Statistical Thinking: Past, Present and Future 2008 Fall Technical Conference ASQ Statistics Division Invited Session Doug Hlavacek, Ecolab STAT Past Chair October 10, 2008 Agenda • Introduction – Doug Hlavacek, Ecolab • Statistical Thinking - Past – Roger Hoerl, GE CRD • Statistical Thinking - Present – Gordon Clark, Ohio State University • Statistical Thinking - Transition – Robert Mitchell, 3M LSSQ • Statistical Thinking - One Future – Roger Hoerl, GE CRD • Panel Discussion Statistical Thinking Statistical Thinking is a philosophy of learning and action based on the following fundamental principles: ¾ All work occurs in a system of interconnected processes, ¾ Variation exists in all processes, and ¾ Understanding and reducing variation are keys to success. Glossary and Tables for Statistical Quality Control Quality Press, 1996 Statistical Thinking • Emphasizes critical thinking • Different from statistical tools... Not number crunching Questions The Statistics Division first published the official definition of Statistical Thinking in the 1996 edition of the Glossary & Tables for Statistical Quality Control. Statistical Thinking is a philosophy of learning and action based on the following fundamental principles: • All work occurs in a system of interconnected processes, • Variation exists in all processes, and • Understanding and reducing variation are keys to success. Three past chairs of the Statistics Division (Roger Hoerl, Gordon Clark, and Bob Mitchell) will share their perspectives about the past, current, and future of Statistical Thinking. What question(s) about Statistical Thinking do you have for the panelists? Statistical Thinking: The Past Roger Hoerl, GE Global Research W. Edwards Deming • To the best of my knowledge, never used the term “Statistical Thinking” • Taught that statistical concepts apply to management, or anything else – e.g., the “red bead” exercise – Focused on understanding, not the formulas • Gradually made the case that we were missing something beyond statistical methods per se – But didn’t articulate it well, in my opinion • The statistical community continued to focus on the math Ron Snee • More than anyone else, broadly popularized and disseminated the concepts of statistical thinking • Defined statistical thinking in 1986 as “thought processes”, not formulas • Clarified the distinction (synergy) between statistical thinking and methods • Later published introductory business statistics text based on statistical thinking (2002) – Statistical Thinking: Improving Business Performance • The statistical community continued to focus on the math Heero Hacquebord • Afrikaner student of Deming’s • Taught public “Statistical Thinking” courses beginning in 1987 – Was more articulate than Deming, in my opinion • Emphasized managerial implications of the concepts – e.g., the hazards of “managing by the last data point” • The statistical community continued to focus on the math Quote From Tom Pohlen, 3M • Attended Hacquebord’s course in 1988 • “I went into the course thinking that I already knew everything I needed to know about SPC. I came out of the course with a whole new perspective on statistics, looking upon SPC and other statistical applications more as a way of thinking about processes so we can learn how to improve them. I also found that I could never again be satisfied with looking at numbers without graphical analysis.” • Pohlen clearly had a “Damascus road” experience • The Statistical community continued to focus on the math Statistics Division Statistical Thinking Tactical Planning Team • Chartered at a Statistics Division long-term planning meeting in 1994 – Developed “5 year plan” • Published formal definition in 1996 Glossary and Tables for Statistical Quality Control - a seminal event! • Wrote a Special Publication on Statistical Thinking for division members in 1996 • Wrote booklet: “Improving Performance Through Statistical Thinking” (Quality Press, 2000) • Organized several conference sessions to “get out the word” • The statistical community continued to focus on the math Impact of These Efforts • Statistical thinking became part of the vocabulary of statistically-oriented quality professionals – Among this group, there is a realization of the uniqueness of statistical thinking versus statistical methods • While not always recognized, statistical thinking principles became a cornerstone of major improvement initiatives, such as TQM and Six Sigma GMC1 Statistical Thinking: Process Improvement Strategy Gordon Clark The Ohio State University October 10, 2008 Illustration of Process Improvement Strategy • Understand the Process – Ricoh’s Numazu plant produced raw materials for paper copier toner • Resin • Consistent quality & volume Collect data on key input, process and output measures Yield = Actual Output Theoretical Output Why is Yield above 100%? Analyze Process Stability Special Cause? • Mechanical problem was special cause – Fixed Evaluate Process Capability • Investigated customer needs for batch output quantity – 4300 kg ± 5 kg Analyze Common-Cause Variation Study Cause & Effect Relationships • Extraction of 2nd phase volume – Resin remained in tank after dividing phase – Line B had less material than line A – Changed dividing procedure • Data showed no detectable difference between batch sizes • Implement change – Variation in output quantity reduced but still too large Solvent Feed Ratio Potential Cause • No relationship should exist • Ratio measurement affected by time solvent sat in tank • Change implemented • Output relationship with feed ratio disappeared • Variation still too high Weighing Process Potential Cause • Found problems affecting weighing process accuracy – In-process (manual) – Final (automatic) • Problems corrected • Change implemented – Output variation met tolerance Output Control Charts ST or Hoerl-Snee Process Improvement Strategy • References – – – – Hoerl, R. W. and R. D. Snee (1995). Redesigning the Introductory Statistics Course. Madison, Wisconsin, University of Wisconsin, Center for Quality and Productivity Improvement. Britz, G. C., D. W. Emerling, et al. (2000). Improving Performance Through Statistical Thinking. Milwaukee, WI, ASQ Quality Press. Hoerl, R. and R. D. Snee (2002). Statistical Thinking - Improving