outgoing chair’s message incoming chair’s...

Outgoing Chair’s Messageby Beth Propst

Incoming

Chair’s Messageby Don Emerling

Why is it that sooften we are facedwith good news-badnews situations? Thegood news is that nomatter how bad thenews, there is usuallysome good news toencourage us. Thebad news is that no

matter how good the news, there aresome undesirable things that diminishour joy. In my case, the good news isthat I am writing the Outgoing Chair’sMessage. The bad news is that I amwriting the Outgoing Chair’s Message.The good news is that I am finished(almost) with the administrivia andpaperwork that a Chair must deal with.The bad news is that I was just startingto be comfortable with the duties andresponsibilities of leadership. I just getto know the ropes and then - poof - I’mgone and someone else has to learnthem.

The good news is the Division hasaccomplished a lot this year. TheStatistical Thinking Tactical PlanningTeam of Roger Hoerl, Don Emerling,Lynne Hare, Galen Britz, and JaniceShade have continued their excellentwork with a session at the 1996 FallTechnical Conference and an article inthe June, 1997, issue of QualityProgress. Past publications of this groupare available through the QualityInformation Center at ASQ, and demandto date has far exceeded our expecta-tions.

Our Web Master (I think he’s a wiz-ard!) Mark Kiel has gotten our divisionweb page (www.asqc.org/statdiv) upand running and it’s absolutely fabu-lous. Education Chair Chris Ayersalong with Jacob Van Bowen and Don

Volume 16, No. 6 Summer, 1997

©

Inside This Issue

Basic Tools - Using a Graphical Technique for Determiningand Displaying Process Capability ..........................................p. 6Creation of Section Liaisons ........................................................p. 11Fall Technical Conference ...........................................................p. 12FTC Pre-Conference Short Courses .............................................p. 14Mini Paper - A Note on EWMA ..................................................p. 15SPAIG Planning Meeting..............................................................p. 20

“Sparky” Williams created the firstmodules of the “Statistical ThinkingVirtual Academy” which debuted atQuality Congress and can now beaccessed off our web page - minussome of the interactive portions whichare not currently supported by the soft-ware. Please visit our Academy (we arelooking for a better name - please con-tribute suggestions) and give us yourcomments and feedback. We need ideasfor new modules as well as volunteersto develop material.

This year for the first time we spon-sored sessions and a short course at theRocky Mountain Quality Conference,thanks to the efforts of Region 13Councillor Rick Schleusener. We hopeto continue this relationship, perhapsexpanding the number of sessions, nextyear. AQC Short Course Chair MarceyAbate did service above and beyond thecall of duty to keep our short courseafloat in the face of the “Orlando effect”and facilitated the crafting of an agree-ment between the Division and head-quarters regarding the financial detailsof the AQC short course. Even thoughwe had a small turnout at the short

I recently started a new job, with anew manager. In my first meeting withhim, I mentioned that I would be theChair of the Statistics Division of ASQand that I would have some time com-mitments for the next few years. Afterasking incredulously, “Why would any-one want to be in a Statistics Division ofanything?” (I obviously have some workto do in communicating the vital impor-tance of statistical thinking), he asked abetter question; “How much of yourtime will be spent on this job?” He wastruly concerned about the amount oftime commitment, between work andStatistics Division activity, throwing mylife out of balance. I assured him I hadthought this through, and was lookingforward to the long hours! (Fortunatelythere is a lot of fun built in!)

His question did get me thinkingabout what “this job” is. What does itmean to be the Chair of the StatisticsDivision? What is “this job” and howwould I know if I was doing a good job.

So I began to wonder… …will I be an administrator, what

Continued on page 5Continued on page 3

2 ASQ STATISTICS DIVISION NEWSLETTER, VOL. 16, No. 6

VISION- Our customers’ needs will be con-tinuously anticipated and met.

- Our members will be proud to be apart of the Division.

- Our Division’s operations will be amodel for other organizations.

- We will be a widely inf luentialauthority on scientific approaches toquality and productivity improve-ment.

MISSION- Promote statistical thinking for qual-ity and productivity improvement.

