The Problem Driven Iterative Adaptation (PDIA) toolkit was developed by Building State Capability (BSC) at Harvard University, and “researches strategies and tactics to build the capability of organizations to implement policies and programs”.
PDIA is a step-by-step approach which “helps you break down your problems into its root causes, identify entry points, search for possible solutions, take action, reflect upon what you have learned, adapt and then act again”.
I would place it in a family of ‘complexity literate’ approaches, which put those insights into practice (others include: Cynefin Framework and Wasafiri’s SystemCraft).
The PDIA approach rests on four principles:
- Local Solutions for Local Problems.
- Pushing Problem-Driven Positive Deviance.
- Try, Learn, Iterate, Adapt.
- Scale through Diffusion.
The PDIA process:
- Constructing your problem. For the PDIA method a good problem: matters to people, can be broken into smaller causal elements, allows for sequenced responses that can be locally-driven.
- Deconstruct the problem using Five Whys, and a fishbone or Ishikawa diagram to visually represent your deconstructed problem.
- Find entry points by understanding the space for change, which is defined by ‘AAA’: the support needed (Authority); how much those affected by reform accept the need for change (Acceptance); and what can be done currently (Ability). You are looking for where there is large AAA as a place to start, with second priority to AAa (and the intervention is: what can be done to increase that small A?).
- Crawl through the design space for possible solutions. The space is Feasible here vs Technically correct somewhere. ‘Crawl’ as in: starting from where you are (A), avoid the siren of just doing what works elsewhere (D), increase capacity by trying better stuff that could be done now with some effort (B) and stuff that are already working but not the norm (C).
- A. Existing Practice (Feasible but Technically not good enough). Scrutinise, learn from, and, maybe, improve.
- D. External best practice (Technically works elsewhere but not feasible here). Identify, translate, adapt, diffuse.
- B. Latent practice (Feasible but Technically mediocre). Provoke through rapid engagement, codify and diffuse.
- C. Positive Deviance (Feasible and Technically correct, but rare). Find, celebrate, codify and diffuse.
- Build and maintain authority for each iteration.
- Design first iteration. Try a number of small interventions in rapid cycles. [Like a Sprint in the Agile methodology.]
- Learn from your iterations. PDIA has no separation between the design and the implementation phase of solving complex problems. On-going iterating is part of the problem-solving. Targeted actions are rapidly tried, lessons are quickly gathered to inform what happened and why, and a next action step is designed and undertaken based on what was learned in prior steps.
I like the overall cycle and specific exercises within it (especially the diagrams above). I saw a family resemblance to the Boisot Social Learning Cycle (as used by IFF), on finding and scaling the Positive Deviance practices.
Least substantive objections: why is the process flow counter-clockwise?
More seriously, there doesn’t seem to be a part of 5 or 6 about selecting which interventions to try (given limited resources). Also, the toolkit illustrates success coming after 6+ iterative cycles. So, I would have liked more attention on designing subsequent iterations, and keeping that going.
Also, there is an implicit assumption that people can be creative, when prompted with all these questions. But maybe the context and culture provides too much of a constraint, or maybe, without practice, coming up with a new thing to do is a big ask.
Still, I can see applying this as part of the ‘Act’ stage in Imagining Influential Trajectories.
This post is part of the #ReadingNotes series, see here for more (including format and use of bulletpoints).
Citation: ‘PDIA toolkit: A DIY Approach to Solving Complex Problems’ Version 1.0 published in October 2018. Editors: Salimah Samji, Matt Andrews, Lant Pritchett and Michael Woolcock
Source: https://bsc.hks.harvard.edu/tools/ [23/04/2024]
David Bent 