Scriptapedia/Initiating and Elaborating a Causal Loop Diagram

Initiating and Elaborating a "Causal Loop Diagram" or "Stock and Flow" Model (Vennix and Rouwette)
This script is used to get an initial idea of central concepts and their relationships at the beginning of a project. If the aim of the project is to construct a formal simulation model, it is recommended to build a stock and flow model. In addition, when accumulations are important in the issue (for instance, when the problem is on a production chain or in human resources that are moving through different states), a stock and flow model is recommended.

Status
Best practices

Primary nature of group task
Convergent

Time
Preparation time: 20 minutes

Time required during session: 20 minutes

Follow-up time: 20 minutes

Materials

 * 1) Either three flip charts, OR wall space on which several flip charts / cling sheets are taped, OR a whiteboard, markers and flip chart, OR a projector and laptop with a suitable modeling software (capable of drawing Causal Loop Diagrams or Stock-and-Flow Diagrams). Note that in the last case a second person is needed to draw the diagram in the modeling software, while in the other situations one person may guide the group.

Inputs
A list of variables

Outputs

 * Interim output/product: increased consensus on dynamic hypothesis, or a possible structural explanation for observed behavior
 * Deliverable: a causal loop diagram which may be described either in a report (in the case that only a qualitative model is built), or be used as a dynamic hypothesis on the basis of which formal modeling starts

Roles

 * Facilitator/modeler with experience in drawing causal loop diagrams, and preferably with experience in building formal models

Steps

 * 1) Remind the group of the problem variable, preferably sketched as a reference mode of behavior. Remind the group of the list of variables elicited before. Place the list of variables in such a way that it is visible to the group of participants. Write the problem variable in the center of the white board.
 * 2) Build the model by following steps a, b, and c below (cf. Vennix, 1996: 120).
 * 3) * (a) Ask participants which variable from the collected list is a cause for changes in the problem variable. When someone makes a suggestion, include this in the drawing of the model in order to visualize what is meant. Then check to see if everyone agrees with the proposed relation. If someone disagrees, ask for clarification and try to determine what the group thinks the relationship should be. If a discussion goes on too long, you can choose to temporarily 'park' this item and continue with another part of the model. Hopefully, there will not only be variables that have a direct relationship with the problem variable, but you will also build a few logical chains of reasoning (via intermediate variables) into the model. In addition, check the polarity (positive or negative) of the relationship.
 * 4) * (b) After spending some time doing this, proceed to the consequences of changes in the problem variable.
 * 5) * (c) At the point where a feedback chain becomes closed, check with the entire group to see if the chain as a whole is correct. Check again to see if a loop is positive or negative. The "Ratio Exercise" script may be used to draw out loops.
 * 6) In the last part of the session, analyze the model by checking the feedback loops one more time. Before you close the group session, make sure you do the following:
 * 7) * If there is a list of 'parked' issues, go through them.
 * 8) * State once more what has been done and what will happen with the final products.
 * 9) * Formulate a few concise conclusions. As Andersen and Richardson (1997) say: "End with a bang!"
 * 10) * Make sure that all the information which is necessary for the report has been noted.

Evaluation Criteria

 * Improvement in quality of communication, insight, consensus on the problem, and commitment with regard to actions
 * Improved causal loop diagram (see Figure 1)



Authors
Jac Vennix 1996, used for bachelor (undergraduate) course by Etiënne Rouwette from September 2007

History
Earlier publications Vennix

Revisions
Explained steps in more detail for bachelor students with limited experience in modeling. Title of the script was revised to reflect the possible outputs more clearly.