The topic of complexity has appeared several times over the last few weeks. Brian Pittman wrote about it in an AEA365; Charles Gasper used it as a topic for his most recent blog. Much food for thought, especially as it relates to the work evaluators do.
Simultaneously, Harold Jarche talks about connections. To me connections and complexity are two side of the same coin. Something which is complex typically has multiple parts. Something which has multiple parts is connected to the other parts. Certainly the work done by evaluators has multiple parts; certainly those parts are connected to each other. The challenge we face is logically defending those connections and in doing so, make explicit the parts. Sound easy? Its not.
That’s why I stress modeling your project before you implement it. If the project is modeled, often the model leads you to discover that what you thought would happen because of what you do, won’t. You have time to fix the model, fix the program, and fix the evaluation protocol. If your model is defensible and logical, you still may find out that the program doesn’t get you where you want to go. Jonny Morell writes about this in his book, Evaluation in the face of uncertainty. There are worse things than having to fix the program or fix the evaluation protocol before implementation. Keep in mind that connections are key; complexity is everywhere. Perhaps you’ll have an Aha! moment.
I’ll be on holiday and there will not be a post next week. Last week was an odd week–an example of complexity and connections leading to unanticipated outcomes.
Thank you, I’ve recently been searching for information about this subject for a while and yours is the greatest I have found out till now. But, what in regards to the conclusion? Are you positive in regards to the source?
I would refer you to the sources I’ve cited. Changing a logic model before you implement a program may save you a lot of headaches and time delays later. It doesn’t take long to craft a preliminary logic model–any model, however brief, is better than no model.