- Serve ASQ, business, industry, acad-emia and government as a resourcefor effective use of statistical meth-ods for quality and productivityimprovement.

- Provide a focal point within ASQ forproblem-driven development andeffective use of new statistical meth-ods.

- Support the growth and develop-ment of Division members.

STRATEGY- Our primary customers are StatisticsDivision members. Other key cus-tomers are:- Management,- Users and potential users of sta-

tistical methods for quality andproductivity improvement,

- Educators of the above cus-tomers.

- Our orientation to customers is cus-tomer focused.

- Our markets, within which weintend to offer products, are weight-ed as follows: greatest weight onintermediate statistical methods,nearly as much weight on basicmethods, and much less weight onadvanced methods.

- Our primary products are educa-tional services.

PRINCIPLES- Focus on a few key things.- Balance short-term and long-term

efforts.- Recognize that we exist for our cus-tomers.

- Value diversity (including geograph-ical and occupational) of our mem-bership.

- Be proactive.- View statistics from the broad viewof quality management.

- Apply statistical thinking ourselves(that is, practice what we preach).

- Uphold professional ethics- Continuously improve

First and foremost, I need toapologize for the tardiness of the1997 Special Publication. Due tounforeseen circumstances regardingother personnel and professionalcommitments, the original author,Tom Swails, will be unable to com-plete the intended publication.Unfortunately, final verification thatthe article would not be completedoccurred in June, several monthspast the expected deadline. We didfind another author to pen the 1997Special Publication.

Davis Balestracci, a StatisticalSpecialist at HealthSystem Minnesota,has agreed to write a SpecialPublication on the topic of “DataSanity” . What is data sanity? It is thepractice of process-oriented thinking,understanding variation, and theintelligent use of data to make deci-sions. Davis will require approxi-mately 6 months to write this piece.Timing for the Publication is slatedfor early December, between the Falland Winter editions. Many thanks toDavis Balestracci for offering his timeand support.

On the topic of the SpecialPublication, the Division is planning

to issue these publications in theWinter instead of the Spring. Wehave learned that much of thedetailed information for the AnnualQuality Congress is not available forthe Winter Newsletter. The movewill permit the Division to betterinform the membership of AQCevents. This change will occur inJanuary 1999. Please note that the1998 Special Publication will bemailed in mid-March, and won’t beaffected by the late mailing of the1997 Publication or the timingchange in 1999.

Lastly to all members: The “articlepool” has gone dry! If anyone has aMini paper or Basic Tools articlewhich meets the criteria listed onthis page, please share it with our12,000 readers. Having done somewriting myself, I know the difficultiesof finding additional time to com-plete and submit an article.However, many of our memberswill benefit from your writings. Andtrust me, there is a big sense ofaccomplishment when you are done.So come on.....Spark up thatcomputer.....Get creative.....We willthank you! ∆

Editor’s Corner

Criteria for Basic Tools and

Mini-Paper ColumnsBasic Tools

Purpose: To inform/teach the “quality prac-titioner” about useful techniques that canbe easily understood, applied andexplained to others.Criteria:1. Application oriented/not theory2. Non-technical in nature3. Techniques that can be understood and

applied by non-statisticians.4. Approximately three to five pages or less

in length (8 1/2” x 11” typewritten, singlespaced.)

5. Should be presented in “how to use it”fashion.

6. Should include applicable examples.Possible Topics:New SPC techniquesGraphical techniquesStatistical thinking principles“Rehash” established methods

Mini-PaperPurpose: To provide insight into applica-tion-oriented techniques of significant valueto quality professionals.Criteria:1. Application oriented.

2. More technical than Basic Tools, but con-tains no mathematical derivations.

3. Focus is on insight into why a techniqueis of value.

4. Approximately six to eight pages or lessin length (8 1/2” x 11” typewritten, singlespaced.)Longer articles may be submitted andpublished in two parts.

5. Not overly controversial.6. Should include applicable examples.

General InformationAuthors should have a conceptual

understanding of the topic and should bewilling to answer questions relating to thearticle through the newsletter. Authors donot have to be members of the StatisticsDivision.

