships; external
validity refers the degree to which the conclusions in your
study would hold for other places, persons, times etc.
need to be agreed depending on the demands and other
situational factors.
Knowledge can also be graded in terms of the four
criteria. As mentioned, bringing together different
perspectives, which is the focus of a Knowledge
Integration approach, can generate more robust
knowledge. This happens, for example, by corroborating
a previously isolated experience or enhancing the
external validity of a conclusion reached under one
particular set of conditions through methodical sharing
or consulting with a set of selected stakeholders. When
KI systematically seeks to bring together different
perspectives about how to achieve change, the robust
knowledge that is gained can be tapped to provide
robust evidence about change interventions. This
requires a process of transcription, i.e. translating the
knowledge produced into a format that can be registered
as evidence, depending on the type of evidence
required10
. The potential offered by such a KI process is
significant: connecting learning with changing can
generate both knowledge and evidence; doing this
systematically can generate knowledge and evidence
that are both more robust, especially when the desired
change concerns complex societal issues.
Bringing together the earlier mentioned helixes of
Learning (Single- and Double-Loop Learning) and
Changing (Freeze-Rebalance-Unfreeze) helps to create a
systematic theory-rooted Knowledge Integration
process. This happens by – crudely speaking – cutting
the helixes into pieces (the steps) and then merging
these together into one KI-helix. How many steps are
used of each original helix and how, in what order, they
are put together to create the windings of a KI-helix
depends on the situation and issue at hand. This is the
unknown element.
Yet, the unknown does not equal unbounded. The
boundaries of the KI process are set by the definition
and delimitation of the four mentioned criteria for
robustness: relevance, corroboration, veracity and
validity. There are no fixed or absolute values of what
each constitutes. For instance, relevance depends on the
interests of involved stakeholders (and sometimes on
the openness of financing agencies). Imagine a three-
year project in which six windings of the organisational
10
According to Davies (2004. In Rieper et al. 2012: 175) there
are many types of evidence including impact evidence,
implementation evidence, descriptive analytical evidence,
statistical modelling, economic evidence, and ethical evidence.

Figure 1: a visual representation of single- and double-loop learning
(source: www.selfleadership.com, consulted 16th
June 2012)
change helix are planned, each with the three sequences
of freeze-rebalance-unfreeze. In that helix, ‘freezing’ of
activities undertaken and results gained could happen
through the writing of half-yearly reports. Those reports
may result in an adjusted half-yearly action plan
(rebalancing). The same can also be achieved by
organising a staff retreat or community review (freezing)
during which corrections or innovations (rebalancing)
are introduced. While in the three years there may be
six windings of the organisational change helix, there
might be just one in the Learning helix. This is, for
example, the case when in those three years no mid-
term review is planned and only a final evaluation. The
latter may be designed like a meta-study that brings
together the results from three similar country projects.
In this case, only single-loop learning is encouraged until
the end-term evaluation, when the aim is double-loop
learning.
Note that the helixes of Learning and of Changing are
both built up by mainly tried and tested elements. Their
sequences or steps may have a similar feel, e.g. multi-
stakeholder workshops. The differences between the
helixes are of focus and emphasis. Report writing can,
for example, be a freezing activity (change helix), a
(collective) learning experience, or even be designed to
contribute to both change and learning. The emphasis at
a certain stage and the focus with which it is undertaken
are leading: if the focus of a writing exercise is on
freezing, then that determines its design.
It is in the art of merging together those known
elements to form the windings of a KI helix that the
unknown element resides. Unfortunately, there exist no
step-by-step recipe for this, only guidelines similar to
‘let the sauce brown until you like it’ in a cookbook.
Nevertheless, let’s try to unpack the unknown element
by looking at an imaginary example of a KI process. The
example illustrates the creation of possible KI windings
out of steps from the two helixes. Different ways of
sequencing the steps or using other tools or instruments
are of course possible. Also, as mentioned earlier, in
Knowledge Integration the steps are not limited to one
organisation only. Therefore, the different steps of both
the learning and changing helix can be designed as
collective exercises, involving changing sets of multiple
parties depending on the objective of a specific step. For
example, enhancing internal validity asks for the
involvement of a different (sub) group of stakeholders
than validating the relevance of data. The central issue
here is to be able to explain the For what and Why of a
certain sequence or Why not if occasion arises.
The imaginary example is represented in figure 2. To
better visualise how the two original helixes are put
Figure 2: merging together the two theory-rooted helixes (Ho, 2013).
©Hivos

