Int. J. Electronic Governance, Vol. x, No. x, 2016 1

Copyright © 2016 Inderscience Enterprises Ltd.

Labelled causal mapping for structuring public

policy problems and design of policy options

Osama Ibrahim1 and Aron Larsson1,2 1Department of Computer and Systems Sciences (DSV), Stockholm

University, Stockholm 16407, Sweden 2Department of Information and Communication Systems (IKS), Mid

Sweden University, Sundsvall 85070, Sweden

Email: osama@dsv.su.se

Email: aron@dsv.su.se

Abstract: Systems analysis allows quantitative, empirical testing of models

that exist in the study of public policy. Simulation and visualisation

techniques can help policy makers to reduce uncertainties on the possible

impacts of policies. This paper presents a new tool for systemic modelling

and simulation of public policy problems. The tool bridges the gap between

the user’s complex mental model and the explicit graphical representation in

order to enable knowledge representation and system analysis. It allows users

to build a systems model of a policy decision situation and simulate the

system behaviour and responses to changing external factors and policy

interventions over time. We propose a policy-oriented problem structuring

method, Labelled causal mapping, which provides a labelling scheme with

iconic representation of the involved actors and key variables and a

quantitative dynamic simulation model of the control flows and causal

dependencies. The ultimate aim of the proposed modeling and simulation

approach aims to support: (i) better understanding and transparency by

clarifying and sharing the modeling assumptions; (ii) evidence-based

policymaking by bringing facts and abstractions from scientific and experts’

knowledge into the modelling process; and (iii) incorporation of the newest

management technologies into public decision-making processes, including:

cognitive strategic thinking, scenario planning and participation. The tool

supports the design of policy options and integrated impact assessment in

terms of social, economic, environmental and other impacts. A web-based

tool prototype has been implemented in a Node.js environment and is

accessible from an online graphical user interface (GUI). Two policy use

cases are presented for demonstration and testing of the prototype.

Keywords: Public Policy Analysis, Exploratory modelling, Simulation and

visualization, Impact Assessment, Scenario Planning, Systems Thinking,

Causal Mapping.

1 Introduction

Much has been written about the complexity of public policy decision-making

problems. Those responsible for creating, implementing and enforcing policies are required

to make decision about ill-defined problems occurring in rapidly changing and complex

Osama Ibrahim and Aron Larsson

environments characterised by uncertainty and conflicting strategic interests among the

multiple involved parties (Mitchell, 2009; Davies, 2004).

The ability to detect problems and emergencies, identify risks and reduce uncertainties

on the possible impacts of policies are among the key challenges facing the policymaking

process (Stefano et al., 2014).

The impact assessment (IA) of policy proposals remains a challenge, since the effects

of the alternative policy options are delayed in time and the ultimate impact is affected by

a multitude of factors. In order to conduct a robust and relevant IA that implements the

principle of sustainable development, it is required to determine the social, economic,

environmental, organizational, legal and financial implications of a new policy (Wacław,

2014). In addition, there are certain key aspects which should be present in order to define

the scope of the policy analysis, including: (i) objective(s) of the policy analysis, (ii)

Geographical area: (global, regional, national, sub-national and local), (iii) Time aspect

(short, medium and long-term), (iv) sectors of the related governmental activities, (v)

participation of actors, and (vi) engagement (involvement) of stakeholders.

The focus of this study is the prescriptive policy analysis, the ex-ante impact assessment

carried out at early stages of policy formulation1, in order to answer forward-looking

questions concerning consequences of alternative actions and possible futures and provide

prescriptions about policy proposals under consideration.

The underlying question in this paper is “how to support public policy decision-making

processes on the different EU policymaking levels using an ICT tool for simulation of

policy consequences and possible future scenarios”? The aim then becomes to design of

an ICT tool for policy makers from the different EU policymaking levels that assists public

decision-making processes through exploratory modelling of a public policy problem,

simulating and visualizing the consequences of possible future scenarios and the societal

impacts of alternative policy (decision) options.

We aim from the tool introduced in this paper to allow users (policy analysts, policy

makers) to build their “own” models of policy problems to facilitate deeper understanding

of the problem and sharing of the problem definition for increased transparency. The design

objectives include the design of integrated, customizable and reusable models, conforming

to proper modelling standards, procedures and methodologies to allow model

interoperability. Of importance is to create more complex or wider perspective models

using existing components or sub-models (blocks) and to ensure long-term thinking by

incorporating time aspect into the simulation model. Other design objectives include:

supporting model validation to ensure the reliability of the model and, consequently, of

policies; engagement of decision-makers and a wide range of stakeholders; interactive

simulation using animations and visualisation techniques to display the model operational

behaviour graphically; and output and feedback analysis to provide a feedback on the

simulation process or on the initial modelling assumptions.

