Useful Theory of Change Models

Abstract

Although frequently discussed in the evaluation literature and general agreement on what a theory of change is conceptually, there is actually little agreement beyond the big picture of just what a theory of change comprises, what does it show, how it can be represented and how it can be used. This article outlines models for theories of change and their development that have proven quite useful for both straightforward and more complex interventions. The models are intuitive, flexible, well-defined in terms of their components and link directly to rigorous models of causality. The models provide a structured framework for developing a theory of change and analysing the intervention it represents

Full text

Useful Theory of Change Models
John Mayne
Ottawa, Ontario
Abstract: Although theories of change are frequently discussed in the evaluation
literature and there is general agreement on what a theory of change is conceptually,
there is actually little agreement beyond the big picture of just what a theory of change
comprises, what it shows, how it can be represented, and how it can be used.  is
article outlines models for theories of change and their development that have proven
quite useful for both straightforward and more complex interventions.  e models
are intuitive,  exible, and well-de ned in terms of their components, and they link
directly to rigorous models of causality.  e models provide a structured framework
for developing useful theories of change and analyzing the intervention they represent.
Keywords: causal links, complex intervention, impact pathways, logic model, results
chain, theory of change, theory of reach
Résumé: Bien que les théories du changement soient souvent débattues dans la
littérature portant sur l’évaluation et qu’il y ait un consensus sur ce qu’est, concep-
tuellement, la théorie du changement, il n’y a, en réalité, aucun accord au-delà de
la dé nition générale de ce que la théorie du changement comprend, de ce qu’elle
démontre ainsi que de la manière dont elle peut être représentée et utilisée. Cet article
donne un aperçu de modèles de théories du changement, ainsi que leur développe-
ment, lesquels se sont montrés très utiles lors d’interventions tant simples que com-
plexes. Les modèles sont intuitifs,  exibles et bien dé nis en matière de composantes.
De plus, ils sont liés à de rigoureux modèles de causalité. Les modèles fournissent un
cadre structuré au développement de théories du changement utiles et à l’analyse de
l’intervention qu’ils représentent.
Mots clés: liens de causalité, intervention complexe, cheminement d’impact, modèle
de logique, chaîne des résultats, théorie du changement, théorie de la portée
Models depicting how interventions are meant to work are frequently discussed
and used in evaluation. See, for example, Patton (2008) ; Chen (2015) ; Rossi,
Lipsey, and Freeman (2004) ; Morra Imas and Rist (2009) ; and Funnell and Rog-
ers (2011) . Some of the earlier discussions are by Suchman (1967) and Bick-
man (1987) . However, the terms used to describe these models vary widely,
and include program theory , logic model , theory of change , results chain , out-
come pathway , action theory , implementation theory , and more, with no general
© 2015 Canadian Journal of Program Evaluation / La Revue canadienne d'évaluation de programme
30.2 (Fall / automne), 119–142 doi: 10.3138/cjpe.230
Corresponding author: John Mayne, john.mayne@rogers.com

120 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
agreement on terms or meaning. Funnell and Rogers (2011 , pp. 15–34) discuss
the range of terms used for these models and their histories, as does Patton (2008 ,
pp. 336–340).
I will be using the term theory of change .  eories of change have a wide range
of possible uses in developing, managing, and evaluating interventions. Mayne
and Johnson (2015) discuss using theories of change in
Designing/planning interventions
1. Designing interventions
2. Understanding and agreeing on interventions with stakeholders
3. Identifying and addressing equity, gender, and empowerment issues
4. Ex ante evaluation of proposed interventions
Managing interventions
5. Designing monitoring systems
6. Understanding implementation, managing adaptively, and learning
Assessing interventions
7. Designing evaluation questions, methods, and tools
8. Making causal claims about impact
9. Reporting performance
Scaling
10. Generalizing to the theory, to other locations and for scaling up and out.
In Part 5 of their book, Funnell and Rogers (2011) discuss using theories
of change (program theories) in monitoring and evaluation and offer many
examples. Some good examples of using theories of change, especially in a
planning and designing mode, can be found in Johnson, Guedenet, and Saltz-
man (2014) .
 e use of theories of change has been reviewed by James (2011) , Vogel
(2012b) , and Stein and Valters (2012) , who all note that while there is general
agreement on the big picture about theories of change—models depicting how
interventions are supposed to work—there is a proliferation of di erent inter-
pretations of just what in practice a theory of change entails, how to develop one,
and how to depict it.
 is article presents and describes a robust and useful model for theories of
change for simple and more complex interventions. It  rst outlines a basic generic
theory of change, followed by a discussion of causation in relation to theories of
change. Models for more complex multifaceted interventions are then presented,
along with a discussion of nested theories of change.  e article discusses three
possible useful versions of a theory of change, discusses simplifying the models,
and o ers a few comments about building theories of change. It concludes by
summarizing why the models discussed are useful.

Useful Theory of Change Models 121
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
THEORIES OF CHANGE AND IMPACT PATHWAYS
Let me  rst de ne a few key terms.  e term results is used to include outputs,
outcomes, and impacts, where impacts are the  nal outcomes a ecting well-being.
 e term intervention is used here to describe speci c activities undertaken to
make a positive di erence in outcomes and impacts of interest. It covers policies,
programs, and projects.
To understand how and if an intervention is working, we need to understand
how the activities of the intervention are expected to lead to the desired results—
both (a) the causal pathway from activities to outputs to a sequence of outcomes
to impacts and (b) the causal assumptions showing why and under what condi-
tions the various links in the causal pathway are expected to work. A variety of
terms are used in the literature to describe the causal pathways, including results
chains, logic models,
1 and impact pathways. I will use the term impact pathways.
Impact pathways describe causal pathways showing the linkages between the
sequence of steps in getting from activities to impact. A theory of change adds
to an impact pathway by describing the causal assumptions behind the links in
the pathway—what has to happen for the causal linkages to be realized. Patton
(2008 , p. 336) makes the same distinction between logic models and theories of
change: “Specifying the causal mechanisms transforms a logic model into a theory
of change.” Chen (2015) , in Chapter 3, makes a similar distinction.  eories of
change are models of how change is expected to happen ( ex ante case) or how
change has happened ( ex post case).
 ere are many ways to depict impact pathways and theories of change.
Funnell and Rogers (2011) illustrate the broad range. Figure 1 illustrates a basic
generic theory of change that has proven useful in several settings.  e sequence
of boxes in the  gure is the associated impact pathway (the results chain), which
is discussed  rst. Figure 1 is a further re nement and improvement of the theory
of change model discussed in Mayne (2014) .
2
Components of an Impact Pathway
Bene ciaries are the target groups whose well-being the intervention intends to
improve.  ese groups might be segmented by income, gender, ethnicity, and/or
geographical area. Consider as an example an intervention aimed at improving
the nutritional diets of children; the children are the intended bene ciaries. Note
that the target groups could include organizations.
In the theory of change model, the activities and results (o en labelled as
outputs, outcomes, and impacts) are depicted in the boxes:
• Activities are actions undertaken by those involved in the intervention.
• Goods and services produced are the direct outputs resulting from the
activities undertaken. In the nutrition example mentioned above, these
might be the innovative education and training material on the bene ts
of a nutritious diet. In this article, the term output is used to refer to t hese
direct goods and services.

