#######################################

     KEYZINE: An e-zine for LEADERS:
   ABOUT THE PEOPLE PART OF BUSINESS
             Volume 97, April 2009
  Publisher: © Key Associates, 2009
              ISSN # 1545-8873
           http://www.mkkey.com/

#######################################

                         

This Issue: On "Process Behavior"

Contents:

"In God we trust.  All others bring data.”
           -- W. Edwards Deming, American Statistician 

"If you can't describe what you're doing as a process, 
then you don't know what you're doing.”
           -- W. Edwards Deming

"When the system is stable, telling the worker about 
mistakes is only tampering.”
           -- W. Edwards Deming

"Confusing common causes with special causes 
will only make things worse.”
           -- W. Edwards Deming

"Managers who do not understand a process 
will manage by the numbers alone.”
            -- W. Edwards Deming

**************************************
WHAT'S HOT IN LEADERSHIP
**************************************  
UNDERSTANDING THAT VARIATION EXISTS 
IN A PROCESS AND WHAT THAT MEANS FOR 
MANAGEMENT DECISIONS.

MONITORING PROCESSES OVER TIME..

NOT REACTING TO SINGLE DATA POINTS
IN DECISION-MAKING.

STUDYING THE EFFECT OF CHANGES ON 
A PROCESS, AS APPLIED SCIENCE.

ALWAYS ASKING:  WHAT COULD WE CHANGE?  
HOW COULD WE MORE ACCURATELY PREDICT?

***************************************
MAINTAINING YOURSELF AS A LEADER
***************************************
For Dr. W. Edwards Deming, everything was an N of 1.  
For leaders, this means that understanding comes from 
observation and appreciation of differences.  Every process 
will vary over time.  Every person is unique, and will 
also change over time.  

Management depends on prediction.  Yet as leaders, we are 
prone to make decisions on one data point--one error, 
one quarterly result, one infraction.  True prediction 
cannot occur without watching data over time.  (Perfect 
prediction would be a straight line, which will never exist).  
Process behavior is variation.  Therefore, a leader must 
observe, value, and work with the uniqueness of 
every process and person.

 *************************************
FREQUENTLY ASKED QUESTIONS
************************************* 
"How can you understand variation?"

 Not through a static display, such as a table or 
average and standard deviation: these are 
descriptive, but they do not lend themselves to prediction.  
A dynamic display occurs when data is plotted 
on a graph over time (using run or control charts).  
Think of it as a snapshot vs. a video.

A further appreciation of what you observe comes 
from teasing out special cause (SC) variation from 
common cause (CC).  CC is the natural, random 
variation inherent in that particular process; SC 
is due to unusual or unnatural causes that do not 
belong to the process, sometimes called "noise." 
SC's should be identified and eliminated (or incorporated), 
in order to improve prediction.  A process with only 
CC variation is considered stable.  (See starred* 
References below)

Attempting to improve processes that contain SC's 
will only increase variation and waste resources.  
An example might be introducing a bereavement 
policy for all employees to control a small number 
of people who abuse the privilege. 

 

 

"I am unfamiliar with this statistical thing as a management theory.  
What's the connection?"

When leaders do not understand process behavior and variation, 
they make several errors: 

1.  They see trends where there are none and act wrongly.
2.  They give blame and credit for things over which they have no control.
3.  Because they never fully understand the past performance, 
they make predictions that are untenable, therefore improvements don't work.
4.  The atmosphere of unfair blame/credit and poor understanding of the work process 
creates a culture of fear and decreased morale (Carey & Lloyd, 2001).

 

 

"We make management decisions based on our quarterly 
reports--which give current quarter compared to last year 
same quarter and year-to-date.  Isn't this sufficient?"

Unless you view data over time using Statistical 
Process Control (SPC: Run/Control Chart) methods, 
it's hard to determine if a process is producing acceptable 
or unacceptable results.  Furthermore, quarterly comparisons 
have aggregated 3 months and 90 days worth of business into 
a single number.  Customers do not care about the average 
order time or average cycle time for the quarter.  They care 
about what is happening right now!  SPC methods are designed 
to provide such an understanding. 

To learn more, please see Quality Healthcare: A Guide 
to Developing and Using Indicators by Robert Lloyd (2004).  

