Page 263 -

262 O. Barreteau et al.

of the modelling process. Rouwette and colleagues analysed 107 cases of such
experiments and proposed a number of guidelines to facilitate consistent reporting
on participatory modelling exercises. These guidelines focus on three categories:
context, mechanisms and results (Rouwette et al. 2002). The second category
focuses predominately on preparation activities and description of meetings, along
with factual elements and the modelling process.
This category of participatory modelling deals with the expectations identified in
the first section in the following manner:
– The participation arena is constituted of a rather small or medium size well-
identified group. The structure of the interaction is rather global: debates
tackle the whole model, and participants are supposed to be concerned by the
model entity as a whole. The connections may convey information on the tacit
knowledge of stakeholders, as well as on their purposes. This is still very diverse
among the experiments. The group of stakeholders is mobilised within specific
events, workshops, which might be repeated. The aim is to feed the model but
also to increase the probability of the use of the models produced.
– The process is predominately controlled by the modellers.
– The format of information is generally not well formalised, even though tech-
niques, such as hexagons brainstorming or causal diagrams (Akkermans 1995),
appear to organise the knowledge brought by stakeholders. This low formalisa-
tion allows the issues related to stakeholder diversity to be tackled and alleviated
in the problem framing phase, but it leaves a large place to the modellers’
interpretation.
Management flight simulators or microworlds constitute a complementary tech-
nique, which focuses more on the stages of use and simulation outcomes analysis,
even though this technique may also be used in a design stage to elicit tacit
knowledge. A key characteristic of this type of technique is to encourage learning
by doing. Participants, who might be the clients or other concerned people without
any formal relation to the modelling team, have to play through a simulation of
the model. Martin and colleagues have used this technique to validate a system
dynamics model on the hen industry (Martin et al. 2007). Participants were asked to
play with some parameters of the model.
When used to elicit knowledge, microworlds attempt to provide events that are
similar to those that participants already face or are likely to face in their activities
related to the issue at stake in the model. Le Bars and colleagues have thus developed
a game setting to lead farmers to understand the dynamics of their territory with
regard to water use and changes in EU common agricultural policy (Le Bars et al.
2004). In flight simulator experiments, interaction between stakeholders and the
simulation model is structured around future users of the model or people whose
stakes are represented in the model, with a slightly deeper connection than with
previous group modelling building approaches. Participants are asked to deal with
parameters of the model and are framed in the categories used in the model.
There is no a priori differentiation among participants. The connections convey
information about the object from the model to participants. It also conveys the