122 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
Behaviour changes
Goods & services
produced (outputs)
External
influences
Capacity changes in
knowledge, attitudes,
aspirations, skills, &
opportunities
Reach &
reaction
Direct
benefits
Direct benefits
assumptions
Behaviour change
assumptions
Capacity change
assumptions
Reach
assumptions
Well-being
changes
Well-being change
assumptions
Activities
Timeline
Unanticipated
results
Figure 1. A Basic Generic Theory of Change
• Reach and reaction are the target groups who are intended to receive
the intervention’s goods and services and their initial reaction. In the
nutrition example, the reach group would be mothers with children
in some geographical region. Reach is important to include as a com-
ponent in causal pathways. As has been argued, “A lack of explicit
thinking about reach in logic models can lead to problems such as
narrow/constricted understanding of impact chains, favoring of ‘nar-
row and efficient’ initiatives over ‘wide and engaging’ initiatives and
biased thinking against equity considerations” ( Montague & Porte-
ous, 2013 , p. 177). I have discussed the usefulness of including reach
in Mayne (2014) .
• Capacity changes are the changes in knowledge, attitudes, skills, as-
pirations, and opportunities of those who have received or used the

Useful Theory of Change Models 123
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
intervention’s goods and services. As discussed later, all of these changes
are needed for new action to be taken.
3
• Behavioural changes are the changes in actual practices that occur, that is,
those in the target reach group do things di erently or use the intervention
products. In our example, this could be the changes in feeding practices of
mothers that occur as a result of the improved knowledge from the train-
ing.  ere typically is feedback between capacity and behavioural changes
(such as with acquiring new knowledge and skills by doing).
• Direct bene ts are the improvements in the state of individual bene -
ciaries.  ese could be such things as increased income, increased use
of health services, more productive farming, more empowerment, or, in
the example, children consuming a more nutritious diet.
• Well-being 4 changes are the longer-term cumulative improvement in
overall well-being of individual bene ciaries, such as better health,
reduced poverty, and better food security. In our example, the improved
diet would contribute to better nutritional and health status.
Note that the causal pathway model in Figure 1 explicitly does not label
the sequence of results as immediate, intermediate, and  nal outcomes (or
impacts)—a much more frequently used model, although these labels could be
added. Because these commonly used terms have little intuitive meaning, on their
own they do not provide much guidance in setting out an impact pathway and,
if used, the result is o en wasted debate about, for example, whether a result is
an immediate or an intermediate outcome. And while I had thought that outputs
was a well-de ned (as I de ne above) and widely accepted term, the recent United
Nations Development Group handbook (2011) confuses that term as well, de n-
ing outputs as goods and services or capacity changes. I am arguing that Figure1
is a more useful representation of an impact pathway than the more common
outcomes-based generic model.
External in uences are events and conditions unrelated to the intervention
that could contribute to the realization of the intended results.  ese could include
other interventions with similar aims, and/or general economic or social trends.
 ey are not part of the intervention theory of change per se. For example, in the
nutrition example, a reduction of the price of vegetables could also account for a
portion of an increase in vegetable consumption that is unrelated to the training
intervention. Industrial forti cation of foods such as sugar or  our could also
contribute to explaining an improvement in micronutrient status.
Figure 1 includes unintended e ects : positive or—more usually—negative
unanticipated e ects that occur as a result of the interventions activities and
results. If these are known possibilities they should be noted. Ex post, unantici-
pated e ects should be actively looked for. Note also that although Figure 1 looks
linear, it explicitly allows for nonlinearity via the feedback between the various
stages. Figure 1 also illustrates a timeline of when the anticipated changes can
be expected to occur. Timelines even with rough dates are useful addition to
impact pathways.

124 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
From an Impact Pathway to a Theory of Change
In developing a theory of change, the  rst step is to develop the impact path-
way. But an impact pathway, results chain, or a logic model is not a theory of
change. Only when we add the assumptions to the causal links in the impact
pathway do we get a theory of change.  e causal link assumptions shown in the
dotted boxes in Figure 1 identify what salient events and conditions have to oc-
cur for each link in the causal pathway to work as expected. What is necessary
for the causal link to work? What factors are critical to these causal processes?
For practical reasons, we only need to consider salient assumptions , that is,
those that stand out for some reason, that are striking and relevant to the situ-
ation. Others, such as the sun rising each day or a revolution not occurring,
are not relevant—although ex post, a revolution would easily explain why the
intervention did not work! Articulating causal link assumptions would entail
a mix of prior evidence, stakeholder experience, and social science theory. For
example, an assumption in the child nutrition example would be that husbands
and mothers-in-law are supportive about what children eat, letting mothers
make those decisions.
 ese causal link assumptions cover all the risks associated with the causal
link; each of the assumptions is a risk to the realization of the ToC.
5 In the nutri-
tion example, risks concerning the availability and a ordability of nutritious
food would be captured by an assumption that nutritious food is available and
a ordable.
• Reach assumptions :  e assumptions are the events and conditions
needed to occur if the outputs delivered are to reach and be positively
received by the reach groups.  ese could include such things as that
the delivery of outputs actually reaches the intended audience and the
outputs are seen as acceptable and worth considering. A key risk here is
that the reach group is not the “right” group, as in the case of the child
nutrition intervention directed at mothers when they do not in fact
make decisions about who gets what food, as well as actually reaching
all of the intended target group and not, for example, just those who
self-select.
• Capacity change assumptions :  ese assumptions are the events that
need to occur and the conditions that need to change if the outputs that
reachthe target populations are to result in changes in their knowledge,
attitudes, skills, aspirations, and opportunities, that is, their capacity
to do things di erently.  ese could include such things as the outputs
being understood, realistic, culturally acceptable, seen as useful, com-
mensurate with the prior abilities and values of the target population,
seen as relevant to the reach group, and so on.
• Behaviour change assumptions :  ese assumptions are the events and
conditions needed to occur if the changes in the capacities of the target
groups are to result in actual changes in their practices.  ese could