Submissions may be made at any timeto the Statistics Division Newsletter Editor.All articles will be reviewed. The editorreserves discretionary right in determinationof which articles are published.

Acceptance of articles does not implyany agreement that a given article will bepublished. ∆

ASQ STATISTICS DIVISION NEWSLETTER, VOL. 16, No. 6 3

course, the response of the participantswas very positive.

Another new venture this year wasthe Ott Scholarship, which is off andrunning under the capable leadership ofAwards Committee Chair Lynne Hare.The first scholarships will be awardedfor Fall, 1997.

The bad news is that there weresome things we weren’t able to accom-plish. The Publications Committee underChair Don Strickert suffered from lackof direction from Division leadership,our failure to involve Quality Press earlyin the process, and our inability to fillseveral key positions. At our tacticalplanning meeting in May we wereforced to seriously rethink our approachto publications. We have backed off ofour original plan and will be taking adifferent approach in the future. Ourspring Special Publication (which, yourecall, we initiated last year in place ofthe spring edition of the newsletter) fellthrough in spite of the work ofNewsletter Editor Janice Shade. (Thegood news is that we are still workingon it and plan to publish it as soon aswe can, so next year you will receivetwo Special Publications instead of one.)

More bad news — an ongoing rela-tionship with a local conference inRegion 12 has been ended. For manyyears the Division has sponsored a trackof speakers at Argonne Quality Day puton by Section 1208 (St. Charles) in con-junction with Argonne NationalLaboratories. Well, Argonne has reengi-neered its Quality Department out ofexistence and so this long-standing con-ference is no more.

But ... along comes the good news:a whole host of people have done agreat job carrying out the day-to-dayoperations of the Division. First andforemost of these are my fellow officers- Chair Elect Don Emerling, TreasurerDon Williams, and Secretary BobMitchell. I couldn’t ask for a better,more delightful group of people to workwith - even if I did, I probably wouldn’tget it. Without them I couldn’t havedone my job. Nor could I have done myjob without coaching, advice, reminders,and support from Past Chair NancyBelunis, who last year guided theDivision once again to the highest levelof the McDermond DivisionManagement Recognition Program(threepeat!). We anticipate reachingLevel 3 again for the 1996-1997 year. Ofcourse, all of our achievements are

based on the excellent work of all thePast Chairs who have laid the founda-tions upon which we continue to build,and from whom I have learned somuch.

There are many others who havemade the Division operations a successthis year. In the interest of brevity, I willonly mention those not previously men-tioned. This year’s Hunter Award wasawarded to Doug Montgomery byHunter Awards Chair Steve Bailey.Steve ended his term of office onDecember 31, and current Chair GalenBritz is busily guiding his committee inthe selection of the 1997 recipient of thisaward. Helping organize our confer-ences and short courses (in addition toMarcey and Rick mentioned above), BillBleau shone as FTC Short Course Chair,Jacob Van Bowen as FTC ProgramCommittee Representative, LarryHaugh as AQC Division SessionManager, and Mel Alexander, SubirChowdbury, John Clulow, FredSpiring, Tsong-how Chang, andElaine Allen as AQC Technical PaperReviewers, and Frank Alt as DemingConference on Applied StatisticsProgram Chair. Helping Don Strickertwith Publications were How-To SeriesEditors Walter Liggett and Bob Brill.Nick Martino continued to serve withdistinction as Certification Chair with thehelp of countless volunteers. In fact, hehas been told to slow down on thenumber of volunteers to work on theCQE exam. They cannot use all of thenames we have supplied. But ... Nickwill be investigating what Division inputwill be required for other certificationexams (e.g., the CQT). Standards ChairEd Schilling and his committee havedone such outstanding work that theDivision has now acquired responsibilityfor several new standards. Bob Perrycontinued as Examining Chair - a posi-tion he has held for many years - andASA Q&P Liaison Joe Voelkel has keptus in touch with what our ASA col-leagues are doing. Membership Chair JLMadrigal has moved into his new posi-tion with grace and aplomb and helpedin the redefinition of the RegionalCouncillor position (which you’ll all behearing more about) and the institution-alization of ongoing member surveys.Many of the Regional Councillors par-ticipated in this effort.

