Systems Thinking, Logic Models and Evaluation
The American Evaluation Association conference is on right now in San Antonio and with hundreds of sessions spread over four days it is hard to focus on just one thing. For those interested in systems approaches to evaluation, the conference has had a wealth of learning opportunities.
The highlight was a session on systems approaches to understanding one of evaluation’s staples: the program logic model.
The speakers, Patricia Rogers from the Royal Melbourne Institute of Technology, and consultants Richard Hummelbrunner and Bob Williams spoke to the challenges posed with the traditional forms of logic models by looking at the concepts of beauty, truth and justice. These model forms tend to take the shape of the box model (the approach most common in North America), the outcome hierarchy model, and the logic framework, which is popular in international development work.
The latter model was the focus of Hummelbrunner’s talk, which critiqued the ‘log frame’ approach and showed how its highly structured approach to conceptualizing programs tends to lead to a preoccupation with the wrong things and a rigidity in the way programs are approached. They work well in environments that are linear, straightforward, and in situations where funders need simple, rapid overviews of programs. But as Hummelbrunner says:
Logframes fail in messy environments
The reason is often that people make assumptions of simplicity when really such programs are complicated or complex. Patricia Rogers illustrated ways of conceptualizing programs using the traditional box models, but showing how different program outcomes could emerge from one program, or that there may be the need to have multiple programs working simultaneously to achieve a particular outcome.
What Rogers emphasized was the need for logic models to have a sense of beauty to it.
Logic models need to be beautiful, to energize people. It’s can’t just be the equivalent of a wiring diagram for a program.
According to Rogers, the process of developing a logic model is most effective when it maintains harmony between the program and the people within it. Too often such model development processes are dispiriting events rather than exciting ones.
Bob Williams concluded the session by furthering the discussion of beauty, truth and justice, by expanding the definitions of these terms within the context of logic models. Beauty is the essence of relationships, which is what logic models show. Truth is about providing opportunities for multiple perspectives on a program. And a boundary critique is a an opportunity for ethical decision making.
On that last point, Williams made some important arguments about how, in systems related research and evaluation, the act of choosing a boundary is a profound ethical decision. Who is in, who is out, what counts and what does not are all critical questions to the issue of justice.
To conclude, Williams also challenged us to look at models in new ways, asking:
Why should models be the servant of data, rather than have data serve the models?
In this last point, Williams highlights the current debates within the knowledge management community, which is dealing with a decade where trillions of points of data have been generated to make policy and programming decisions, yet better decisions still elude us. Is more data, better?
The session was a wonderful puctuation to the day and really advanced the discussion on something so fundamental as logic models, yet took us to a new set of places by considering them as things of artful design, beauty, ethical decision making tools, and vehicles for exploring the truths that we live. Pretty profound stuff for a session on something seemingly benign as a planning tool.
The session ended with a great question from Bob Williams to the audience that speaks to why systems are also about the people within them and emplored evaluators to consider:
Why don’t we start with the people first instead of the intervention, rather than the other way around like we normally do?