Useful Theory of Change Models 125
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
include such things as  nancial capacity to make the practice changes,
acceptance by others (such as peers, social, cultural and religious leaders,
family) to make the changes, the practice changes shown to be useful,
the policy or natural environment allowing the practices to be adopted,
access to needed assets and supplies, and so on.
• Direct bene ts assumptions :  ese assumptions are the events and con-
ditions needed to occur if the practice changes are to be realized as a
direct bene t to the conditions of the targeted bene ciaries.  ese could
include such things as change practices result in a net increase in income,
routine use of health services, involvement in decision-making, and so
on. In the nutrition example, there may be an assumption that the only
change in the diet is the one recommended by the training program. If
the improved practices (e.g., more vegetables) are incorporated but then
other foods are reduced, the expected bene t may not occur.
• Well-being change assumptions :  e assumptions are the events and
conditions that need to occur if the direct bene ts are going to lead to
changes in the well-being of the bene ciaries. For example, if children
consume a better diet and if they have access to basic health care and
improved sanitation, they will improve their nutritional and health sta-
tus. If as a result of the intervention women begin to play a greater role
in food consumption decisions and if the intervention is seen as success-
ful, this could contribute to a change in gender norms that empowers
women.
Note that these causal link assumptions are not descriptions of the causal link.
A description of a causal link in Figure 1 (the solid arrows) would be, for exam-
ple, that the changes in knowledge skills and so on (capacity) will result in the
expected behaviour changes in actual practices. Causal link assumptions explain
how and why the causal link works.
Bringing about changes in behaviour can be quite challenging and has been
the subject of much research. Darnton (2008) reviews much of this literature.
A typical model is the NOA (needs, opportunities, and abilities) model in
Gatersleben and Vlek (1998) . It posits that behaviour change is brought about
by motivation and behaviour control (agency). In turn, motivation results
from needs and opportunities, and agency from opportunities and abilities.
All these elements are captured in the generic theory of change (knowledge,
skills, aspirations, attitudes, and opportunities) with di erent terms, albeit not
in as much causal detail. But the research suggests that the causal package for
behaviour change needs to include each of these components. Some are what
the intervention aims to change in terms of capacity. Others would be captured
as relevant in the behaviour change causal link assumptions. In the nutrition
example, it can be safely assumed that mothers do want to improve the health
of their children (motivation) and that the intervention aims to provide the
opportunities and abilities.

126 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
It can be useful to recognize two di erent types of capacity and behavioural
changes.  e  rst are incremental or additional changes to the current state,
such as learning new techniques and skills or adopting new practices.  ese are
relatively easier to bring about than more fundamental changes , such as think-
ing about problems di erently or changing current practices. In the nutrition
example, if what is required is acquiring new food products for children, this is
an additional practice that is relatively straightforward. On the other hand, if the
practice change required is a redistribution of food among household members,
then this changes how food was distributed previously and raises power issues.
It is a fundamental change and likely considerably more di cult to bring about;
thus the associated causal link assumptions would need to be more robust and
challenging.
 e discussion so far has been in deterministic terms (e.g., an assumption
is either necessary or it is not). However, we may want to re ect the probabil-
istic nature of causality. Mahoney (2008 , p. 421) argues that “a treatment is a
cause when its presence raises the probability of an outcome occurring in any
given case.” He introduces the useful ideas of probabilistically necessary causes—
“factors that usually or almost always have to be present for the outcome to
occur”—and probabilistically su cient causes—“a cause that much of the time
on its own will produce the e ect” (pp. 425–426). For many interventions being
evaluated, these are more realistic interpretations of necessity and su ciency.
 us the causal link assumptions can be thought of as likely necessary assump-
tions, events and conditions that almost always have to occur for the causal link
to work.
Setting out assumptions for a theory of change can be confusing because
there are di erent types of assumptions associated with an intervention. In par-
ticular, in addition to the causal link assumptions discussed above, there are also
rationale assumptions that identify the underlying hypothesis or premise on which
the intervention is founded, such as the assumption that informing household
decision-makers about the bene ts of nutrition for their children will change
their behaviour and result in children getting a better diet. It would be expected
that the rationale for most interventions would be based on some prior evidence
and experience.
Figure 2 sets out the theory of change for the nutrition example.
6 A l t h o u g h
the nutrition example is not based on an actual case, see White (2009) for a discus-
sion of just such an intervention in Bangladesh.
I am arguing that in most interventions each of these components of the
generic theory of change—activities, outputs, capacity changes, behavioural
changes, direct bene ts, and well-being change, along with the associated causal
link assumptions—can be, and should be, identi ed and thought through when
developing impact pathways and theories of change.  e structure of the model
forces one to consider just how it is expected that the intended results will be
brought about: What is the causal process at work and what does it take to make
it happen?  e model is a framework for analyzing how an intervention works.

Useful Theory of Change Models 127
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
THEORIES OF CHANGE AND CAUSAL PACKAGES
 eories of change represent how and why it is expected that an intervention will
contribute to an intended result. But it is clear that rather more than the inter-
vention activities are needed; also needed is the realization of the causal assump-
tions.  e intervention activities are rarely the sole cause of a result.  e theory
of change depicts a causal package of activities plus assumptions that together
are expected—are su cient—to contribute to the intended results. Cartwright
and Hardie (2012) call these assumptions support factors : events and conditions
needed to bring about a contribution to the e ect of a cause.  e expectation is
also that the intervention activities in particular are an essential—a necessary—
part of this su cient causal package.  at is, without the intervention activities,
realization of the causal link assumptions would not be su cient to make a
contribution.  e intervention activities can then be said to be a contributory
Figure 2. A Nutrition Intervention Theory of Change
Mothers adopt
new feeding
practices
Innovative workshops
and information
External influences
•Lower prices for food
•Other staples become
more nutritious
Mothers acquire new
knowledge about
nutrition benefits and
feeding practices
Mothers reached
Children
consume a more
nutritious diet
Capacity change assumptions
•Nutrition beneits understood
•Feeding practices understood
and relevant
Reach assumptions
•Targeted mothers with young
children reached
•Approach & material seem
appropriate
Children’s
nutrition status &
health improve
Training & informing
on nutrition benefits &
feeding practices
Timeline
Unanticipated
results
•Mothers become
more empowered
•Husbands become
resentful
Behavioural change assumptions
•Mothers want to improve the
health of their children
•Mothers make decisions about
children’s food
•New practices supported by
husbands and mothers-in-law
•Nutritious food available and
affordable
Direct benefits assumptions
•Practices prove practical
•No reduction in other nutritious
food intake
Well-being change assumptions
•Children have access to health
care
r
oth
r
s
a
o
M
o
d
do
n
’
s
s
&
r
a
m
u
n
e
d
i
n
g
p
r
a
c
th
o
n
M
ot
he
o
g
d
a
a
d
Nested Theory
of Reach for
Mothers
p
t
the
ot
th
Ch
n
u
Nested
Theories of
Reach for Girls
& Boys