This work is carried out within the contexts of the project Sense4us2, aiming for the

creation of a web portal integrating public policy decision support tools, linked open data

search tools and social media analytics, in order to enable easy access to information

sources and knowledge creation.

User requirements in a policy modelling and simulation tool were assessed using both

interviews and an online survey in a study done within the Sense4us project, (Taylor et al.,

1 Policy formulation is the process of standardizing, or rating, the proposed policy as a viable, practical, relevant solution to the identified problem. The draft of legislation is usually considered as the second activity of policy

formulation but is left out of focus herein. 2 EU FP7 research project ‘Data insights for policymakers and citizens’, http://sense4us.eu/

Labelled Causal Mapping

2014) that targeted potential interested policy makers, analysts and researchers in

governmental and legislative institutions.

Trustworthiness was the most discussed requirement around policy modelling and

simulation. Most of the end users wished to understand the methodology the tool will use

and want to be able to see all the source data that is used to present the predicted outcomes

in order for them to trust the tool. Secondly, policy makers explicitly referred to the

dynamism of underlying factors in certain policy fields, therefore flexibility to choose

variables in the tool on their policy issues, in order to make comparisons between different

prognoses is important, leading to the idea of a member account being incorporated into

the tool to enable users to individually shape policy simulation.

Models and simulations are too often perceived as black boxes, unintelligible to the

users. This is a particular challenge in public policy where decisions need to be taken on

the basis of transparent information. There is a lack of policy-oriented modelling and

simulation tools, whereas the existing econometric models are unable to account for human

behaviour and unexpected events and the new social simulations are fragmented, single-

purposed, suffer from lack of scalability to the macro level and require high level of

technical competency (Stefano et al., 2014). There exist several software packages for

processing causal data, graphing and analysing causal maps (e.g., CMAP3 and Decision-

Explorer). In addition, there exist software packages for quantitative system dynamics

simulations, in a strict sense for system performance analysis and prediction (e.g.,

STELLA). None of them, however, is dedicated to policy analysis and decision support for

policymaking.

This study takes a design-science research, (Hevner et al., 2004), approach to

developing a public policy scenario planning and impact assessment simulation tool. In the

design-science paradigm, knowledge and understanding of a problem domain and its

solution are achieved in the building and application of the designed artefact. The design

science methodology is based on the following main activities: (i) Explicate the problem;

(ii) Outline artefact and define requirements; (iii) Design and develop artefact; (iv)

Demonstrate artefact; and (v) Evaluate artefact (Johannesson and Perjons, 2012).

We classify this study as a development and evaluation focused design-science study.

The focus is the design and development of the artefact using both research and creative

methods. The prototype system is assessed based on the user requirements and the design

objectives. In addition, the viability of the tool is demonstrated and a lightweight evaluation

is performed to assess the completeness of the prototype in handling correctly the semantics

and processing of the labelled causal mapping method.

2 Systems approach and Policy modelling

Easton (1965) envisioned the ‘Systems approach’ as a framework and model to address

the central problem of empirical political study. Systems approach argues that public policy

is the product of a system and that system is a compilation of both intra-societal

components and extra-societal components and actors within an environment. This

compilation of components and actors, along with the interjected demands, form inputs to

the political system (decision-making processes) – which in turn generate outputs (public

policy). The effectiveness of the outputs, as measured by feedback, form new inputs that

are acted upon by the system (Easton, 1965).

Stewart and Ayres (2001) have argued for the use of systems approach in policy making

by the following points: (i) the systems approach offers policy makers a fresh set of

perspectives on the fundamentals of policy analysis; (ii) policy design is as much a matter

of choosing structures and relationships as of choosing instruments; and (iii) understanding

Osama Ibrahim and Aron Larsson

causation means acknowledging two-way influence and the role of feedback. They

introduced both the terms: ‘systems analysis of policy’ (understanding what is happening

when policy is made) and ‘systems analysis for policy’ (generating concepts, ideas, and

modes of action to make recommendations about policy problems), (Stewart and Ayres,

2001).

Stefano et al. (2014) addressed the challenges facing the model-based collaborative

governance and inferred that the systems thinking and system dynamics approach may

prove a useful dynamic tool for next generation policy making, which can be applied in

conjunction with other modelling techniques to produce hybrid models for public policy

analyses.

Researchers in System Dynamics modelling (Andersen et al., 2007; Bryson et al., 2004)

and causal mapping (Ackermann and Eden, 2005) have been working together on finding

the right mix between methodological power and richness with a “Systems Thinking”

approach. Richmond (1994) identifies three impediments to representing complex mental

models as the series of stocks and flows in system dynamics modelling:

The gap between the explicit graphical representation and the mental model;

The complexity of the explicit model that results in cognitive overload;

The ambiguity of the semantics of System Dynamics diagramming.