 

 

“I learned ANOVA, t-tests and chi-square in school.  Isn’t this 
sufficient to understand variation?”

Most  professionals receive some training in “enumerative statistics,” 
such as descriptive statistics, tests of significance and regression analysis.  
SPC is a distinct branch of statistics ( initially developed by Dr. 
Walter Shewhart in the 1920's).  According to Dr. Bob Lloyd, the key 
distinction is that enumerative examines aggregate data at fixed points in 
time, to determine if one group of data is statistically different from another.  
Whereas analytic statistics seeks to understand the variation that occurs 
with the data over time--through the use of run and control charts.  
The question becomes whether the data reflect common or special causes 
of variation and is prediction possible, not whether two data points are 
different.  This is applied science, not controlled research or experimentation.

 

 

EXERCISES AND ACTION ITEMS:

* Decide on the best measure (signal) for your 
most critical processes.  Plot 20-25 points on a chart, 
by hand, with a pencil, and see what you observe, 
without runs tests.

* Learn:

-  What is a run?
-  Find the tests for special cause (Carey, 2002; Carey & Lloyd, 2001).
-  Analyze your chart.  Are there SC's and where are they coming from?
-  What can you do to stabilize your process?
**************************************
EDUCATIONAL OPPORTUNITIES
**************************************
Articles and newsletters on every type of chart-SPC for EXCEL
http://www.spcforexcel.com/articles-newsletters

Plotting data over time and other CQI instruction
http://www.ihi.org/IHI/Topics/Improvement/ImprovementMethods/Measures/ 

Key Associates offers introductory courses in 
Continuous Improvement and Innovation
and Statistical Thinking .
Contact us at:
1-888-655-3901 or keyassocs@mindspring.com. 

**************************************
OTHER USEFUL WEBSITES 
**************************************
 Web-based SPC Workout for basics on Control Charts
http://www.qualitycoach.net/webbased/webasedspc.htm.

Free trial of ChartRunner by PQ Systems
http://www.pqsystems.com/products/SPC/CHARTrunner/CHARTrunner.php.

Statistical software for Windows, MINITAB
http://www.minitab.com/products/minitab/14/.

QI Analyst--an SPC product with more industrial 
and real-time applications 
http://trainweb.wonderware.com/getstartqia/index.htm

Charts, graphs, and diagrams as add-ons to EXCEL
http://www.qimacros.com/Macros.html (QI Macros).

Former Keyzines on related topics:
Volume 47, February 2005 - Whither Quality
 Volume 48, March 2005 - The Strategic Plan
Volume 49, April 2005 - Measurement
Volume 50, May 2005 - Picture of a Process
Volume 80, November 2007 - Balanced Scorecard
Volume 89, August 2008 - Accountability
Volume 93, December, 2008 - Bad Systems, Good People

**************************************
ARTICLES/PUBLICATIONS                              
************************************** 
* Carey, Raymond G.  Improving Healthcare 
with Control Charts: Basic and Advanced 
SPC Methods and Case Studies.  American 
Society for Quality, 2002.

* Carey, Raymond G. & Robert C. Lloyd. 
Measuring Quality Improvement in Healthcare: 
A Guide to Statistical Process Control Applications.  
 American Hospital Association, 2001.

Chambers, David S. & Donald J. Wheeler.  
Understanding Statistical Process Control.  
SPC Press, 1992.

Deming, W. Edwards.  Out of the Crisis.
MIT Press, 2000.

Deming, W. Edwards.  The New Economics.  
MIT Press, 2000.

* Lloyd, Robert.  Quality Healthcare: A Guide 
to Developing and Using Indicators, 2004.  

Scholtes, Peter. The Team Handbook. Madison, WI: 
Joiner Associates, Inc., 2003.

Shuster, H. David. Teaming for Quality Improvement: 
A Process for Innovation and Consensus. 
Englewood Cliffs, NJ: Prentice Hall, 1990.

Wheeler, Donald J.  Understanding Variation: 
The Key to Managing Chaos.  SPC Press, 2000.

Wheeler, Donald J.   Advanced Topics in Statistical 
Process Control: The Power of Shewhart's Charts.
SPC Press, 2004.