While I believe the Model for Improvement provides a useful general foundation for improving any situation, specific problem-solving methods that call for standardization are particularly useful. These methods can help people to act to reduce defects and build reliable performance into on-going operations.
Improvement and problem-solving are linked but not identical in application
You can pose any problem as an improvement opportunity: a partial or full solution to a problem will yield a better state of affairs than the unsolved problem. Hence a solved problem will have improved the system in which the problem arose.
You can also pose any improvement opportunity as a problem: in improvement, you seek to get closer to an aim, starting from where you are. Hence, you start with a gap between where you are and where you want to get. That gap is the problem.
So it looks like improvement and problem-solving are equivalent in the sense that you can express any problem as an improvement opportunity and any improvement opportunity as a problem.
The Model for Improvement is an algorithm to improve any situation. If you accept that improvement and problem-solving are equivalent, the Model for Improvement can be applied to any problem.
Does that mean the Model for Improvement is the best algorithm for solving any problem? No. In some settings, it may be more efficient to tackle problems directly.
Types of Problems
Gerry Nadler in The Planning and Design Approach (Wiley: 1981) — abbreviated PDA — describes seven types of problems, aligned with a list of seven “human purposeful activities”:
1. Assure self-preservation and survival of the species: self-preservation
2. Operate and supervise an existing solution or system: operate and supervise
3. Create or restructure a situation-specific solution or system: plan and design
4. Search for generalizations: research
5. Evaluate performance of previous solutions or other purposeful activities: evaluate
6. Gain skills or acquire knowledge about existing information and generalizations: learn
7. Experience leisure: leisure.
As Nadler points out, the “types are not mutually exclusive: each may be involved with, and depend on, the other. For example, successful planning and design frequently requires, at various points in a project, research, evaluation, learning, and operating and supervising.” (p. 19, PDA)
The Model for Improvement can be applied to each of the seven types, more or less efficiently.
Operating and Supervising; Planning and Design
Two activities from Nadler’s list account for much of the work by organizations: Operating and Supervising (O&S) and Planning and Design (P&D).
“O&S concerns systems and solutions that people participate in routinely, expecting fairly standardized results.” (p. 19 PDA)
“Planning and design activities result in custom-made solutions, policies, and designs, that restructure existing systems or create new ones….P&D is concerned with imagining, designing, and implementing new and restructured systems and solutions, O&S with maintaining them. The latter stresses standardization and routine, the former flexibility and innovation.” (p. 20, PDA)
The Model for Improvement works well right off the shelf for Planning and Design activities as defined by Nadler.
However, Model for Improvement applied to Operating and Supervising problems can cause users to stumble. Here’s why: Holding the gains is implicit in the Model for Improvement but improvers usually need explicit guidance.
On the other hand, there are various frameworks developed to solve problems in O&S settings that give explicit advice about holding the gains--for example, Lean’s A3 problem-solving, Six Sigma's Design-Measure-Analyze-Improve-Control model, and the QC Story (Chapter 10, Statistical Methods for Quality Improvement, AOTS, Tokyo 1985.)
Standardized work is one way to hold the gains of a proposed solution, a systematic approach to teaching, monitoring and assuring reliable work practices. A useful reference for healthcare applications: Getting to Standard Work in Healthcare (2012) by Patrick Graupp and Martha Purrier, CRC Press, Boca Raton, FL.
A hint that you may be working in an O&S setting is whether it feels natural to characterize problems in terms of defects. A defect is a specific type of problem, an undesired result of a job, as succinctly described by Kume and Takehashi, authors of the QC Story chapter in the AOTS book cited above.
In work with colleagues these days, I now try to understand the type of situation or problem they’re tackling. I use the Model for Improvement as my fundamental mental framework across all settings but I especially appreciate and support the application of specific problem-solving frameworks in operating and supervising situations.
My 15 year-old daughter Grace got fired up by Hans Rosling’s first Ted talk. She took up the challenge of learning a little bit of R code to build a web display using Google’s Motion Chart for her final European History project, June 2015. Her project display involves time series of population, infant mortality, and per capita GDP for nine countries, 1820-1992.
This was a real data analysis problem. For Grace, most of the work involved gathering the data, getting it into the right format for display and thinking about the data limitations. As she worked to understand the meaning of the numbers, I got to trot out Lord Kelvin’s advice, "the more you understand what is wrong with a figure, the more valuable that figure becomes.“
I showed Grace how to insert the R code into a Shiny app and then open a shinyapps.io account to share with her class. (Here’s the link for her project.)
The overall pattern is decline in infant mortality and increase in wealth from the start of each country’s series through 1992.
Interact with the display for individual countries and you can see the impact of World Wars, the Great Depression and the break up of the Soviet Union. For example, the Netherlands experienced a sharp drop in estimated income and increase in infant mortality at the end of World War II.
Grace used population and constant dollar GDP per capita data from the Madison historical series.
Infant mortality series--deaths before age 1 per 1000 live births--came from a physical book, International Historical Statistics: Europe 1750-1993 (4th edition, 1998) by B.R. Mitchell, London: Macmillan Reference; New York, N.Y.: Stockton Press, accessed at the University of Wisconsin-Madison research library. That source meant a bit of time reading and typing numbers into a spreadsheet, a throwback to 20th century data analysis.
We found that the Google chart element sometimes loads slowly--be patient and if you run out of patience, hit the reload button. Since the Motion Chart is rendered in Flash, you can use the free browser download from Puffin if you want to see the chart on an iOS device.
Grace allowed me to share her code, I have posted it here on GitHub.
David Leonhardt in today's New York Times invites readers to play a little game--guess a rule for a sequence of three numbers. Here's the link.
Invented by psychologist Peter Cathart in 1960 ("On the failure to eliminate hypotheses in a conceptual task", Quarterly Journal of Experimental Psychology, 12, 3, 129-140, link here), I first learned the game from my colleague Jerry Langley in 2001. Jerry uses it to teach people about the importance of testing ideas and predicting outcomes, as motivation for learning the Model for Improvement.
Leonhardt usefully discusses the game in the original context of "confirmation bias"-there seems to be a natural human tendency to seek confirmation of views we already have and to avoid disconfirmations.
The upside of the bias is that we're sometimes right and a quick confirmation is all we need to make progress. The downside of the bias is there are few if any universal rules related to human systems so at some point our mental model will fail. If we never determine situations where our views fail, we're quite constrained in our ability to learn and develop systems that do work, in the appropriate circumstances.
"When you seek to disprove your idea, you sometimes end up proving it — and other times you can save yourself from making a big mistake. But you need to start by being willing to hear no. And even if you think that you are right, you need to make sure you’re asking questions that might actually produce an answer of no. If you still need to work on this trait, don’t worry: You’re only human."