The problem structuring methods (PSM) provides an alternative path. PSMs deal with

unstructured problems characterised by the existence of multiple actors, multiple

perspectives, incommensurable and/or conflicting interests, and key uncertainties. See

(Rosenhead and Mingers 2001) for an overview of the characteristics of a wide range of

PSMs.

The use of cognitive or causal maps is among the early approaches to enable for

problem understanding, originally intended for representing social science knowledge, see

(Axelrod 1976). What causal maps contribute is a visual, mental imagery-based simulation

of the system's behaviour for system analysis and social communication. It is obvious that

such maps can be useful for analysing, developing and sharing views and understanding

among key actors also for creating some preconditions for intervention (Wang and

Laukkanen, 2015). Large-scale causal maps can be used to model complex policy

problems, representing what a decision-maker thinks about the drivers, barriers,

instruments and consequences of change achieved by a certain policy proposal. In addition

to the policy analyst knowledge and domain expertise, data for building such maps are

acquired from various online sources, including: (i) published policy evaluation and impact

assessment reports; (ii) related research literature and reports from research institutes and

NGOs; (iii) open governmental data; and (iv) public online political discussions.

To deal with the dynamic complexity inherent in social systems and to infer dynamic

behaviour, quantitative simulation is required. Therefore, and particularly in those

situations where it is important to understand the interactions among the variables over

time, the value added by causal maps can be significantly increased if they are

complemented with simulation modeling (Senge, 1990; Sterman, 2000).

The long-term implications of policy making imply the need to consider the range of

possible futures, sometimes characterised by large uncertainties. “Scenarios” are a main

method of projection, trying to show more than one picture of the future. “The analysis of

scenarios of change allows the design of strategies to take place in spite of the messiness

of the situation” (Schoemaker, 2002). Scenario-driven planning closes the gap between

problem framing which depends on qualitative analysis and problem solving which

depends on quantitative analysis by blending qualitative and quantitative analytics into a

unified methodology (Georgantzas, and Acar, 1995).

Labelled Causal Mapping

3 Labelled causal mapping method

The proposed approach for policy modelling and simulation is based on: (i) systems

approach to policy analysis; (ii) problem structuring using causal maps; (iii) graphical

representation of complex problem situations as both a knowledge representation technique

and Systems analysis tool; (iv) scenario-based dynamic simulation. The main rationale is

to support a flexible, informative and a more rational and structured policy making process

identifying effective policies by gaining insight from conceptual and empirical analysis of

the system.

Acar, (1983) introduced the ‘causal mapping and situation formulation’ method in his

PhD dissertation as an analytical method for formulation and analysis of unstructured

strategic problems. The method enhances the causal mapping with rich computational

properties by including indications not only of the directions and signs of the presumed

causal influences, but also of their intensities, minimum threshold values and the possible

time lags, see (Acar and Druckenmiller 2006). The rich computational semantics of Acar’s

causal mapping support automated modelling and simulation in ways that other varieties

of cognitive mapping approaches do not. The automated simulation capabilities for Acar’s

causal mapping was explored in (Acar and Druckenmiller 2006; Druckenmiller and Acar,

2009). The complexity of causal loops in the models and calculations of successive waves

of change through the model when simulated presented a substantial design challenge for

developing automated scenario support using object-oriented techniques alone. A

prototype system for the development and simulation of causal maps that uses RePast 2.0,

a Java agent-based modelling (ABM) and simulation library. In this implementation

individual nodes of the causal map are represented as agents and links between agents

constitute the environment of the system, analogous to a communication network. It

focuses on business strategy development, interaction and behaviour of the involved actors

as autonomous agents.

This research proposes a policy-oriented version of Acar’s causal mapping, the

‘Labelled causal mapping’ as a systematic method for structuring of public policy decision

situations and simulation of possible futures and consequences of alternative policies. The

method defined policy-oriented categories with iconic representation for the model

elements. The following subsections provide a detailed description of the model elements

and the concepts of the simulation process.

3.1 Model elements and labels

3.1.1 Actors

Institutions, organizations, committees or individuals involved in the decision-making

process. The actors can be labelled as: (i) Official actors – including both: [legislative

actors (Parliament committees, political parties) and executive actors (Governmental

bodies, departments and institutions, chief Executive, staff/officials, agencies, bureaucrats

and civil servants)]; and (ii) Unofficial actors: [Interest groups, political parties, citizen

representative bodies, NGOs, industry/trade Unions, think tanks, media].

