Slicing up the world September 25, 2007
Posted by Paul Duignan in : DoView, Outcomes models , trackback
The other day I was talking to Laurie Porima, an evaluation colleague, about building outcomes models with DoView outcomes software. We started discussing the best way of dividing an outcomes model (program logic, results chain, strategy map, ends-means diagram) up into sections. Obviously the real world of causality is not divided up into tidy conceptual pieces suitable for humans to quickly grasp. Many traditional approaches to outcomes models do not have to grapple with this problem because they limit themselves to a single printed page. There’s no question of having to divide the model up in any particular way. However if you allow your models to be larger than a single page – which seems to me to be essential if we are to develop models which can even start to do justice to the causal complexity of the world we’re trying to operate in – you need to think about how you can visualize this.
One approach is to just let the canvas you draw on in your drawing software to get bigger and bigger. This does allow for a more complex model. However, there remains the problem of how people can actually see this model, particularly in a group session when the model is being projected. Equally important there is the question of helping people to quickly grasp the content of the model. A set of thirty or forty steps and outcomes in a model which are not grouped in any particular way is very hard to get our head around. DoView uses the concept of ‘slices’ to try and address this important problem.
A slice is both a single DoView ‘page’ or ‘diagram’ and also a metaphor for a cut through the world of causality which you’re trying to model. When teaching workshops on drawing outcomes models I often suggest that we can imagine the complex and interrelate causal relationships behind the outcomes we’re trying to achieve as being like an undifferentiated sphere – say like an orange. Our job in building an outcomes model is to cut some ‘slices’ through that sphere which make conceptual sense to ourselves and to the stakeholders to whom we’re trying to communicate our model.
There’s no particular right or wrong about how many slices you should take or the group of outcomes they should cover. However, each slice should make some conceptual sense to someone looking at your outcomes model. For instance, after having built hundreds of outcomes models what I notice often emerging for many programs are the following slices:
- A national level policy and action slice – covering the steps national level stakeholders are trying to bring about (in a corporate setting this could be seen as a ‘head office’ kind of slice).
- A ‘locality’ slice – covering the steps which need to occur more at a single locality level.
- An organizational or institutional slice – covering the steps which need to occur within an individual organization or institution.
- An individual slice – covering the steps which need to occur in regard to an individual.
- A ‘mechanisms’ slice – this one sometimes emerges, it’s harder to describe but it covers steps in the outcomes models which have not been covered sufficiently in the other slices.
Depending on the project, there can be a range of other slices, e.g. a family slice and even slices on certain themes e.g. a sport and recreation slices, a business development slice etc. We’ll be posting a lot of examples of outcomes models at www.outcomesmodels.org so in due course you’ll be able to check out various examples of how models on different topics have been sliced up.
In developing slices it’s important to realise that they do not need to be totally exclusive of each other. If you force them to be exclusive you’re forcing a ‘siloed’ outcomes model where a lower level outcomes is only allowed to contribute to a single (or a limited number of) higher level outcomes. I’ll talk about forcing a ‘siloed’ model and other outcomes model sins in a later blog posting (you can see some discussion of this in the Easy Outcomes outcomes model guidelines). DoView lets you put ‘clone’ steps on as many different slices as you think are needed to represent the real-world of causality you’re trying to model. A clone step is a step or outcome which is a ‘live copy’ so it represents exactly the same step or outcome when it appears on different slices. Working out how to slice up the outcomes world for a particular project is an art as much as a science, but it’s very important to get a set of slices which stakeholders can quickly relate to if you’re wanting to have a model which they’ll be able to quickly read and understand.
Paul Duignan (outcomesblog.org)
Comments»
[...] is the ’slice globbler’. In DoView each diagram is called a slice (see my blog entry on slicing up the world to build outcomes models). There are some words used in some context which globble up large parts [...]
[...] are the slices I identified in an earlier post – national, locality, organizational/institutional, individual and one or more [...]