The bad news is that two people areretiring from their positions because ofjob responsibilities and other conflicts.They are Education Chair Chris Ayersand How-To Editor Bob Brill. BothChris and Bob contributed a lot to the

Division. We are sorry to see them goand wish them well in their futureendeavors. Other people are leavingtheir current positions to move into newpositions in the Division, and that isgood - for them and for us.

Some really good news - I am turningover the reins to Incoming Chair DonEmerling, who is so good he is alreadymaking significant improvements to theoperation of the Division. (The badnews is that I didn’t think of them a yearago so the Division could have benefit-ted sooner!) He has a capable group ofofficers to help him: Don Williams,Bob Mitchell, and Janice Shade. Thebad news is that as Past Chair myinvolvement with this exceptional groupwill be tapering off. The good news isthat I will now be free to pursue otheractivities within ASQ. Next year I will beserving on the Education and TrainingBoard and as a GTC Facilitator. The badnews is that the Division won’t be rid ofme that easily — I’ll be ba-a-a-ack toserve in other capacities.

One last piece of bad news - goodnews. The good news is that I’ve hadmany opportunities to share thoughtswith all of you — the members, the rea-son behind all of the work we do. Thishas occurred through my Chair’sMessages, phone conversations, e-mailconversations, and personal contact atconferences. I have really enjoyed thatpart of my job — it has been a wonder-ful experience. If that were the extent ofthe job description, I could be Chair fora lifetime and never tire of it! The badnews is that this is my last officialopportunity to share with you and I’vealready filled my air time by recognizingthe efforts of the volunteers that haveenabled the Division to be successful.The good news is that conversationsand relationships, once begun, can con-tinue indefinitely and I look forward tocontinuing those I have begun this pastyear. I’d like to leave you with someprofound words of wisdom, but I’m notfeeling particularly wise or profound atthe moment. Instead, a quote fromTennyson’s Ulysses springs to mind:I am a part of all that I have met;Yet all experience is an arch wherethro’Gleams that untravell’d world whose

margin fadesFor ever and for ever when I move.

Even so, I am a part of everything Ihave done and everyone I have met, asyou are all a part of me, but my jobhere is done and so I ride off slowlyinto the sunset in search of thoseuntraveled worlds that are now beforeme. (That’s the good news!) ∆

OUTGOING CHAIR’S MESSAGE

Continued from page 1


1996 Annual Evaluation, ASQ Statistics DivisionDon Emerling, Chair-Elect

1. IntroductionThis is the sixth annual evaluation of the Statistics

Division.The annual evaluation is intended to “facilitate continu-

ous improvement of the Statistics Division by measuringcustomer satisfaction and progress of the Division relativeto McDermond guidelines and Statistics Division Vision,Mission, Strategy, Principles, Systems and Annual TacticalPlans” (Statistics Division Operating Manual).

Evaluation: The annual evaluation is not meeting it’sintended purpose. We are not using it to facilitate continu-ous improvement of the Statistics Division.

Upgrade: Develop a tactical plan by this year’s FTC tooperationalize the continuous improvement process, basedon the annual evaluation.

2. Vision, Mission, Strategy and PrinciplesOur vision is the desired end state the division is striving

to achieve, and our mission is our reason for existence.The strategy defines our customers and markets. Our prin-ciples are rules to follow in the pursuit of our vision andmission. No changes have been made to these since thelast evaluation.

Evaluation: The five-year plan is directed towardsachieving our vision and fulfilling our mission. The plan isnow three years old and needs to be revisited and revisedto a new five-year plan.

Upgrade: 1. Review and evaluate the progress towardthe five-year plan at this years’ AQC tactical planning meet-ing.

2. Schedule a long-range planning meeting in the firstquarter of 1998 to review our progress toward our visionand develop a new five-year plan.

3. Customer SatisfactionDivision Membership continues to remain flat or declin-

ing slightly. Our quarterly membership survey’s indicate ahigh level of satisfaction and the exit interviews almostalways give reasons other than dissatisfaction with theStatistics Division.