Table 1 Labels and icons for actors

Executive actor

Legislative actor

Unofficial actor

Osama Ibrahim and Aron Larsson

3.1.2 Variables

Variables are factors or events that structure, constrain, guide, influence and indicate

impacts of actions taken by actors. These are idealised as quantitative variables, or

quantified using value scales, so that it is meaningful to talk about change in the form of

increases or decreases in their levels. Variables are divided into:

Independent variables: sources of change (graph origins)

Policy instrumental variables (decision variables): These are variables controlled by

policy actors through various types of policy instruments, which can be used to classify

these variables. Scenarios of change in these variables represent action alternatives (policy

options). These changes reflect the allocation of natural, human and capital resources, the

regulatory role of the government, regional and international cooperation.

Table 2 Labels and icons for policy instruments – controllable sources of change

1. Economic Instruments:

1.1 Financial instruments:

1.1.1 Public expenditure, investment or funding

1.1.2 Public ownership

1.1.3 Subsidies

1.2 Fiscal instruments:

1.2.1 Taxes, Fees and User charges

1.2.2 Incentives

1.2.3 Loans / Loan guarantees

1.3 Market instruments:

1.3.1 Property rights

1.3.2 Contracts

1.3.3 Tradable permits / Certificate trading

1.3.4 Insurance

2. Regulatory Instruments:

2.1 Norms and standards

2.2 Control and enforcement

2.3 Liability

3. Informational Instruments:

3.1 Public information centers

3.2 Sustainability monitoring & reporting

3.3 Public awareness campaigns

3.4 Consumer advice services

3.5 Advertising & Symbolic gestures

4. Capacity-building Instruments:

4.1 Scientific research

4.2 Technology and skills

4.3 Training and employment

5. Cooperation Instruments:

5.1 Technology transfer

5.2 Voluntary agreements

Labelled Causal Mapping

External factors (state variables): These are variables not under control of any of the actors.

Scenarios of change in these variables represent the possible futures.

Table 3 : Labels and icons for uncontrollable sources of change

1. Drivers and barriers

The drivers and barriers of change, are either associated to the political

context (e.g., the political ideology and strategic priorities of the

government of the day, the preferences and demands of politicians) or

the economic context (e.g., the availability of resources, the economic

growth, the economic climate, current and future commitments).

2. External environment’s disturbances and conditions

The system representing the policy problem is surrounded by physical,

biological, social, and psychological environments that the system

needs to adapt to.

3. Social, demographic and behavioural change

e.g., Population growth, immigration, culture, attitudes and

behaviours.

Dependent variables: impacts of change (middle and end nodes)

These variables represent the consequences of change in the independent variables, are

either direct impacts, associated with the sources of change, or indirect impacts, associated

with the direct impacts. Table 4 shows the different categories of impact variables as related

to one of the policy areas.

Table 4 : Labels and icons for policy impacts

Economy

Finance

Environment

Community/Social

Energy Infrastructure Transportation Healthcare

Education

Technology

Judiciary and Law

National Security

The scope of the model is defined by the involved actors and the variables of interest.

The system analysis must consider the involved actors as coupled with either an abstract

or concrete component. This way, the influence of the actor within and upon the system

clearly reveals itself. Actors have powers (control over some decision variables) and

interests in some outcome variables (goal variables).

Osama Ibrahim and Aron Larsson

3.1.3 Change transmission channels

Links or change transmission channels are cause-effect relationships connecting the model

variables, defined by:

a) direction, from an upstream variable to a downstream variable

b) sign (positive is changes in the same direction or negative if changes are in

opposite directions)

c) change transfer coefficient: intensity of the causal relationship in terms of the

proportionality ratio of change transfer, how much change is transferred to the

downstream variable in case of 1% change in the upstream variable

d) time lag (if change transmission is not instantaneous) expressed as

weeks/months/years; and

e) minimum threshold for the change in the upstream variable (if applicable).

Two types of change transmission channels:

(i) full channel: a double arrow from an upstream variable X to a downstream variable

Y, if X is sufficient to induce change in Y

(ii) half channel: a single arrow from an upstream variable X to a downstream variable

Y, if X is necessary but not sufficient to induce change in variable Y. Half channels

from a set of variables such as {U, V, W} to variable Z, need to be all activated

before change can be transferred to Z.

Figure 1 : Change transmission channels

“Additivity” and “Transitivity” are two main characteristics of change transmission in

the model that allows running scenarios of change on the causal map. Once initiated in one

or more of the independent variables, change is transmitted throughout the network given

the transfer ratio and time lag for each channel. The main assumption for transmission of

change is that "the percentage relative change in a downstream variable Y is a linear

function of the percentage relative change in an upstream variable X". The definition of

time lags, minimum thresholds quantifications of the change transmission channels besides

the existence of the half channels add a meaningful dimension of nonlinearity to the model.

