January 11, 2018
I am thinking of getting together a panel presentation on this topic for The European Evaluation Society conference in Greece 1-5 October 2018: http://www.ees2018.eu/abstract-submission-guidelines.htm. Something like this:
Visualisation of Theories of Change
This panel is at the intersection of two trending topics in evaluation.
Data visualisation: evaluators have to present their results in an accessible way, and good visuals can complement a narrative and provide new insights.
Theories of Change, whether presented as visuals or text, are central to many modern understandings of evaluation and its tasks.
But what happens at their intersection? How do we, should we, visualise Theories of Change?
- Can we apply what we have learned from visualising data to visualising theories?
- How do informal approaches to sketching/visualising the processes and results of participatory workshops fit into this picture?
- If graphical Theories of Change don’t (primarily) visualise data, what do they represent? Causal networks (Pearl, J. (2000). Causality: Models, reasoning and inference. Cambridge Univ Press.)?
- Could visualisations of causal connections be useful to evaluators apart from presenting formal, complete Theories of Change?
- Is there a role for a common visual language or alphabet in the visualisation of data? What about in the visualisation of Theories of Change? Could there be a common language across both? What would be the benefits and drawbacks?
- How can adaptive mechanisms be represented graphically?
- How can we deal with interactive, dynamic Theories of Change which allow desk-based experiments that let us see the consequences of different variations in the design of theories of change?
This panel welcomes both practical presentations of tips and case studies as well as more theoretical papers.
October 18, 2017
AEA program listing.
A short post here on this blog
The book is not an easy read
But there is an Epilogue right at the back of the book which provides a great summary. You can read it in an hour or two and it will definitely change your thinking about evaluation.
Free web apps:
Here is the Cheat Sheet as slides:
September 21, 2017
Sometimes in Theories of Change it is very useful to be able to present simple developments over time, for example an increase or decrease or stagnation. Theories of Change are not famous for their ability to deal with time at all, for which they are often criticised. I’ve written before and here and here about the problem of Variables which may or may not exist across a whole project duration - here, I’m tackling the problem of reporting changes in the value of a Variable when it does stretch across the whole duration of a project.
Tufte’s Sparklines can do this well.
But I am looking for the simplest kind of display which is readable on Variables within a Theory of Change or Logframe even from afar. Sparklines, simple as they are, are still too detailed for a crowded diagram with many Variables which can get quite small. Plus, they suggest a degree of accuracy which I’m not looking for - I just want to be able to indicate the roughest rises and falls.
Plus you can’t easily cut and paste sparklines into a report, say.
So, enter Unicode Sparklines - the Flintstones of data display for Theories of Change.
So for theorymaker.info and the new interactive version I’ve implemented simple Unicode Sparklines just with these five blocks: ▇ ▆ ▅ ▃ ▂. To produce them at theorymaker.info, when typing the names of your Variables, you just type
!1 produces a small bar and so on up to
!5 which produces the tallest bar. To get a sequence of bars you can write for example
!1!2!3 but abbreviating to
!123 works as well.
Here’s an example diagram.
Because these five little blocks are plain Unicode, so you aren’t tied to Theorymaker. You can write them directly in text, like this: temperatures have been changing lately, ▂▅▇. You can use them directly in Office documents - find them in your “insert/symbol” dialogue in your word processor or just copy them from this post. In MS Word, there’s a special trick - type 2582 and then immediately type Alt+x - that will get the smallest bar. For the others you have to type 2583, 2585, 2586 and 2587. Primitive, but expressive - like the Flintstones.
Funnily enough, the Unicode implementation of these kinds of block isn’t very consistent - I had to mess about to find just these five which look consistent everywhere, which explains why unfortunately the height intervals aren’t quite even.
(Turns out the idea of Unicode sparklines isn’t new: see here.)