Representing complexity in outcomes models - 'simple but not simplistic’ principle

principlesbutton_med

Academic references to outcomes theory

screenshot_1915_med


The principle

Visual outcomes models used in outcomes theory should be fit-for-purpose for their audiences. Small overly simplistic one page models are usually inadequate. On the other hand, overly complex models with too high a cognitive load (the mental effort required to understand them) may alienate high-level decision-maker and stakeholder audiences. At this early stage in socializing the use of visual models as the basis for real-time strategic discussions it is essential that stakeholders do not reject the visual-based approach as 'overly complex'. The attempt should be made to keep models very user-friendly by being 'simple but not simplistic'. Complexity can slowly be increased over a number of years as audiences' appetite; skills at interpreting complexity; visualization conventions; and the sophistication of visualization software advances further. However, this principle does not preclude building more complex visual models for back-room technical, rather than general stakeholder presentations and discussions.

 

The problem

In many outcomes systems, the relationship between the steps being used to achieve outcomes and the outcomes being sought is highly complex. At the moment the attempt is usually made to communicate this complexity through the use of extensive text-and-table-based documents sometimes supplemented by visual models varying in their level of user-friendliness. The cognitive load (the amount of mental effort it takes to understand something) of this largely text-and-table-based approach is too high for most program staff, decision-makers and stakeholders. It also does not provide a useful platform for real-time strategy discussions because of the difficulty of quickly accessing the text-and-table documentation.

Outcomes theory ultimately seeks to provide programs, decision-makers and stakeholders with a much more sophisticated way of thinking about outcomes systems than that possible using present approaches and forms of documentation. This new approach is based on extensive use of visualizations of outcomes models. However this is a staged multi-year socialization process of: 1) getting people increasingly comfortable working with visual rather than just text-and-table documentation; 2) building their skills at interpreting visual models; 3) developing conventions for visualizing outcomes models; and, 4) progressively improving visualization software and platforms as technology advances and users become better at using this technology. 

There are four problems which can arise in regard to complexity and visual modeling in the current stakeholder communication environment. The first is that there is a risk at this early stage in the process of using this new visual way of working that the outcomes models that lie at the heart of outcomes theory, end up being represented in a way that has an excessively high cognitive load. The danger is that the visual modeling approach may be rejected by high-level stakeholders because it is seen as 'overly complex'.This will push back the outcomes theory agenda of introducing a heavily visual approach to outcomes work. Technical staff can complexity visual models because of their level of comfort with highly detailed visual representations. However these models can become inaccessible by decision-makers and other stakeholders.

A second practical problem with attempting to include all causal links and feedback loops in visual outcomes models being used for stakeholder communication is the length of time this takes. If a group is building a model in real-time it is usually not possible within the time available for all the causal links and feedback loops to be incorporated. There are limits to the amount of time stakeholders will allocate to drawing an outcomes model in real-time. Real-time modeling with stakeholders is the approach encouraged within outcomes theory because it is the most effective way of getting stakeholders to buy in to the model as representing the reality of their program. Insisting on adding all of the causal links and feedback loops within the model while building it slows down the process and difficulty of model construction. This often results in it being no longer practical to construct models within the length of time available for most groups to work on such models and can lead to stakeholders rejecting the process as too complex. It is very important to avoid this happening at this stage in introducing a heavily visual approach to stakeholders.

The third problem is that there is likely to be different levels of detail to which causal links and feedback loops can be specified in different parts of an outcomes model. If all known causal links and feedback loops are modeled in one part of the model and some are not modeled in another part of the model because of current uncertainty about causal processes, the visual model misrepresents the true situation in regard to complexity within some parts of the model.

The fourth problem is that it is important that any visual outcomes model invites stakeholders to engage in amending the model. From a psychological point of view, a visual model which is densely populated with line and arrow links discourages the viewer from changing the model because of the visual chaos which is produced when the boxes are manipulated within the model.   


The solution

Therefore at the current time the suggested approach to presenting visual outcomes models is to make them 'simple but not simplistic'. It is counter-productive to overwhelm decision-makers and stakeholders with too much complexity too soon. The objective at this stage in the long-term socialization process is to just get them comfortable working with large-scale visual models rather than the traditional text-and-table documentation. The complexity of the models being used can be increased once large numbers of stakeholders in different sectors are comfortable with the new fully visual approach. Below is an example of a DoView outcomes model with all of the causal links and feedback loops included within it. Following this is a user-friendly DoView outcomes model which is much more suited for working with a group. However, this does not mean that diagrams such as the first one cannot be developed for back-room technical work. It is just that at the moment they are generally not the most effective way of using outcomes models for communication and strategy discussions with decision-makers and stakeholders.