128 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
cause to the results. In these terms, a theory of change is a model of the interven-
tion as a contributory cause ; it is a model of the causal package showing just how
the contribution to the results are to be brought about. Mayne (2012) discusses
contributory causes and causal packages in the context of theories of causation,
and in particular INUS causes.
7
 e theory of change is a model of the contribution to and not cause per
se of the intended result, because there may be other external factors also con-
tributing to the intended results, as noted in the external in uences box. Only
if there are no external in uences at work is the theory of change a model of
causation. As with an intervention, an external in uence usually does work just
on its own, but rather as part of another causal package that might include some
of the supporting factors in the intervention causal package. External in uences
can have positive or negative e ects on the level of results attained. Depending
on the strengths of the intervention, the external in uences may explain some
or all of the observed results. Signi cant negative e ects, that is, risks that could
undermine the intervention’s theory of change, are included in the causal link
assumptions.
In probabilistic terms, we can speak of likely su cient to describe the suf-
 ciency of the intervention causal package, meaning that, in this case, the causal
package most likely produced a contribution to the observed result. To show that
the intervention is a contributory cause is to show that the intervention’s causal
package is likely su cient, and that the intervention is itself a likely necessary
element of the su cient package.
In discussing theories of change, it is useful to distinguish the ex ante from
the ex post case . Ex ante, there is a need to speak of probabilistic causes and likely
su ciency. Ex ante, one has a postulated or prior theory of change setting out the
argument that if the intervention is implemented as designed and if the assump-
tions associated with the ToC hold, then the intended contribution to the results
will be realized. It sets forth the assumed reality and complexity of the interven-
tion. It is a prediction of e ectiveness.
Ex post, you are verifying that the theory of change did occur with evidence
on the results and assumptions that were realized. When you make a causal claim,
you know which factors were at work and whether something in addition was at
play. If you conclude that the package was likely su cient, here it means that you
recognize that you may have missed something in your analysis, but that reason-
able people would conclude that the causal package was indeed su cient. Ex post
you are testing the ex ante causal hypothesis. As noted, ex post you are likely to be
able to identify if there were other external in uences at work. If there were none,
then the intervention causal package can be said to have caused the observed
result, not just contributed to it.
 e intervention itself is one among several causal factors in the causal pack-
age necessary to bring about change. In that sense, all are equal. Yet our interest
is on the intervention as an instrument of change—activities deliberately done
to get or continue change happening where adequate change was not happening

Useful Theory of Change Models 129
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
before. We can ask ex post what role the intervention played in bringing about the
changes. We may expect that at a minimum the intervention acts as a trigger to
start the causal chain. In such cases, an intervention can be said to be a principal
contributory cause. In other cases, the intervention might see itself as playing
a more modest supporting role, joining others in an already ongoing process,
enhancing a change process already underway so that better or more timely results
are achieved ( Mayne, 2008 ).
MULTIFACETED SUFFICIENT INTERVENTIONS
Although there is a lot in Figure 1 , it was referred to as a “basic” generic theory
of change.  is is because it only shows one actor undertaking activities, and the
model may su ce for many straightforward interventions. But for many, more
complicated interventions, this is generally not the case. To make a di erence, an
intervention needs to engage and work with a variety of other intermediaries—
delivery partners, governments, the private sector, and NGOs—and in uence
their behaviour.  e theory of change shown in Figure 1 identi es a possibly
wide range of causal link assumptions that need to occur if the direct bene ts
and well-being changes are to be realized. Leaving these to chance may not be an
option, and the intervention should work with relevant intermediaries, including
delivery partners, to try to make sure that the intermediaries undertake actions
to ensure (or go a long way to ensuring) that the numerous causal link assump-
tions are brought about.  ese supporting activities carried out by the intervention
actors are in addition to its main or core activities. We can thus speak of the core
intervention and the overall intervention . In an agriculture research for develop-
ment intervention, the core intervention is the research activities, while those
plus engagement supporting activities that are carried out to get the research
used constitute the overall intervention. In other cases, there are no identi able
core or main activities, and the intervention works with a variety of partners to
collectively deliver a su cient set of activities.
Typically, these causal link assumptions can cover a range of events or condi-
tions that create an enabling environment for the intervention activities to contrib-
ute to well-being.  is results in a much more multifaceted overall intervention
but with the aim of ensuring that it is su cient : that the collection of (core) inter-
vention e orts, its engagement activities, and the resulting actions by intermedi-
aries are su cient to contribute to the expected bene ts and well-being changes.
 at is, the set of engagement activities are aimed at ensuring that the causal link
assumptions—the support factors—are realized.
We can still ask if the core intervention was a principle contributory cause,
that is, did it play a trigger role in getting change started. And in the multifaceted
su cient contexts, the intervention will also involve other subsequent support-
ing actions taken along the causal pathway to sustain the causal pathway.  us,
we would like to assess whether the core intervention is a triggering contributory
cause and a sustaining contributory cause.

130 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
A strong causal claim about a multifaceted su cient intervention would be
that the intervention was a principal contributory cause of the relevant observed
results.  at is,
 e intervention was a necessary component of a package of causal factors that to-
gether were su cient to contribute to an observed result. In other words, the interven-
tion made a di erence . In addition, the intervention played a key role; it was the trigger
that initiated the chain of events and through its supporting activities sustained the
chain of events that contributed to the observed results.
Figure 3 illustrates the generic theory of change for this more complex,
indeed multifaceted su cient intervention . 8
In building a ToC, it can be useful to identify the degree of control one has
or might have over the causal link assumptions. Assumptions can be labelled as
[O], over which the intervention has no or very little in uence; [I], where the
intervention can (should) have an in uence, direct or indirect; or [C], where
the intervention should be able to directly control.  is helps to identify where
additional supporting actions might be useful to better ensure the assumptions
are realized and hence the risks to the intervention minimized, perhaps leading
to a multifaceted intervention.
Using the nutrition example, Table 1 illustrates the type of ex ante causal link
analysis that can be undertaken. Each of the assumptions in Figure 2 is assessed
Figure 3. A Basic Theory of Change for Multifaceted Su cient Interventions
Reach &
reaction
Capacity changes
Behaviour changes
Well-being changes
Direct benefits
assumptions
Behavioural
change
assumptions
Capacity
change
assumptions
Reach
assumptions
External
influences
Engagement activities and outputs
Direct benefits
Well-being
change
assumptions
Timeline
Core activities and outputs
Timeline
Reach, Capacity change
Behavioural change of intermediaries