For full channels the transmission is automatic, as soon as a variable incurs a change, the

channels proceeding from it become activated and transmit to the downstream variables,

according to the indicated time lags. The same applies for half channels as soon as all half

channels converging to a node are activated. Assuming that 𝑑𝑋/𝑋 represents the relative

change in a variable 𝑋, then the relative change 𝑑𝑌/𝑌 in a downstream variable 𝑌 is given

by: 𝑑𝑌

𝑌= 𝑎

𝑑𝑋

𝑋, where 𝑎 is the real valued change transfer coefficient for the link XY. Given

a set of activated channels {𝑋1𝑍, 𝑋2𝑍, …, 𝑋𝑛𝑍} going into 𝑍, then: 𝑑𝑌

𝑌= ∑ 𝑎𝑖

𝑑𝑋𝑖

𝑋𝑖

𝑛𝑖=1 .

Labelled Causal Mapping

3.1.3 Demonstration policy use case

Figure 2 presents the causal mapping model for the use case policy problem, ‘Regulation

on Personal protective equipment (PPE)1’. The general policy objectives are: (i) high level

of health and safety protection for PPE users; (ii) free movement of PPE and a fair playing

field for PPE economic operators; and (iii) simplifying the EU regulatory environment

related to the field of PPE. The model shows the actors’ participation in a co-decision

legislative procedure (European commission’s directorate generals (DG): Enterprise and

Industry, DG-ENTR; Secretariat-General, DG-SG; Health and Food Safety, DG-SANCO).

The model shows two policy options, with the legislative option more effective in

achieving the policy objectives. Links are variably marked with positive and negative signs

indicating signs and intensity of the causal relationships.

Figure 2 : Causal map of the PPE use case

1 Source: European commission (EC), (2014) Impact Assessment report, Industry and Entrepreneurship,

‘Regulation on personal protective equipment’, available on:

http://ec.europa.eu/smart-regulation/impact/ia_carried_out/cia_2014_en.htm#entr

Osama Ibrahim and Aron Larsson

3.2 State of the system

The baseline or the “status quo level of the system”, is a projection of the

characteristics and behaviour of the system at the beginning of the analysis. It is used as a

point of comparison. The changes/responses are expressed in terms of percentage changes

relative to the baseline. Thus, change transfer coefficients are dimensionless and the initial

values of relative change for all variables in the system are set to zero, i.e. the initial state

in terms of the relative change is the null vector (0, 0, …, 0). The simulation runs are based

on discrete time points (T0, T1, T2, …, Tn), and the user is required to define a time step T,

time period between two consecutive time points as a number of weeks, months or years;

and maximum number of iterations n for the simulation. The state of the system at time Ti

is then defined as the values of all variables in the system at time point Ti of the simulation

run, given a specific scenario of change. The desired state of the system is given by targeted

changes in a set of impact variables of interest to the decision-maker, represented in a goal

vector. Each actor has a goal vector reflecting the targets of that actor.

3.3 Scenarios of Change

Quantitative analysis of the causal mapping model can show the opportunities for

policy interventions to achieve targeted changes in impact variables defined as policy

goals. The policy options represent different combined and controlled changes in the

system inputs to produce the targeted outcomes. Scenarios of change can be defined as a

combination of specified percentage changes occurring at a specific time point or at

successive time points. Defining quantified policy goals, in the form of a goal vector for

each actor, allows the analysis of these scenarios with respect to goal achievement. In

addition, structural analysis of the causal map, can support scenario analysis (e.g.,

reachability analysis shows the ability of a scenario of change triggered at an independent

variable to achieve a particular goal if the goal variable is reachable from this variable).

A “pure scenario” is a scenario of change in one particular independent variable, while a

“mixed scenario” is a scenario of change in more one than one independent variable. We

also need to differentiate between “alternative futures”, scenarios of change in

uncontrollable independent variables caused by natural or external forces, and “alternative

actions”, scenarios of change in controllable independent variables willed by actors.

The alternative futures reflect the concept of “uncertainty”: each possible future provides

a projection of changes in uncontrollable sources of change, a probability is assigned to

each of the possible futures. The alternative actions by actors reflect the different policy

options, i.e., the changes of controllable sources of change as policy instruments, given

some future circumstances. The user can define as many scenarios of change as needed,

using combinations of alternative futures and courses of action by actors. Simulation of the

alternative futures would bring up the various vulnerabilities of the system.

Analysing the behaviour of the system under various conditions yields a much deeper

understanding of the problem and supports the design of policy options.

3.4 Goal feasibility and compatibility

The formulation and evaluation of policy options can be addressed by answering

questions like: What variables are relevant and controlled by actors of the problem and

what constraints apply to them? What variables are relevant but are not subject to control?

How do controlled and uncontrolled variables interact to produce an outcome? What are

the actions needed to generate a desirable state of the system or to block the occurrence of

an undesired one? What are the costs and benefits of these actions?

Labelled Causal Mapping

Decision-makers at senior levels of government operate within a finite set of available

resources and timelines. Furthermore, there are inherent constraints that the decision-maker

needs to consider, such as annual cycles for strategic planning, budget, and legislation.

