Six methods, one Design Framework

Foundations / Choosing a method

Method choice and design

The six causal systems mapping methods the MRC and CSO guidance compares, and the 13 questions that decide between them.

Six methods, three families

The guidance compares six causal systems mapping methods. None is best. Each carries a different amount of information, and that decides which one your project can use.

Qualitative. Systems-based theory of change maps, causal loop diagrams and CECAN participatory systems mapping. The links carry direction and polarity, but no numbers.
Semi-quantitative. Fuzzy cognitive maps. The links carry numbers, but the map is not a simulation.
Quantitative. Systems dynamics models and Bayesian belief networks. Numbers, and for systems dynamics, simulation.

The guidance presents them from conceptually simplest to most challenging, and that order is a fair guide to effort.

The six methods side by side

Read the fourth column as the decision column.

Method	What the map is	Participation in practice	Choose it when
Systems-based theory of change maps	Inputs, activities, outputs and outcomes, plus assumptions, linked non-linearly. Usually acyclic, so no feedback loops.	Flexible and low-demand. Group or individual building, with no formal analysis to explain.	You need to frame or evaluate an intervention, and make its logic and assumptions visible.
Causal loop diagrams	A qualitative picture of structure: factors linked by positive or negative connections, forming reinforcing and balancing loops.	Usually built as a group, then validated with the same stakeholders. The most used method in population health.	You want to explain how the system produces the outcomes it produces, and the group can work visually.
CECAN participatory systems mapping	A large inclusive map, with participant-driven network analysis used to draw out sub-maps and causal chains.	Participation is core, not optional, and the analysis is often co-designed with stakeholders.	You start with thin data but still need structure, and you want more than a picture but less than a model.
Fuzzy cognitive maps	Factors and connections weighted by participants, either for degree of certainty or for strength of influence.	Usually map building only. Quick, visual and engaging for participants.	You want to compare mental models or rank which factors are most influential, and you have one or two sessions, not six.
Systems dynamics models	A causal loop diagram extended into stocks and flows, specified as differential equations and run as a simulation.	Stakeholders can build and validate the map. The modelling sits with specialists.	You have a dynamic question about a delimited problem, with quantity and time in it, and the resources to model it.
Bayesian belief networks	An outcome with parent factors feeding it, each with states and conditional probabilities. Typically acyclic.	Least often participatory in population health. Stakeholders may join only at validation.	You need probable causes or probable outcomes under uncertainty, with limited data and no need to model feedback.

Each method also has a warning label. Theory of change maps can stay narrow and reinforce optimism. Causal loop diagrams rarely show the relative strength of links, and loop focus can crowd out other factors. CECAN PSM captures one point in time. Fuzzy cognitive map numbers are easily over-interpreted. Systems dynamics models turn on their assumptions. Bayesian belief networks demand technical skill and are hard to update live.

How to choose: three questions, in this order

What data do you already have?

Theory of change maps, causal loop diagrams and fuzzy cognitive maps have no data requirements. CECAN PSM can grow from a prior map, which suits thin data. Systems dynamics models and Bayesian belief networks need quantifiable data.

What must the output do?

Simulate change over time and you need systems dynamics. Produce probable causes and risks and you need a Bayesian belief network. Rank the influence of factors and you need fuzzy cognitive maps or CECAN PSM. Explain the logic of an intervention and a theory of change map is the natural form. Explain how the system produces its outcomes and you want a causal loop diagram.

Who is in the room, and for how long?

Capacity building grows along the list: low to moderate for theory of change maps, causal loop diagrams and CECAN PSM; moderate for fuzzy cognitive maps; moderate to extensive for systems dynamics; extensive for Bayesian belief networks. Skills follow the same gradient. Network analysis is useful for causal loop diagrams, CECAN PSM and fuzzy cognitive maps. Mathematical specification and specialist software are needed for systems dynamics, and statistical work for Bayesian belief networks.

Pick the simplest method that answers your question. Method choice is not a quality ranking. A causal loop diagram that a group trusts and uses beats a simulation nobody can maintain.

The Design Framework: three considerations, 13 questions

The Framework asks two questions: which method meets your needs, and what feasibility issues you must accept. It organises the 13 questions under three considerations, in any order you like. An editable version comes with the guidance.

Consideration 1: what is the scope of the project?

1. The intended purpose of systems mapping in your project.
2. The knowledge or information gaps you seek to address.
3. The types of output, including map features, that will be useful.
4. What you intend to do with the map.

Consideration 2: what is the added value of a participatory approach?

5. How the participatory approach will benefit the project.
6. Who the stakeholders are, and who you will involve.
7. The emphasis you will place on participation.
8. The stages at which you will involve participants.

Consideration 3: what affects your capacity to use a method?

9. The capacity building that meaningful involvement will require.
10. The data or information already available.
11. Where the mapping process will take place.
12. The skills, resources and expertise required.
13. The data collection methods, for the mapping and the project.

A comparative table closes each consideration sub-section. Use it to check your answers against what each method can do.

Read the Framework throughFamiliarise yourself with the content and the 13 questions first.
Answer the easy ones firstUse the editable template and fill in every question with a clear answer today.
Go back to the hard onesCollect feedback from your team, the literature or methods experts.
Check against the comparative tablesSee which methods fit your purpose and your resources.
Decide, then adaptChoose your method, and record how you will adapt it to your context.

One test before you go further. If you selected very few or no items under question 1 or question 3, systems mapping may be less suitable for your project than you assumed. That is a useful result to get early.

What the scoping review found

The guidance was supported by a scoping review of peer-reviewed literature from 2000 to 2021. 73 papers met the inclusion criteria. Three findings matter.

Causal loop diagrams dominate. 54 of the 73 papers used them, and a third of those used them within systems dynamics modelling. Theory of change maps, fuzzy cognitive maps and Bayesian belief networks appeared in a minority of studies, and the review found no population health studies using CECAN PSM.
Understanding was the most common purpose. 44 studies used mapping to understand an issue or context, against 13 for evaluation, 10 for intervention development and 10 for simulating a future state.
Reporting of the participatory process is weak. Most projects involved stakeholders in both map building and validation, and group building was far more common than individual. But only about a third reflected on the participatory approach.

Choose a causal loop diagram with group building and validation and you follow the well-trodden path, with examples to copy. Choose a Bayesian belief network or CECAN PSM and you are ahead of the published literature. Either way, write down who took part, when, how the boundary was set, and what the group disagreed about. Most projects omit that part.

Check your understanding & apply it · 4 questions, 2 essays

1A team wants to compare mental models and rank which factors are most influential along their discharge pathway. They have one workshop session rather than six, and no quantitative dataset. Which method fits?

2A team has a dynamic question about a delimited problem, with quantity and time in it, and the resources to model it. Which method does the Design Framework point to?

3You are starting the three choosing questions with a new project group. Which question comes first?

4A team works through the Design Framework and selects very few items under question 1, the intended purpose, and question 3, the types of output. What does this page say that means?

5Essay A trust wants to develop an intervention to reduce unplanned admissions along an existing discharge pathway. Staff know the pathway well, no quantitative dataset is ready, the group can work visually, and the aim is to explain how the current system produces the outcomes it produces. Choose a method and justify it against at least one alternative.

6Essay A colleague argues for a Bayesian belief network because it looks the most rigorous option. Using the three choosing questions and the comparative material on this page, argue the case against that decision.