Useful Theory of Change Models 131
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
Table 1. Analysis of Nutrition Intervention Causal Link Assumptions
Causal link assumptions Degree of
control Supporting actions needed
beyondcore activities
A1 Reach Assumptions
• Targeted mothers with
young children reached Medium [I] Intervention needs to know its target
population and how to reach them.
Action: Likely requires outreach e orts.
• Approach and material
seems appropriate High [C] Requires good planning and knowing
the speci c context.
A2 Capacity Change Assumptions
• Nutrition bene ts
understood High [C] Requires good planning and knowing
the speci c context.
• Feeding practices
understood and relevant High [C] Requires good planning and knowing
the speci c context.
A3 Behavioural Change Assumptions
• Mothers want to
improvethe health of their
children
n/a Can be assumed.
• Mothers make decisions
about children’s food Unknown Would require knowledge of the
speci c context.
• New practices supported
by husbands and
mothers-in-law
Low [I] Action: Need for engagement with
husbands/mothers-in-law on need for
better nutritional diets for children.
• Nutritious food
availableand a ordable High [C] A prerequisite for the intervention.
Ifnot likely available, need a
di erenttype of intervention such
assubsidies.
A4 Direct Bene ts Assumptions
• Practices prove practical Medium [I] Action: Could require monitoring to see if
practices do prove practical in the speci c
context.
• No reduction in other
nutritious food intake High? [C] Should be part of the training: don’t
stop consuming other nutritious food.
But, risk exists that husbands and
mothers-in-law in the poor households
will insist on substituting.
Action: Need to engage with husbands/
mothers-in-law.
A5 Well-being Change Assumptions
• Children have access to
health care ?? [O] Would probably just be assumed. If
health is a major problem, then might
question the intervention.

132 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
Figure 4. A Multifaceted Nutrition Intervention
An
Mothers adopt
new feeding
practices
Innovative workshops
and information
External influences
•Lower prices for food
•Other staples become
more nutritious
Mothers acquire new
knowledge about
nutrition benefits and
feeding practices
Mothers reached
Children
consume a more
nutritious diet
Children’s
nutrition status &
health improves
Training & informing
on nutrition benefits &
feeding practices
Time line
Unanticipated
results
•Mothers become
more empowered
•Husbands become
resentful
A1
A4
A
A3
A5
A2
Engagement
with NGO
Husbands /
mothers-in-
law
appreciate
the need for
nutritious
diets
NGO
engages with
husbands/
mothers-in-
law
Research and outreach
efforts undertaken to
well-identify target
groups and best
means for engaging
with them
NGO establishes
appropriate M&E
on practices tried
Intervention activities
Efforts supporting an
assumption
Causal link
assumptions
Legend
Nested Theory
of Change for
NGO
as to the degree to which the intervention could undertake e orts to strengthen
the likelihood that the assumption will materialize.
In the example, it may be that husbands and/or mothers-in-law are not likely
to support their wives making decisions about who gets what food.  en, in order
for the intervention to work, some form of education of husbands and mothers-
in-law about the bene ts of a nutritious diet for their children is needed.  e
intervention agency may need to get others, perhaps an NGO, more accustomed
to dealing with culture and gender issues to engage with husbands and mothers-
in-law to in uence their behaviour. Figure 4 illustrates the resulting multifaceted
nutrition intervention.
Ex ante causal link analysis can also be used to a priori assess the extent
to which there is empirical evidence to support each link in a ToC. In many

Useful Theory of Change Models 133
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
cases it is likely that indeed there are supporting prior research and evaluation
 ndings that support some of the causal links, and equally it is o en the case
that such prior evidence is weak or not existing.  e latter case would suggest
that some new research be undertaken to better con rm the causal link before
implementing the initiative and/or that the assumptions for these links be care-
fully monitored as the intervention is implemented.  is type of causal link
analysis is discussed by Mayne and Johnson (2015) and by Johnson, Mayne,
Grace, and Wyatt (2015) .
As interesting, of course, is ex post causal link analysis of a theory of change,
determining the extent to which a causal link and its assumptions have occurred
and a credible causal claim be made.  is is the essence of contribution analysis
( Mayne, 2008 ; Mayne, 2012 ).
DEALING WITH MESSY INTERVENTIONS
Figures 1 and 3 could be seen as targeting one group of bene ciaries, such as chil-
dren in the nutrition example ( Figure 2 ). However, interventions o en have sever-
al target groups in mind (such as mothers and children) and/or subgroups within
a general group (such as boys and girls). For multifaceted su cient interventions
( Figure 3 ), there are usually several di erent intermediaries (governments, or-
ganizations, and partners) targeted. For these multitargeted interventions, one
approach would be to try to develop a theory of change that captures all these ac-
tivities on the various target groups and the resulting result sequences, capturing
the links among the various pathways. However, developing and setting out such
a model other than as an overview—which is helpful —can be quite challenging,
and the resulting quite messy theory of change model can become cumbersome
and hard to work with, either in terms of explaining the intervention or for help-
ing design the evaluation.
Nested Theories of Change
Instead, it would be much more useful to develop a subtheory of change for each
key target group—a nested theory of change or theory of reach 9 —recognizing that
these theories of reach may interact with each other in bringing about the desired
results. Figure 2 shows nested theories of reach (the oval shapes) for mothers and
for girls and boys—boys might be treated di erently than girls, and having theo-
ries of reach for each would ensure a focus on these di erences.
Figure 4 identi es the nested theory of change for NGOs in the nutri-
tion example, which is illustrated in Figure 5 .  e assumptions in the NGO
theory of change ( Figure 5 ) are ones that the NGO should be able to control or
strongly in uence. Figure 6 illustrates nested theories of reach for the generic
multifaceted su cient intervention. Nested theories of change o er a way to
break down a more messy theory of change into something more understand-
able and practical.

134 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
Figure 5. Nested NGO Theory of Change
Support for mothers
managing children’s
diets
Engagement material
and information
Changes in knowledge
& attitudes toward
nutrition & diets for
children
Husbands and
mothers-in-law
reached
Behaviour Change Assumptions
•No loss of prestige seen
•Beneits seen from mothers
deciding on diets
Capacity Change Assumptions
•Support from community
leaders and peers
Reach Assumptions
•Engagement is culturally
sensitive
NGO activities
Intervention
engaging with NGO
Figure 6. Nested Theories of Change
Reach and
reaction
Capacity changes
Behaviour changes
Well-being changes
Direct
benefits
assumptions
Behavioural
change
assumptions
Capacity
change
assumptions
Reach
assumptions
External
influences
Supporting activities and outputs
Direct benefits
Well-being
change
assumptions
Timeline Activities and outputs
Timeline
Reach, Capacity change
Behavioural Change
Different
Beneiciary
Nested ToC
Different
Intermediary
Nested ToC
A
A
A
AA
A
A
A
A
A
A
A causal link with
assumptions