Legislative and political imperatives add to the complexity of government policy decision-

making and the selection of policies. For an actor, the triggering of change in one the

controllable variables imposes the expenditure of funds and resources. If an actor has the

required resources and/or funds available for a course of action that realizes his goals,

assuming inactivity of other actors and external environment, then his goal vector is

“internally feasible”. If the required resources are not available, then there is an intrinsic

problem of consistency between the actor’s goals and capabilities.

When considering the moves of other actors and the changing external environment,

if no pure or mixed scenario can be found to realize the actor’s goals, then his goal vector

is said to be “infeasible”. If it was found as an “internally feasible” goal, then the actor has

a problem to synchronize with the other actors. If a scenario could be found to realize the

actor’s goals, then his goal vector is said to be “feasible”. If it was not found as an

“internally feasible” goal, then the actor is benefiting from interacting with the whole

system in turning potential problems and constraints into opportunities.

The concept of compatibility is connected to the concept of feasibility. Components

of a single goal vector, as well as goal vectors of different actors are called “compatible”

if a scenario of changes can be found to realize them jointly. Goal compatibility is a graded

concept; two goals that can be realised by the same pure scenario are more compatible than

two goals requiring a mixture of pure scenarios.

3.5 Tactics and game theoretic analysis

A policy option can be viewed as either a combination of actions taken by the involved

actors that may achieve the policy goals in a cooperative decision-making situation (a co-

decision legislative procedure); or the actions taken by the focal actor to preempt/counter

natural, external forces and/or other competitors’ moves in a competitive decision-making

situation.

A tactic is a course of action logically paired with an alternative future and what other

actors might do. It is simply a change or sequence of changes triggered at controlled

variables by an actor that would help realize his goal vector. The effectiveness of a tactic

of an actor is a measure or at least an evaluation of the degree to which it helps him realize

his goal. The efficiency of a tactic is a measure of the use of resources to realize the goals.

Both the tactical effectiveness and efficiency need to be measured in comparison to the

competing tactics the actor might choose from and also in connection to the competitors’

tactics and alternative futures the actor aims at countering or preempting.

The competitive analysis allows policy makers to shape policies that takes into account

their competitors’ likely responses when deciding on their own actions, by quantifying and

estimating the utility each actor has in the alternative courses of action, while accounting

for the possible alternative futures. Game theoretic analyses are based on the idea of a

‘tactic’ as a sequence of moves to preempt or counter nature’s or other competitors’ moves.

It then allows either to devise one tactic for the set of actors involved in a cooperative

decision-making situation which would yield optimal result in achieving the policy

objectives; or to devise one tactic which would yield optimal results for the focal actor in

a competitive decision-making situation. For instance, Brandenburger and Nalebuff (1995)

provided a useful tool for identifying the various courses of action leaders have available.

They advocated using the acronym PARTS (Players, Added values, Rules, Tactics, and

Scope).

Osama Ibrahim and Aron Larsson

4 The Artefact description

Based on the labelled causal mapping method, a policy modelling and simulation tool

is implemented using a gaming simulation approach. Having a goal, the focal actor (in

individual mode) or each participant (in a group/participatory mode), tries out different

scenarios to identify the ones that best achieve the goals or intended consequences, with

the unintended and unexpected consequences used as a lasting learning method. Here, the

simulation model for policy analysis has an exploratory rationale rather than being a

prediction model. The individual simulation runs are not being treated as providing

predictions or explicit answers to the policy makers' questions. Instead, new information

that was implicit in their prior knowledge that defined what is plausible, is generated to

support an informed policy decision. The aim is to explore the implications of alternative

actions, varying assumptions and cause-and-effect hypotheses to deal with insufficient data

or unresolved uncertainties. Such exploratory approach, described in (Bankes, 1993), can

provide a basis for decision making in complex and uncertain problems.

The simulation-based impact assessment of user-created policy scenarios results in a set of

considered policy options and their simulated social, economic, environmental and other

impacts.

We suggest a step-wise procedure that enables the user to build a causal mapping

simulation model for the policy problem at hand. The cases of policy analysis are defined

as ‘policy problems’ and several impact assessment models can be defined for each policy

problem. ‘Ultra-Low Emission Vehicles (ULEV) Uptake1’ policy problem is used as a test

case.

Step 1. Define a policy problem A policy problem is defined using a title and a policy question and is given an ID by the

system. Figure 3 shows user interface for defining a policy problem, including:

1- Policymaking level: (EU level, National level, Local level)

2- Geographical area: (A set of EU countries – EU level; A country and a set of local

regions – National level; A local region – Local level).

