Statistical Design and Control: New Book by Kieron Dey

“The advantages of statistical design and control, using only techniques available in the literature, are self-evident.  The basic idea is simple.  The main disadvantage is that it requires some technical skill and management support.  Those might also be argued as advantages over the competition.”

“The essential elements in application include establishing homogeneity and stability in both the process and the measurement error, although there are workarounds if unattainable.   Statistical control isolates then exploits uncontrolled variation, keeping an eye on the economic aspect.  Developing strong interventions and correct randomization are central to statistical design.  Implementation that would ordinarily be difficult uses statistical control again, as a simple, real-time, feedback-control system.”

(Kieron Dey, Competitive Innovation and Improvement, CRC Press 2014, p. 198)

My colleague Kieron Dey has written a new book that demonstrates the synergy between statistical process control and statistically designed experiments in efforts to improve services and products.  The case studies and extended examples feature mostly service applications rather than the engineering/manufacturing examples that often dominate the applied statistical literature on these topics.

The audience is intended to be anyone interested in improving competitive position.

The first chapter offers a detailed example, improving case management by nurses, which illustrates the main ideas of the book.   There are no mathematical formulas, following direct advice of George Box, as Kieron states in his acknowledgements.

The chapter 1 case is worth the price of the book.

Studying the experiment—a screening design that tested 19 interventions by 20 nurses--Kieron estimated a 25% decrease in admissions rate of the patients being managed.  About half the interventions were meaningful and were implemented.   At the end of the implementation period reported in the case, the nurses accomplished a 15-20% decrease in admissions, while simultaneously increasing their case load by 50% to serve more patients by adjusting the frequency of calls to patients.

Impressive results, skillfully described.

Subsequent chapters give more details and examples, as Kieron explains the steps used in the first case, still with a minimum of formulas.  The few formulas that do appear are helpfully broken out of the main text and can be considered by technically minded people or skipped by readers interested in the core lessons.  Here is a link to the publisher’s web page where you can see the table of contents.

In the first five chapters, there will be few surprises in terms of the statistical content if you have experience and knowledge of control charts and screening designs.  You should still keep your eyes open for Kieron’s advice and insights—for example, the important practical advantage of having “many” interventions under test simultaneously.  This attribute of screening designs reduces the biasing effect that arises when people attend to a special situation (related to the Hawthorne Effect), which can lead to over-estimates of impact when changes are made part of standard practice.

Chapter Six summarizes 12 cases, giving the reader a sense of the range of topics that can be tackled with advantage by the book’s methods.

Chapter 7 outlines Kieron’s advice on simultaneous design, by which he means the strategy to run concurrent screening experiments on appropriate test units.   He starts with a relatively simple example that nests one screening design within another--interventions to improve science education of students--and then moves to a more complicated example involving sales channels.   Kieron outlines how he thinks about simultaneous designs.   I plan to re-read this chapter and think more about his advice.

In Chapter 9, Management Improvement and Innovation, Kieron gives his perspective on the management context for his approach.  He presents an outline of 20 steps to tackle problems like those he has described earlier in the book.

Kieron’s book presents a range of useful insights and the case examples should inspire you to try your hand.  He offers specific advice and encouragement to help many more people and organizations get results through the powerful combination of statistical design and control that he has mastered.

Value in Healthcare

Deming's Chain Reaction: Better Quality, Lower Costs, Higher Productivity