together, different colours are used to represent the
sewn-together helix of the KI process:
- For the learning helix: yellow for the single-loop
sequences, and orange for the double-loop part
- For the organisational change helix: green for the
Rebalance sequences, light blue for the Freeze
and darker blue for the Unfreeze sequences.
This imaginary example of a KI process is sequenced as
follows:
1. Creation of a first single-loop learning winding:
a. Scanning – this scan can cover, among others,
existing literature or other experiences or
projects, and addresses the following questions:
i. for programme XXX, what are tried and tested
concepts, methodologies and instruments;
ii. what can be the added value of these
concepts, methodologies and instruments to
bring about the changes aimed for in
programme XXX;
iii. what characteristics of the environment need
to be taken into consideration when using
these concepts, methodologies and
instruments.
b. Designing the pilot – formulating working
hypotheses for the pilots of programme XXX
(theory of change and theory of intervention).
Depending on the degree of, for example,
geographical variation, a decision on the number
of pilots is taken. The higher the variation, the
more pilots are needed
c. Implement pilots programme XXX.
d. Single-loop learning – day-to-day, weekly or even
monthly adjustments of what is implemented
according to the plan. This can take place at an
individual level or during operational team
reviews. A pattern of cumulative adaptations
emerges.
2. Transition to a first double-loop learning winding by
delving deeper and merging learning with the
organisational change helix:
a. Freeze – momentary stop of the implementation
process to identify and make explicit certain
patterns and create a narrative of what happened.
An example of a narrative can be ‘This has taken
place, because...’, or: ‘As that happened, we have
adjusted...’. Besides a reflection on results, the
validity of the hypotheses and resulting policies
formulated in the pilot phase are analysed in
relation to the adaptations introduced during the
implementation and the effects observed. Values
underlying the hypotheses are also questioned.
This analysis or questioning happens therefore at
two levels:
i. articulation of lessons learnt in relation to
hypotheses: these lessons lead to the creation
of stronger case studies (and result in a
strengthened empirical base). It is not
sufficient to have a simple story of change:
the case studies should be used to contribute
in the rewriting or further concretising of the
hypotheses formulated at the beginning. Also,
the case studies provide the embedding of a
story, a narrative that links the hypotheses
with (past and future) change. Articulation at
this level ensures that the team does not get
side-tracked. It, however, should leave room
for lessons about unexpected events or effects
to avoid a myopic freezing process.
ii. articulation of lessons learnt in relation to the
knowledge base obtained so far, that is the
lessons are compared with what emerged from
the first scan of literature or other
experiences. This may seem tiresome and
undo-able in ‘real-life’, a complaint often
uttered by practitioners. However, this
particular step provides some kind of
guarantee that the lesson learnt is not just
rediscovered. In fact, this step is essential to
strengthen and advance the theoretical
foundation that can underpin social change
interventions.
To contribute to Knowledge Integration, this step is
made into a shared exercise. This can be achieved, for
example, by collectively creating a story of change. In
the shared freezing sequence, the different knowledge
perspectives of involved parties are brought together.
Not all elements of the freezing need to be undertaken
jointly. Areas to be analysed can, for example, be
divided among groups or group members. However, it is
important that collective moments are built in as well. In
this manner, a joint freezing contributes to the required
synthesis of different perspectives for a common
knowledge base with all its associated benefits. For
instance, through the design and timing of these shared
activities, addressing, among others, corroboration or
veracity these can be made more robust.
b. Rebalance – While freezing entails looking at the
past (for example, to create images and stories of
what has happened, during rebalancing the focus
shifts towards the future. Based on the outcomes
of the freezing step, the earlier ideas are
reformulated about what needs to be done and
how, based on the insights in why and towards