Useful Theory of Change Models 135
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
Causal Link Analysis
Another way to handle a messy theory of change is to discuss the di erent major
causal links in the theory of change separately, such as discussing the link “get-
ting from capacity changes to behavioural changes.” In addition, discussion of, for
example, the di erent causal link assumptions can be done in an accompanying
narrative that could also provide suitable references to prior research and evalu-
ation that support the underlying assumptions, as done in Table 1 . Mayne and
Johnson (2015) illustrate this approach.
Uncertainty and Emerging Results
Interventions vary in their “messiness”—from more complicated, such as the
multifaceted su cient interventions discussed earlier, to truly complex interven-
tions exhibiting uncertainty and emergent properties. In evaluating truly complex
interventions, using evaluation for incremental learning and adapting over time
is usually suggested ( Ling, 2012 ; Mayne, 2011 , pp. 82–84; Rogers, 2011 ; Sander-
son, 2000 ). Consistent with that thinking would be developing initial theories of
change such as the ones discussed here, which are then revised and adapted as
new knowledge is acquired. Rogers (2008) and Ling (2012) discuss using program
theories/theories of change in complex settings.
DIFFERENT VERSIONS OF A THEORY OF CHANGE
FORDIFFERENT PURPOSES
 ere are clearly limits to how much detail can be usefully depicted in a theory
of change model, especially of a large and multifaceted intervention. Something
more manageable is needed, both to work with and for communication purposes.
It can be useful to have at least three versions of each theory of change.
 e  rst is a text version, describing in a sentence or two how the speci c
intervention being planned or implemented is intended to work, a theory of change
narrative .  is version explains in a straightforward manner how the intervention
is supposed to work and can identify the underlying rationale assumptions behind
the intervention.  is is the basic description or “story line” given by managers
(or politicians) when asked to describe why they think the intervention will work,
or set out in a policy-type documentation, usually entailing a few sentences. An
example might be the simple theory for an anti-smoking TV ad intervention: by
describing on TV the dangers of smoking, smokers will stop smoking. In the child
nutrition example, the theory of change narrative would be something like: “By
educating and informing mothers about the importance of a nutritious diet for
their children, mothers will change their past behaviour and seek to improve the
diets of their children.”  e rationale assumption here is that better information
will change behaviour.
 e theory of change narrative plays an important role, because it sets out
how the intervention will be publicly described and defended. It is in essence the
public theory of change.

136 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
 e second theory of change version is a simpli ed overview theory of change
to show the big picture for a multifaceted intervention.  is is especially useful
for multifaceted su cient interventions.  e overview theory of change can just
be a simpli ed impact pathway showing as relevant any nested theories of change,
along with the rationale assumptions. Figure 7 illustrates the nutrition example
with the theory of change for engaging with husbands and mother-in-law noted
in the triangle.
 e third and more detailed version of a causal theory of change is usually a
diagram model such as those shown in Figures 1 through 5 , showing the impact
pathways and the causal link assumptions details of the theory of change. Each
of these versions of a theory of change has its uses, and o en all three are helpful
to have at hand.
A further way to simplify a theory of change model by dropping “boxes”
and including their essence in the causal link assumptions—essentially rear-
ranging the causal package for the link. For many interventions, displaying all
the elements of their impact pathway or theory of change in a single diagram
can be cumbersome, resulting in a too-complex diagram of arrows and boxes.
Figure 8 shows a “simpli ed” version of Figure 1 in which the Reach and Capac-
ity Change boxes have been dropped.  is is o en tempting to do since it is the
behavioural changes that are thought to be the key outcomes along the impact
pathway.
Figure 8 still shows the essence of the impact pathway, but in developing
it as a theory of change, it needs to be remembered that the reach and capacity
change aspects are not explicitly shown. In this case, the assumptions behind
the arrows leading from activities and outputs to Behavioural Changes need to
include assumptions about reach and capacity change.  at is, the causal pack-
ages associated with each link remain intact , just positioned di erently. Ignoring
the reach and capacity change issues will signi cantly weaken the theory of
change.
Figure 7. Overview of Nutrition Intervention
Mother adopts new
feeding practices
Mother acquires new
knowledge about nutrition
benefits and feeding practices
Children’s nutrition
status & health
improves
ToC for
engaging with husbands &
mothers-in-law (Fig 6)

Useful Theory of Change Models 137
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
BUILDING IMPACT PATHWAYS AND THEORIES OF CHANGE
10
 e concept and application of theories of change can appear complicated, but
only because “theory of change” is not one thing per se.  is is similar to the con-
cept of “evaluation” that can be many things, depending on a variety of situations.
 eories of change
• are time dependent—can vary over time
• have di erent purposes
• need to recognize uncertainties and nonlinearities
• can be ex ante and ex post.
 ere are now numerous sources available for guidance on developing theo-
ries of change. Vogel (2012a) and Barnett and Gregorowski (2013) discusses theo-
ries of change in relation to research interventions.  ere is an extensive website
on theories of change at www.theoryofchange.org with references to many other
guides and relevant literature.
In getting to a robust prior theory of change, initial versions should be tested
against the logic and assumptions set out, as well as against any available evidence
from previous research or evaluations that might (or might not) support the way
the theory of change is being depicted.  is challenging of a theory of change
is what Brousselle and Champagne (2011) and Kautto and Silila (2005) discuss,
arguing the value of this type of theory of change analysis even before testing it
in the  eld against the actual results of the speci c intervention in question.  e
Figure 8. A Simpli ed Generic Theory of Change
Behaviour
changes
Outputs
External
influences
Direct
benefits
Direct benefits
assumptions
Behavioural change
assumptions
•Assumptions about
reaching target groups
•Assumptions about
bringing about capacity
change
Well-being
changes
Well-being change
assumptions
Activities
Timeline
Unanticipated
results

138 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
analysis undertaken to develop the theory of change may uncover weaknesses in
initial ideas and assumptions about how the intervention is supposed to work.
As noted at the outset, a theory of change can also be used as a framework for
designing the intervention, developing a monitoring regime, and developing an
evaluation plan.
In developing impact pathways and theories of change, several points should
be kept in mind:
•  ey are probably best developed in a participatory manner, but this is
not always possible.
• In a participatory process, one can start with a blank page and build
from soliciting views or, perhaps more e ciently, with a straw impact
pathway/theory of change that is developed by a few people and then
used as the basis for comment, challenge, and revision.
• In discussion with stakeholders, more than one version of an interven-
tion’s theory of change may emerge ( Hansen & Vedung, 2010 ; Weiss,
1997 ). It may then be useful to test both versions against reality.
• It is important to explicitly or implicitly include all the theory of change
elements.
• Developing impact pathways and theories of change should be seen as a
process, evolving over time as more insight is gained.
• Aim for a “good enough” impact pathway/theory of change, rather than
the perfect one.
•  e capacity and behavioural changes are o en key.
• To the extent possible, impact pathways and theories of change should
be based on prior research in addition to stakeholder views.
• Nested impact pathways and theories of change/theories of reach can be
quite useful, developed around the types of intervention strategies being
used and/or target groups.
•  eories of change can be displayed in a variety of ways and can be set
out at di erent levels of detail.
• Generic impact pathways and theories of change can be quite useful as
building blocks when similar interventions occur at di erent locations.
CLOSING REMARKS
Credible theories of change are essential for undertaking theory-based evalu-
ations.  e models discussed here are meant to be  exible enough to apply to
a wide range of interventions.  e article argues that the model of a theory of
change illustrated generically in Figures 1 and 3 are “useful.”  ey are useful for
several reasons:
•  e models are o en a “good enough” representation of a theory of
change and not overly complex.  ey lay the basis for a logical perfor-
mance story ( Mayne, 2004 ).

