Scriptapedia/Graphs over Time

Graphs Over Time

This script is used to engage participants in a group model building session in framing the problem, initiating mapping, eliciting variables, and gathering input to decide the reference modes for the workshop or modeling process. It is performed at the beginning of a group model building session as it is a springboard for discussion about the problem to be modeled.


Best practices

Primary nature of group task



Preparation time: 15 minutes

Time required during session: 45 minutes

Follow-up time: 0 minutes


  1. Stacks of 8.5x11 white paper with X and Y axes drawn on them
  2. Large blank wall (8'x10')
  3. Thick tipped markers
  4. Blue painter's tape, glue sticks, or tacks
  5. Camera or other method to capture the graphs




Candidate variables for the dynamic model or the map (informal causal map/modeling)


  • Facilitator who has some experience with SD to work with the group
  • Modeler who listens to what is being graphed, the way people are talking about the graphs, and who is also be able to conceptualize the early seeds of system structure
  • Wall-builder with little or no experience in SD who will cluster graphs and talk about themes
  • Recorder to document the session through written notes and photograph the clustered graphs
  • Runner (optional) to carry the graphs from the participants to the facilitator if the group is large


  1. Based on group size, decide whether to break participants into subgroups. In smaller groups (N<10), allow individuals to work and present independently. In larger groups (N >10), divide participants into subgroups of roughly 10. Ask the subgroups to sit together.
  2. The modeling team hands out sheets of white paper to each participant.
  3. The facilitator gives an example of how to draw a graph over time, carefully labeling the X-axis as “Time,” and adding a start time, end time, and the present time indicated with a vertical dashed line. The Y-axis is labeled with a variable name. The facilitator then sketches the behavior over time.
  4. The facilitator then asks participants to draw one variable over time per piece of paper. The participants should be given the option of including hoped for behavior, expected behavior ("business as usual"), and feared behavior on the same graph.
  5. The facilitator and wall-builder walk around and help participants with the task as needed. Allow 15 minutes, or until the group runs out of steam, to complete the task.
  6. Reconvene as a large group:
    • If N<10, instruct participants to arrange their graphs in a stack with the most important graphs ("the best stuff") on top. The facilitator takes one graph at a time from each participant, holds it up in front of entire group, and asks the participant to talk about it. Ask for participants to share the “best stuff” first. Clarify timescale, variable names, etc.
    • If N>10, instruct subgroups to share their graphs with each other and choose the ones they think are most important. They should then arrange their graphs in a stack with the most important ones on top. The facilitator then goes to each subgroup and holds the first graph they have selected up in front of entire group. The subgroup spokesperson talks about the graph. Ask subgroups to share the “best stuff” first. Clarify timescale, variable names, etc.
  7. The facilitator then hands the graph to the wall-builder.
  8. The facilitator repeats steps 6 and 7 with each participant or subgroup, taking one graph at a time until all graphs are shown or time has run out. Finish by asking if any participant has something else that really ought to be shown.
  9. During steps 7-8, each graph is posted on the wall. The wall-builder tries to cluster the graphs meaningfully on the fly based on themes and variables.
  10. The facilitator asks the wall-builder to explain the clusters of graphs on the wall. The wall-builder tries to summarize dynamics that help to characterize the problem that emerges from the participants’ graphs.
  11. The facilitator enables the participants to talk about the clusters and the characterization of the problem they imply.
  12. Consider labeling the clusters based on themes or related variables. There is potential for the modeler to close by highlighting the beginnings of feedback thinking in the dynamic problem.

Evaluation Criteria

  • Interesting, self-sustaining group discussion about clusters (Example 2) described by the wall-builder
  • Meaningful clusters identified
  • Graphs tend to converge to a clear dynamic problem
  • Some key dynamic variables emerge from reflecting on the graphs and thematic clusters
  • Modeling team can begin to see key stocks and perhaps important feedback loops
  • Members of the group appear to have a better understanding of the problem being modeled

Example 1: Graph over Time
Example 2: Graphs over time


George P. Richardson and David F. Andersen


First described in Luna-Reyes et al (2006).




Andersen, D. F., & Richardson, G. P. (1997). Scripts for group model building. System Dynamics Review, 13(2), 107-129.


  • When participants are discussing their graphs with the facilitator, the wallbuilder may be otherwise engaged in identifying themes and arranging graphs on the wall. When the facilitator passes the graphs to the wallbuilder, they may need to pause and wait for the wallbuilder to be ready to turn away from the wall and receive the graph, and should check to make sure the wallbuilder heard the discussion and understands the meaning of the graph.
  • There is a tendency to want to avoid making discrete "clusters" of themes on the wall. However, experience shows that it is most helpful to participants if the wallbuilder does indeed create discrete thematic clusters, as opposed to having a graph or two serve as a "link" between different themes. Remembering that participants will be able to revise the themes can help alleviate tensions the wallbuilder may feel in making concrete decisions about themes.
  • When clarifying/verifying the meaning of participants' graphs, it can be helpful to think about "rephrasing with meaning added." For example, it might not be helpful to repeat back to the participant exactly what they stated about the graph, but it is important to be sure that the facilitator is interpreting what the participant says correctly.
  • Consider providing 2 examples with different time intervals and different hoped/feared line shapes (i.e. one example might have a hope that increases quickly and a fear that increases slowly, while another example has one increasing hope line, one decreasing fear line, and another fear line that oscillates)
  • Emphasize that this is supposed to be challenging. Our brains don’t usually think this way!