Figure 3 : User interface for defining a new policy problem

1 Source: (Online) House of Commons, Transport Committee, (2012),‘Plug-in vehicles, plugged in policy?’,

Fourth Report of Session 2012–13, available on:

http://www.parliament.uk/documents/commons-committees/transport/Plug-in%20vehicles%20239.pdf (Online) Department for Transport, UK, (2015),‘ Uptake of Ultra Low Emission Vehicles in the UK’. A Rapid

Evidence Assessment, available on:

https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/464763/uptake-of-ulev-uk.pdf

Labelled Causal Mapping

Step 2. Define scope of the policy analysis model The user sets the boundaries of the policy impact assessment, as a guideline for the

modeling process and as an input to information search tools. This includes defining the

objective(s) of the policy analysis, the time aspect of the analysis and the related policy

domain or governmental activity, as shown in figure 4. The aim of this description or

classification is to facilitate the query processes to reuse model components from stored

policy models and to identify trends in groups of policy problems.

Figure 4 : User interface for defining scope of the policy model

Step 3. Create an information model of the policy problem A check-list editor, shown in figure 5, provides a tree structure for the concepts that

result from the surveying or information searching process under the main issues of the

policy problem and the different categories for the model elements. The editor allows the

user to edit the tree structure by adding or deleting elements.

Figure 5 : User interface for a policy check-list model

Osama Ibrahim and Aron Larsson

Step 3. Define indicators, measures and links to data sets for variables This step involves the definition of how to measure the concepts and decide which

indicators should be used. The indicators should be relevant to the scope, easy to track over

time, measurable through quantitative metrics or value scales. Open governmental data

portals provide statistical data and indicators for the different policy domains and themes.

Figure 6: User interface for defining measures and indicators for model variables

Step 4: Import or add elements to the model builder graphing canvas. In this step, the user adds elements to the model builder canvas or selects elements from

the information model formed by the information search results and imports them to the

graphing canvas as model elements with icons assigned to these elements based on the

defined categories. The labelled causal map graphing canvas operates in two modes: the

‘modelling’ mode and the ‘simulation’ mode. In the modelling mode, the user can create,

edit, save, load and delete models. The user can add elements, i.e., actors and variables to

the model by clicking on the respective icon from the model elements toolbar on the left

panel.

Step 5. Build hypotheses on the causal links and interrelationships: The user can refine the model structure by adding control flows and causal links (both full

and half change transmission channels), by the dragging the red dot that appears beside the

element by clicking on it or by a right-click and drag to the respective downstream variable.

Using.

Step 6. Edit nodes’ and links’ properties – Define a baseline By clicking a node, the user can edit the node properties and perform functions that appear

on the left panel, including: (i) For actors: name, description, delete actor, delete control

flows, randomise colour of the control flow; (ii) For variables: name, description, baseline

level, unit of measurement and delete node. Quantify the causal links and verify them

through statistical analysis of the data sets for indicators. In case of unavailability of

statistical data, the quantification can be done based on expert’s judgment.

Step 7. Define actors’ goals For each actor of the model, define the actor’s goal vectors and preferences of goals

realization (i.e., ranking of the goal vector components).

Step 8. Define and simulate scenarios of change In the simulation mode, the user can define the time step for the simulation model in terms

of a number of weeks, months or years and a maximum number of iterations, which appears

as the limit for the simulation time slider.

Labelled Causal Mapping

Figure 7 : Model for the policy case on the causal map graphing canvas

Osama Ibrahim and Aron Larsson

A scenario editor window can be used to define scenarios of change for the independent

variables (policy instruments and external factors), for which the change scenario appears

as part of the node properties, in terms of timely percentage changes in their levels relative

to baseline levels.

The ‘simulate’ button populates the model variables throughout the model with the

transferred changes, which are updated by moving the time slider. The ‘line graph’ button

shows a line graph for the simulation results for each of the model variables in different

colours, by selecting the respective nodes.

Figure 8, shows a model example for a demonstration policy case in the simulation mode

with the scenario editor window to define and edit multiple scenarios of change in the

independent variables. Figure 9 show the simulation results for a scenario of change and

Figure 10 shows the visualization of results using line graphs for selected nodes.

Figure 8 : Scenario editor window in the simulation mode

Figure 9 : Simulation results for a scenario of change

Labelled Causal Mapping

Figure 10 : Visualization of simulation results for selected variables

5 Conclusion

In the labelled causal mapping method, presented in this research as a policy-oriented

version of Acar’s causal mapping, a public policy problem is abstracted as deviations from

goals and/or standards, those deviations originate in a change that propagates itself through

causal connections, while a policy option is seen as a purposeful action for change that can

resolve these deviations in order to achieve the policy objectives. The analysis of the

transfer of change throughout the system can be used to assess the effectiveness and

efficiency of alternative scenarios of change in achieving objectives.