Useful Theory of Change Models 139
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
•  e models deliberately avoid explicit labelling of results along the im-
pact pathway as di erent levels of outputs and especially outcomes, such
as immediate, intermediate, and  nal outcomes.  ese output and out-
come labels have no inherent meaning and are not helpful in developing
a theory of change—indeed they o en lead to wasted debate. Rather, it
is the sequence that is important.  e goods and services, reach, capac-
ity change, behavioural change, and other labels in Figures 1 and 3 have
intuitive meaning and provide a good analytical structure for developing
a theory of change.
• Causal link assumptions can be well de ned, describing what is necessary
for the link to work, and they are front and centre in a theory of change.
•  e use of the causal link assumption boxes allows for a more straight-
forward looking representation. Otherwise, many more boxes and ar-
rows would be needed.
•  e theory of change model can be o en simpli ed somewhat by drop-
ping a “result box” and including it as an assumption instead.
• More complicated theories of change can be simpli ed by using the idea
of nested theories of change and reach to focus on key nested impact
pathways.
•  e model with assumptions as support factors links directly with the
concepts of causal packages and contributory causes, providing a rigor-
ous basis for making causal claims.
NOTES
1
 e term “logic model” is sometimes used synonymously with program theory or
theory of change ( Funnell and Rogers, 2011 ), but o en is identi ed with only the
causal pathway.  us, for example, the Canadian federal government de nes logic
model as “a depiction of the causal or logical relationships between activities inputs,
outputs and the outcomes of a given policy, program or initiative, e.g., Results Chain”
( Treasury Board Secretariat, 2012 ).
2  e main revisions were to simplify the representation of the model by dropping
explicit references to “risks,” “other explanatory factors,” and “incentives” and making
explicit reference to unanticipated results.
3 Knowledge pertains to learned information or accepted advice; attitudes focus on
beliefs, opinions, feelings, or perspectives; skills refer to mental and physical abilities
to use new or alternative practices; aspirations refer to ambitions, hopes, objectives,
or desires. Adapted from Bennett and Rockwell (1995 , p. 6).
4 Well-being is the broad term used here for the end result aimed for. Livelihood is
another term that could be used.
5 In previous articles, I had o en explicitly included “risks” in the assumption boxes,
noting that some assumptions are more easily understood and written as risks.  is
can be useful, but does clutter the boxes somewhat.
6  e nested theories of reach in Figure 2 are discussed later.
7 INUS stands for an Insu cient but Necessary part of a condition that is itself Unneces-
sary but Su cient for the occurrence of the e ect ( Mackie, 1974 ).

140 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
8 Figure 3 illustrates one type of complicated and complex intervention, with core and
supporting activities. Another (not shown) would be a multicomponent su cient in-
tervention made up of a number of quite separate and distinct component activities,
which together are expected to lead to improved well-being. Here the components
would be nested theories of change within a larger overview theory of change.
9  eories of reach are discussed in Mayne (2014) .
1 0  e points in this section are further elaborated on in Mayne and Johnson (2015) .
REFERENCES
Barnett , C. , & Gregorowski , R. ( 2013 ). Learning about theories of change for the monitoring
and evaluation of research uptake (IDS Practice Paper in Brief, ILT Brief 14: Insti-
tute of Development Studies). Retrieved from http://opendocs.ids.ac.uk/opendocs/
bitstream/handle/123456789/2995/PP%20InBrief%2014%20FINAL.pdf;jsessionid=
4E0FEAAE919534B1B2A65F654E121023?sequence=1
Bennett , C. , & Rockwell , K. ( 1995 ). Targeting outcomes for programs: A hierarchy for tar-
geting outcomes and evaluating their acheivement (University of Nebraska). Retrieved
from ftp://bsesrv214.bse.vt.edu/grisso/Program_Logic/Targeting_Outcome_
Programming.pdf
Bickman , L. (Ed.). ( 1987 ). Using program theory in evaluation (New Directions for Evalu-
ation, no. 35). San Francisco, CA : Jossey-Bass .
Brousselle , A. , & Champagne , F. ( 2011 ). Program theory evaluation: Logic analysis . Evalu-
ation and Program Planning , 34 , 69 – 78 .
Cartwright , N. , & Hardie , J. ( 2012 ). Evidence-based policy: Doing it better. A practical guide
to predicting if a policy will work for you . Oxford, UK : Oxford University Press . http://
dx.doi.org/10.1093/acprof:osobl/9780199841608.001.0001
Chen , H. T. ( 2015 ). Practical program evaluation:  eory-driven evaluation and the inte-
grated evaluation perspective ( 2nd ed. ).  ousand Oaks, CA : Sage .
Darnton , A. ( 2008 ). Practical guide: An overview of behaviour change models and their
uses. Retrieved from http://www.civilservice.gov.uk/wp-content/uploads/2011/09/
Behaviour-change_practical_guide_tcm6-9696.pdf
Funnell , S. , & Rogers , P. ( 2011 ). Purposeful program theory: E ective use of theories of change
and logic models . San Francisco, CA : Jossey-Bass .
Gatersleben , B. , & Vlek , C. ( 1998 ). Household consumption, quality of life and environ-
mental impacts . In K. Noorman and A. Schoot-Uiterkamp , Eds., Green households?
Domestic consumers, environment and sustainability (pp. 141–183). London, UK :
Earthscan .
Hansen , M. B. , & Vedung , E. ( 2010 ).  eory-based stakeholder evaluation. American
Journal of Evaluation , 31 ( 3 ), 295 – 313 . http://dx.doi.org/10.1177/1098214010366174
James , C. ( 2011 ).  eory of change review (Report commissioned by Comic Relief).
Retrieved from http://mande.co.uk/blog/wp-content/uploads/2012/03/2012-Comic-
Relief- eory-of-Change-Review-FINAL.pdf
Johnson , N. , Guedenet , H. , & Saltzman , A. ( 2014 ). What will it take for bioforti cation to
have impact on the ground?  eories of change for three crop-country combinations

