I’ve got a theory (of change)

(video: I’ve got a theory – from Buffy the Musical)

Duncan Green just wrote an excellent and detailed post “What does a theory of change look like” which examines the reasoning behind having a theory of change and outlines a range of different approaches and change theories that are in use.

I was planning on posting a few thoughts as a reply – but it got a bit long so I’ve written a blog post instead!

As Duncan mentions, theories of change (toc for short) are all the rage in development now and so you better be able to articulate yours. Yet it’s worth ruminating for a moment on what theories of change actually are and what they are (and are not) good for.

Theories of change are just that – theories. In other words they are mental models we use to explain reality since our brains and our current level of knowledge and understanding of the way the world works are limited. Theories are useful because they allow us to simplify and therefore “understand” the problems we are trying to tackle, and use this predictively to help us decide what to do next. But it’s important to acknowledge that they do not really explain the world – but rather give us a manageable snapshot that we can handle and which is useful to us, and is a close enough approximation to reality that we can overlook those areas where it is a bit off.

This is particularly important when thinking about “wicked” or “complex” problems in development. While we can recognize that we are not truly able to establish linear causal relationships between inputs and outcomes that explain and predict everything that is happening, a theory of change can give us a useful and relatively simple approximation of what is happening in part of a complex system – in particular the part we are planning to do our project on. This is appealing since it breaks down a potentially intractable whole into pieces we can tackle.

But how do we develop our theory of change? I’d argue that the choice of theory we use is, in practice, heavily influenced by our background – our beliefs about the world which can come from our upbringing, or our professional training, or the prevailing organizational culture where we work. For example an economist is more likely to use econometric based models, an anthropologist anthropologically based ones, a religious person might well see a divine presence behind the change.  The point here is that the choice of theory is not something that is impartially decided based on the nature of the problem we are addressing, it is also heavily tied up with our own identity and beliefs.

This means that it can be difficult to get a diverse group of stakeholders to see a problem the same way and get them to agree that a particular toc is the best one to use in any particular circumstance. That said, having an articulated toc is an important way to make the reasoning behind a programme design more explicit to others – and to create a dialogue around different potential ways to look at and address a problem in order to surface and address the different world views that often exist around  a particular development challenge.

But it also points to the need for the toc not only to be useful in terms of helping make decisions to those who “believe” it – but also on the need for the toc to be sufficiently simple, and plausible in order to to be able to be understood by and to be able to convince others. This means that there can be a trade off between having a technically thorough model, and having one that is sufficiently straightforward that you can convince others with differing perspectives to accept it.

Another important consideration is that any toc is useful only insofar as it can be used predictively i.e. to help inform planning and decision making for the future. Although it’s possible, and tempting, to develop a model that fits the data when looked at after the event retrospectively and thus believe it can be applied in the future – this retrospective coherence can be misleading and needs to be thoroughly tested in it’s ability to foresee the future with any accuracy (in stock market parlance – past performance is not necessarily a guarantee of future returns).

From a development practitioner’s perspective, it’s probably wise to be flexible in the use of theories of change  – as Duncan says  -to be prepared to have more than one – in particular to be able to choose tocs that i) seems to work based on the evidence we have ii) are ones around which you can get the different stakeholders to agree, or at least agree to try. They need to be both explanatory AND persuasive. We might also need to consider two or more different explanations concurrently and test them out against the evidence – allowing for the ambiguity both between the theories but also that the results might not clearly support one theory over another.

In some instances you might not need a theory of change at all. Most large aid programmes will require one in order to justify, at least to the donor, why we are investing our money where we are – but many development actors – in particular individual innovators and entrepreneurs might not need a well developed theory. Maybe all that is needed is an idea, a recognition that they can offer something that doesn’t already exist, without a clear idea of its potential benefit. The test of whether the idea is any good or not will be whether people in developing countries make use of it or not, and the benefit will be whatever the benefit was that led to adoption.  In other words, while aid agencies, governments and researchers will find it important to have a theory of change- we shouldn’t necessarily impose this “requirement” on all our counterparts, but rather leave some space for unplanned, untheorized entrepreneurialism.