which objectives. This is not a new invention.
However, it is easy to just jump through the hoop
and keep on scratching at the level of activities.
To go deeper and encourage double-loop learning,
the reformulation or rebalancing should be done
at two levels: at the level of hypotheses and of
the knowledge base. Rebalancing can be done as
exhaustively as the means available (time,
participants, interest etc.) will allow, as long as
the central objective is kept in mind: reduce the
number of blockages by re-labelling and re-
sequencing at the level of hypotheses and of
approaches.
3. Creation of a second single-loop winding by merging
with the organisational change helix: unfreeze –
furthering practice through approaches that are
reformulated to help unfold events in more
meaningful ways. These approaches incorporate the
double-loop learning that has been achieved through
freezing and rebalancing.
In the case of programme XXX, in a second project
phase, the focus may change from piloting to an up-
scaled level of implementation, with an adjusted and
refined theory of change or theory of intervention.
Other single-loop sequences can follow, e.g. monthly
community monitoring or report writing.
4. Creation of a second double-loop winding by merging
learning with the organisational change helix:
a. Freeze – during this second momentary stop
again patterns are identified and made explicit
and a narrative of what happened is created.
Depending on the experience bases that have
been built so far in programme XXX, freezing can,
for example, take the form of collective analysis
and reflection across a number of pilots. This
means that the articulation covers again the two
levels:
i. articulation of lessons learnt in relation to
hypotheses, and systematization of these
lessons across pilots in order to strengthen the
base from which conclusions can be drawn.
ii. articulation of lessons learnt in relation to the
knowledge base (from the scan and the first
freeze); and at a meta-theoretical level to link
the insights, for example, from different
disciplines.
Care has to be taken to delimit and define that
knowledge base, especially considering the issue
of external validity (Cartwright and Hardie,
2011). Pilots with a strong external validity need
not produce exactly the same results. Cartwright
and Hardie therefore promote the idea of looking
at what matters to adjust and fine-tune, for
example, a theory of intervention or a theory of
change, to define the knowledge base.
b. Rebalance – again the focus moves from the past
towards the future by reformulating ideas
(hypotheses, policies etc.) about what needs to be
done. The objective of this second rebalancing
remains the same: encourage double-loop learning
at the level of hypotheses and approaches so that
the change process can restart from a
reformulated, reinvigorated basis. As mentioned,
building in collective activities contributes to KI
and can make the process more robust.
5. Creation of the third single-loop winding: unfreeze –
furthering practice through approaches that
incorporate the double-loop learning that has been
achieved through freezing and rebalancing.
Note that in the above example, the meta-level is
reached only in the second double-loop (step 4) where
lessons across pilots and at a meta-theoretical level are
formulated. This activity can be moved forward or
backwards in time depending on the situation
(requirements and resources available, degree of
robustness to be achieved). However, it is important to
arrive at this meta-level at some stage by building in,
for example, a systematic review or a meta-analysis.
At this point the team can scrutinise if and under what
circumstances certain interventions may work, and
show to others that it has not been cherry-picking.
Earlier, relevance, corroboration, veracity and validity
were mentioned as the four criteria used commonly to
judge the robustness of knowledge and evidence. Some
hints have been given in the above example of how a