Useful Theory of Change Models 141
CJPE 30.2, 119–142 © 2015doi: 10.3138/cjpe.230
(IFPRI discussion paper). Washington, DC : International Food Policy Research Insti-
tute . Retrieved from http://ebrary.ifpri.org/cdm/singleitem/collection/p15738coll2/
id/129089/rec/1
Johnson , N. , Mayne , J. , Grace , D. , & Wyatt , A. ( 2015 ). How will training traders con-
tribute to improved food safety in informal markets for meat and milk? A theory of
change analysis (IFPRI discussion paper 1451). Washington, DC : International Food
Policy Research Institute . Retrieved from http://ebrary.ifpri.org/cdm/ref/collection/
p15738coll2/id/129293
Kautto , P. , & Silila , J. ( 2005 ). Recently introduced policy instruments and intervention
theories . Evaluation , 11 ( 1 ), 55 – 68 .
Ling , T. ( 2012 ). Evaluating complex and unfolding interventions in real time. Evaluation ,
18 ( 1 ), 79 – 91 . http://dx.doi.org/10.1177/1356389011429629
Mackie , J. L. ( 1974 ).  e cement of the universe: A study of causation . Oxford, UK : Oxford
University Press .
Mahoney , J. ( 2008 ). Toward a uni ed theory of causality. Comparative Political Studies ,
41 ( 4-5 ), 412 – 436 . http://dx.doi.org/10.1177/0010414007313115
Mayne , J. ( 2004 ). Reporting on outcomes: Setting performance expectations and telling
performance stories. Canadian Journal of Program Evaluation , 19 ( 1 ), 31 – 60 .
Mayne , J. ( 2008 ). Contribution analysis: An approach to exploring cause and e ect , (ILAC
Brief No. 16).  e Institutional Learning and Change Initiative. Retrieved from http://
www.cgiar-ilac.org/ les/publications/briefs/ILAC_Brief16_Contribution_Analysis.pdf
Mayne , J. ( 2011 ). Contribution analysis: Addressing cause and e ect . In R. Schwartz , K.
Forss , & M. Marra (Eds.), Evaluating the complex (pp. 53 – 96 ). New Brunswick, NJ :
Transaction .
Mayne , J. ( 2012 ). Contribution analysis: Coming of age? Evaluation , 18 ( 3 ), 270 – 280 . http://
dx.doi.org/10.1177/1356389012451663
Mayne , J. ( 2014 ). Using theories of reach to enhance equity considerations in evaluation .
In K. Forss & M. Marra (Eds.), Speaking justice to power: Ethical and methodological
challenges for evaluators (pp. 133 – 152 ). New Brunswick, NJ : Transaction .
Mayne , J. , & Johnson , N. ( 2015 ). Using theories of change in the Agriculture for Nutrition
and Health CGIAR research program . Evaluation , 21 ( 4 ), 407–428.
Montague , S. , & Porteous , N. ( 2013 ).  e case for including reach as a key element of
program theory. Evaluation and Program Planning , 36 ( 1 ), 177 – 183 . http://dx.doi.
org/10.1016/j.evalprogplan.2012.03.005
Morra Imas , L. G. , & Rist , R. C. ( 2009 ).  e road to results: Designing and conducting e ec-
tive development evaluations . Washington, DC : World Bank .
Patton , M. Q. ( 2008 ). Utilization-focused evaluation ( 4th ed. ).  ousand Oaks, CA : Sage .
Rogers , P. ( 2008 ). Using programme theory to evaluate complicated and complex aspects of
interventions. Evaluation , 14 ( 1 ), 29 – 48 . http://dx.doi.org/10.1177/1356389007084674
Rogers , P. ( 2011 ). Implications of complicated and complex characteristics for key
tasks in evaluation . In K. Forss , M. Marra , & R. Schwartz (Eds.), Evaluating the
complex: Attribution, contribution and beyond (pp. 33 – 52 ). New Brunswick, NJ :
Transaction .

142 Mayne
© 2015 CJPE 30.2, 119–142 doi: 10.3138/cjpe.230
Rossi , P. H. , Lipsey , M. W. , & Freeman , H. E. ( 2004 ). Evaluation: A systematic approach ( 7th
ed. ).  ousand Oaks, CA : Sage .
Sanderson , I. ( 2000 ). Evaluation in complex policy systems. Evaluation , 6 ( 4 ), 433 – 454 .
http://dx.doi.org/10.1177/13563890022209415
Stein , D. , & Valters , C. ( 2012 ). Understanding ‘theory of change’ in international develop-
ment: A review of existing knowledge . Asian Institute and the Justice and Security
Research Programme. Retrieved from http://www.theoryofchange.org/wp-content/
uploads/toco_library/pdf/UNDERSTANDINGTHEORYOFChangeSteinValtersPN.
pdf
Suchman , E. D. ( 1967 ). Evaluative research: Principles and practice in public service & social
action programs . New York, NY : Russell Sage Foundation .
Treasury Board Secretariat . ( 2012 ). Results-based management lexicon . Ottawa: Author.
Retrieved from http://www.tbs-sct.gc.ca/cee/pubs/lex-eng.asp
United Nations Development Group . ( 2011 ). Results-based management handbook: Har-
monizing RBM concepts and approaches for improved development results at country
level . Retrieved from https://undg.org/wp-content/uploads/2015/01/UNDG-RBM-
Handbook-2012.pdf
Vogel , I. ( 2012a ). ESPA guide to working with theory of change for research projects . Ecosys-
tem Services for Alleviation of Poverty. Retrieved from http://www.espa.ac.uk/ les/
espa/ESPA- eory-of-Change-Manual-FINAL.pdf
Vogel , I. ( 2012b ). Review of the use of ‘theory of change’ in international development .
Department for International Development. Retrieved from http://www.oxfamblogs.
org/fp2p/wp-content/uploads/DFID-ToC-Review_VogelV4.pdf
Weiss , C. H. ( 1997 ). How can theory-based evaluation make greater headway? Evaluation
Review , 21 ( 4 ), 501 – 524 . http://dx.doi.org/10.1177/0193841X9702100405
White , H. ( 2009 ). Theory-based impact evaluation: Principles and practice (Work-
ing Paper 3). International Initiative for Impact Evaluation (3ie). Retrieved
from http://www.3ieimpact.org/admin/pdfs_papers/48.pdf . http://dx.doi.
org/10.1080/19439340903114628
AUTHOR INFORMATION
John M ayne is an independent advisor on public sector performance. Over the last 11 years
he has focused largely on international development evaluation and results-based manage-
ment work. He has been working with a number of government, NGOs, and international
organizations. He has authored numerous articles and reports, including several on con-
tribution analysis, and co-edited eight books on program evaluation and performance
monitoring. In 1989 and in 1995, he was awarded the Canadian Evaluation Society Aw ar d
for Contribution to Evaluation in Canada . In 2006, he was made a Canadian Evaluation
Society Fellow. Dr. Mayne’s current research interests are on approaches for strengthening
impact evaluation, useful theories of change, and dealing with attribution.