The basic linearity of the system can be a source of criticism, but it allows increasing

the level of detail of the model by adding as many elements as needed to represent relevant

factors that affect the outcome of a decision. Therefore, it increases the model’s capability

to be used for policy analysis, without affecting its feasibility in terms of mathematical

tractability and data requirements.

The prototype system implemented in this research is assessed based on the design

objectives, as follows:

- User-created policy scenarios: Although it is a serious and time consuming task, it helps

the user to develop advanced skills and gain a deeper understanding.

- Integrated, customisable and reusable models: Creating more complex or wider

perspective models using existing components or sub-models (blocks).

- Model validation: Defining data requirements for sensitivity analysis and validation.

Engaging in a cumulative learning through experimentation. The model validation, which

is to be done by the user, is reduced to validation of sub-models, determining model

parameters from validated sources and ensuring the plausibility of outcomes. And the

model quality control is limited to the plausibility and the degree of completeness of the

model (inclusion of all factors and phenomena that might influence outcomes).

- Engagement of decision-makers and stakeholders: The resulting policy model being

simple, robust, easy to control, adaptive, easy to communicate with and as complete as

possible.

Osama Ibrahim and Aron Larsson

- Interactive simulation: Using a graphical representation of the policy problem is

important for end-users for facilitating communication about the problem understanding

among different governmental departments

- Output and feedback analysis: Synthesising new knowledge on the system, when

ultimately, a satisfying result has been achieved or when a complete understanding of the

system has been gained.

6 References

Acar W., (1983). Toward a Theory of Problem Formulation and the Planning of Change:

Causal Mapping and Dialectical Debate in Situation Formulation. Ann Arbor,

Michigan: U.M.I.

Acar, W. and Druckenmiller, D. (2006). Endowing cognitive mapping with computational

properties for strategic analysis, Futures 38:993-1009.

Bankes, S. (1993). Exploratory modeling for policy analysis. Operations Research, 41(3),

435-449.

Brandenburger, A. M., & Nalebuff, B. J. (1995). The right game: Use game theory to shape

strategy. Harvard Business Review.

Davies PT (2004), “Is evidence-based government possible?” Jerry Lee lecture. 4th Annual

Campbell Collaboration Colloquium Washington D.C.

Easton, D. (1965). A systems analysis of political life. New York: John- Wiley & Sons.

Georgantzas, N. C. and W. Acar (1995). Scenario-Driven Planning: Learning to Manage

Strategic Uncertainty. Westport Connecticut: Quorum Books.

Hevner, A., March, S., Park, J., and Ram, S. (2004). “Design Science in Information

Systems Research,” MIS Quarterly (28:1), pp. 75-105.

Johannesson, Paul & Perjons, Erik (2012). A Design Science Primer.

Koliba C., Zia A., (2011), Theory Testing Using Complex Systems Modeling in Public

Administration and Policy Studies: Challenges and Opportunities for a Meta-

Theoretical Research Program, Proceedings of the 2011 Public Management Research

Conference, Maxwell School of citizenship and public affairs at Syracuse University.

Mingde Wang and Mauri Laukkanen (2015), Comparative Causal mapping: The CMAP3

method. Ashgate Publishing, Ltd..

Mitchell B. (2009). Policy-making process. Available: (retrieved on 10/03/2015)

http://www.waterencyclopedia.com/Oc-Po/Policy-Making-Process.html

Rouse, W.B & N.M. Morris (1986): On Looking Into the Black Box: Prospects and Limits

in the Search for Mental Models. Psychological Bulletin, Vol. 100, No.3, 349363

Schoemaker, P. J. H. (2002). Profiting From Uncertainty: Strategies for Succeeding No

Matter What the Future Brings. New York: Free Press.

Senge P.M. (1990). The Fifth Discipline: The Art and Practice of the Learning

Organization. New York: Doubleday Currency.

Labelled Causal Mapping

Stefano A., Camello C., Riccardo O. & Pietro S. A., (2014): Policy Modeling as a new area

for research: perspectives for a Systems Thinking and System Dynamics approach?

Proceedings of the Business systems Laboratory 2nd International Symposium.

Sterman J. D. (2000). Business Dynamics: Systems Thinking and Modelling for a Complex

World. Irwin/McGraw-Hill: Boston.

Taylor S., Uzdavinyte R., Wandhöfer T., Fox R., (2015). Issues Arising from the

Specification of an Information Acquisition and Analysis Toolkit for Policy Makers in

Governmental and Legislative Institutions, Accepted paper to be published/presented

at: The eChallenges e2015, Vilnius, Lithuania.

Wacław B., (2014), Regulation Impact Assessment (RIA) at Poland and at Some EU

Countries. Procedia - Social and Behavioral Sciences Volume 109, Pages 45–50. 2nd

World Conference on Business, Economics and Management.