higher grading on relevance, corroboration, veracity
and even validity can be achieved through a careful
design and sequencing of the KI process to
methodically involve selected parties. A higher grading
on internal validity comes, for example, from checking
the consistency of conclusions regarding results and
effects with the parties involved in a programme. Doing
so across different programmes comparing, for
example, locations, socio-economic groups etc., helps
to improve the external validity. Similarly, a design that
is mindful of[/takes into account] power issues can
boost the grading on relevance, corroboration and
veracity by building in steps to prepare minority voices
to express what is relevant to them.
4. Final remarks
A close look at the different steps shows that separately,
they are not really new. They are already being
undertaken by a range of organisations. What is novel is
the purposeful ordering of these steps, the intimate
twining of changing, learning, and sense-making, and the
systematic building-in of interactions between different
stakeholders honouring the principles of knowledge
integration. This purposeful sequencing produces lessons
that can be used as rigorous evidence, because it is
created by methodically learning from different
knowledge perspectives.
Often in what can be counted as evidence, certain types
of knowledge, such as academic knowledge, carry more
weight than subjective, experience-based knowledge. By
intentionally interjecting collective interactions with
subgroups and individual events, it is possible to create a
more even balance of what is appreciated and included.
Contrary to what may be thought, the amount of
information generated through reviews by social change
organisations, among others, is often far too much to use
sensibly and learn from it. This is often the case with big
organisations. It is even more applicable to knowledge
integration approaches when several organisations are
involved. Closely tying together changing, learning and
sense-making can make the process ‘lean and mean’
because loose data that do not make sense will quickly
be discarded or not even collected.
With a demanding and often mystifying approach like
Knowledge Integration, it can help to give the ‘beast’ a
name, for example, by using a term like sequencing, or
looking at the why of a particular winding as part of a
longer process rather than as another activity. Finding
the right wording or concept is often part of the problem
(and the solution) as it can help to put on the right lens
and provide an abstract process with manageable handles.
However, the crux is not whether one uses difficult,
abstract terms. What is key to optimally using KI for
robust evidence-creation is that one does not treat the
different processes, such as collective sense-making or
learning, as standalone pillars. The main message here is
the importance of systematically establishing and
reinforcing logical connections: between purposefully
ordered steps or activities, as well as between central
organising processes. This theory-rooted connecting
mindset provides an affordable remedy against losing track
of the logic of different steps, over-enthusiastically
undertaking a range of activities with a weak internal logic
(fuelled by the do-mode of a project-oriented
organisation), or confusion about the longer-term horizon
in a diffuse process such as knowledge integration. Using
knowledge integration in this logically sequenced and
twined way will result in a more robust way of evidence-
creation by practitioners, managers and evaluators.
To contact the author: howws@wxs.nl
References
Argyris, C. & D.A. Schön, 1978. Organizational Learning:
A theory of action perspective. Reading MA: Addison-
Wesley.
Cartwright, N. & J. Hardie, 2011. Evidence-based policy:
a practical guide to doing it better. Oxford University
Press.
Davies, P. 2004. Is evidence-based government
possible? Jerry Lee Lecture. 4th
Annual Campbell
Collaboration Colloquium, Washington DC. In: Rieper,
O., F. L., Leeuw and T. Ling (editors). 2012. The
evidence book: concepts, generation, and use of
evidence. Comparative Policy Evaluation, Volume 15.
Transaction Publishers, New Brunswick, New
Jersey: 175.

10 | Creating robust evidence through Knowledge Integration | October 2013
Centre for Development Innovation
PO Box 88
6700 AA Wageningen
The Netherlands
T +31 317 486 800
F +31 317 486 801
info.cdi@wur.nl
www.wageningenUR.nl/cdi
Hivos, 2012. Can knowledge trigger change? Briefing
note.
Ho, W., 2011. Like a bridge over troubled waters:
dialogues of policy, practitioners, and academic
knowledges. A think piece. Hivos and IKM Emergent.
Ho W., 2013. Handles on Knowledge Integration:
Guidance Notes in the Myriad of Facts and
Perspectives. Hivos.
Loorbach, D.A. and R. van Raak (2006). Strategic niche
management and transition management: different,
but complementary approaches. FSW, EUR,
Rotterdam.
Regeer B. & J. Bunders,2009. Knowledge Co-creation:
Interaction between science and society. A
transdisciplinary approach to complex societal issues.
Athena Institute, VU University Amsterdam.
Rieper, O., F. L., Leeuw & T. Ling (editors), 2012. The
evidence book: concepts, generation, and use of
evidence. Comparative Policy Evaluation, Volume 15.
Transaction Publishers, New Brunswick, New Jersey.
Weick K.E. & R.E. Quinn, 1999. Organizational change
and development. Annual Review of Psychology 50:
361-386.
This working paper has benefitted from earlier
cooperation with Hivos. While the author is the sole
responsible, observing Hivos practice and discussions
with Hivos staff have enriched the thoughts expressed in
this paper.
Please cite as
Ho, W. (2013) Creating robust evidence through
Knowledge Integration. Working Paper. Centre for
Development Innovation, Wageningen UR (University &
Research centre), Wageningen, and Hivos, The Hague.

2013 wp evidence-creation-through-knowledge-integration_ho

More Related Content

Viewers also liked (14)

Similar to 2013 wp evidence-creation-through-knowledge-integration_ho (20)

Recently uploaded (20)

2013 wp evidence-creation-through-knowledge-integration_ho