Business Performance. Pacific Grove, CA, Duxburry Blog: http://www4.asq.org/blogs/statistics Process Improvement Strategy Comparison with DMAIC Strategy • Improvement occurs in iterative sequential steps – Enhanced PDCA approach to improvement • Emphasis on removing special-cause variation first – Analysis of special cause variation differs from common-cause variation Current Scope of SQC Douglas Montgomery (2005). Introduction to Statistical Quality Control, Fifth Edition – “Quality is inversely proportional to variability” – “Quality improvement is the reduction of variability in processes and products” – “quality improvement … three major areas..statistical process control, design of experiments, … acceptance sampling.” Observation • Lacks an overall process improvement strategy Statistical Process Control Montgomery (2005) – “Statistical Process Control (SPC) is a powerful collection of problem-solving tools useful in achieving process stability and improving capability through the reduction of variability” Observations • Lacks an overall process improvement strategy • In practice, focus is on control charts • More emphasis needed on reducing common-cause variation Statistical Process Improvement • Upgrade to SQC and SPC • Use Statistical Thinking • Use Hoerl-Snee Process Improvement Strategy Statistical Thinking: The Transition to Entitlement Quality Robert Mitchell - 3M LSSQ October 10, 2008 Quality Journey • A typical example Six Sigma – Lean – Innovation – Human Sigma -- Entitlement Quality DMAIC • • • • • • Project-by-Project Improvement Eliminate defects (nonconformance) Business Critical Y: Cost, Cash, Growth Project length: 6-9 months Tools focused Metrics: Primary, Secondary, Counterbalance Learnings: – Internally-focused. Where is the customer? – Lack of systems thinking... Sub-optimization – Must focus on building process capability Lean • • • • • Eliminate waste (8 forms of muda) Eliminate non value-added activities Improve flow, Reduce cycle-time Tools focused... Not a philosophy, like TPS Metrics: Yield, Time, Productivity, Inventory Learnings: – Cannot reliably improve flow unless process is stable – Lack of knowledge of variation – Cultural norms and behaviors (LMS) must be created Commercialization NPI Framework Idea Concept Feasibility Dev’t Scale Up Launch PostLaunch • Collect and translate fuzzy VOC • Understand variation in markets and customer segments • DFSS tools to design and deliver value-added products and services Learnings: – Commercialization is a business strategy – Innovation is messy, not linear – ST concepts drive robust product design Process Approach Process Inputs Suppliers Outputs Customers A series of activities that converts inputs into outputs The business should see the improvement ($), the customer should “feel” the improvement ST in All Improvement Initiatives All work is a process Processes are variable Change Process Analyze process variation Develop process knowledge Reduce Variation Improved Quality Control Process Roger Hoerl, Ron Snee, Statistical Thinking Improving Business Performance, pg 13 (ISBN 0-534-38158-8) Satisfied: • Employees • Customers • Shareholders • Community Human Sigma John H. Fleming Gallup Consulting • An holistic approach to optimizing the vital signs of a company’s human systems • Focus on reducing variability in performance and improve organizational effectiveness – The human aspects that drive profitability and growth • In a service economy, value is created when an employee meets and interacts with the customer • Variation = Danger 3M Entitlement Quality • Improvement methodologies are often treated as “floats in a parade” (Jim Buckman, Juran Center, U of MN Carlson School). • But the improvement principles and tools are bedrock... building blocks to continual improvement. • EQ integrates Statistical Thinking into a system of continuous improvement approaches of Quality-Lean-Six Sigma-Innovation methods to optimize customer value. 3M Entitlement Quality Back to basics... – Focus on key business processes, value streams, and customer CTQs – Characterize process behavior (average and variation, structure) • “Plot the dots... and look at the plots” (Lynne Hare) – Assess process state and capability – Apply a critical thought process – Address the root causes using the appropriate tool regardless of the improvement toolkit. 3M Leadership Attributes Building the Culture • • • • • • Thinks from the Outside In Drives Innovation and Growth Develops, Teaches and Engages Others Make Courageous Decisions Leads with Energy, Passion and Urgency Lives 3M Values Statistical Thinking: The Future Roger Hoerl, GE Global Research Statistical Thinking – What Next? • There will always be “the next big thing” in the business world – Total Quality Management – Reengineering – Six Sigma – Lean – Innovation – ??? Statistical Thinking – What Next? • However, some things never go out of style: – – – – – Chocolate High heels Diamond rings Pizza and beer Spending holidays with the family • Business improvement, including the use of statistical thinking, is one of those things – The concepts are timeless, and they work! One Specific Thought • The first principle of the Statistical Thinking definition: – “All work occurs in a system of interconnected processes” • This critical principle has not yet been applied broadly to continuous improvement initiatives – We have tended to focus on one improvement process: Six Sigma, Lean, Reengineering, etc. – We need initiatives that emphasize the system of interconnected improvement processes A System of Interconnected Improvement Processes* Process Performance Data Customers Process Improvements The Process Feedback Feedback Reports & Information to Management Process Control Periodic Analysis and Reviews Feedback Improvement Projects Continuous Improvement System When Needed *From Snee and Hoerl Leading Six Sigma (2003) Product & Process Redesign System of Improvement Processes • Must be managed and optimized as a system – Not sub-optimized at the process level – No competition among “favorite methods” • Covers “Juran Trilogy”: – Design, improvement, control – Long-term, medium-term, short-term improvement • Avoids the “fad of the month” trap • “Back to the future” – an idea whose time has come (again)