Evaluation: The membership survey will be important inassessing members’ needs. The booth activity at AQC hasbeen used to provide information on how the divisioncould be more responsive.

Upgrade: Results from the survey and previous member-ship surveys should be included in the Operating Manual.Additional ways to measure customer satisfaction need tobe identified. Information obtained from the booth activi-ties needs to be reviewed.

4. Systems

4.1 Infrastructure RenewalPurpose: Continuously renew and improve the organiza-

tional structure of the Division.Procedure: Past Chair is responsible.Evaluation: We should attempt to include as many past

chairs as possible in the infrastructure meeting at the AQC.This will give us a better pool of possible future leaders ofthe division. We need to develop the process for respond-ing to, and taking advantage of, members who complete

member interest forms. This process should include metricsto track the quality of the process.

Upgrade: Develop a tactical plan by this year’s FTC todevelop the process for responding to members who com-plete member interest forms.

4.2 Conference CallsPurpose: Facilitate communicate among division leader-

ship outside of scheduled meetingsProcedure: Conference calls among division leadership

are held on a monthly basis.Evaluation: The conference calls have been a valuable

process for keeping the division leadership communicatingon a regular basis. We should include a regular review ofthe Statistics Division Table of Activities (pp. 4-7 of theStatistics Division Chair Job Description) as a means of“keeping on top” the activities throughout the year. Weneed to find a way to more actively manage the activitiesof the division, such as tactical plans, and committeeactivity.

Upgrade: 1. Incorporate a regular review of the Table ofActivities into the conference call agenda.

2. Have a 1-2 day operational planning meeting (nolater than July 15, 1997) of the division leadership to builda calendar-based operational plan for the fiscal year.

4.3 McDermond Division Management RecognitionProgram

Purpose: Encourage divisions to focus their efforts onmeeting the needs of their members, customers and ASQ.

Procedure: Documented in ASQ’s policies and proce-dures. Chair is responsible for submitting the application.

Evaluation: Statistics Division achieved the highest level- Level 3 for 1995-1996. We are applying for Level 3 in1996-1997. We have designed our division infrastructure toachieve Level 3 every year.

Upgrade: Follow the process.

4.4 Operating ManualPurpose: Provide the information needed to run the divi-

sion.Procedure: Updated by Chair-Elect in first quarter of the

calendar year.Evaluation: The system is working well.Upgrade: 1. Put the operating manual on the Statistics

division home page.2. Provide an opportunity for members to receive an

electronic copy of the operating manual.

4.5 Tactical PlanningPurpose: To develop and implement tactical plans sup-

porting the division strategy.Procedure: Tactical Planning meetings are held at the

AQC and FTC. Follow-up meeting may be held as neces-sary.

Evaluation: A. The tactical planning meetings need tobe balanced between providing time to work on tacticalplans and providing time to review status and discuss newtactical plan candidates.

B. The tactical plans that have been generated from thefive-year plan have tended to be multi-dimensional and

Continued on page 5


Beth calls “dealing with the administriv-ia…”? I really hope not! — If that’s themost important job of the Chair, I’ll becounting the days until next July 1!

…will I be leading an organization of12,000 people? I don’t really think so.— To be honest, I probably only knowabout 100 members, it’s hard to imaginea leader who doesn’t know 99% of thepeople he is supposed to be leading!That’s why I’ve always had trouble withthe concept of the President of theUnited States being the “leader of thefree world.” I personally have never feltthat I’ve been led anywhere by a politi-cian, unless it’s down the primrose path!

…will I be the “keeper of theMission?” That sure sounds like the jobof more than one person. — The mis-sion statement of the Statistics Divisionis to:

• Promote statistical thinking forquality and productivity improve-ment.

• Serve ASQ, business, industry,academia and government as aresource for effective use of statisti-cal methods for quality and produc-tivity improvement.

• Provide a focal point within ASQfor problem-driven developmentand effective use of new statisticalmethods.

• Support the growth and develop-ment of Division members.

That list tells me that a whole lot ofpeople need to be working together ifthere is any hope that the mission willimpact society.

OK, so now I’m really confused,WHAT IS “THIS JOB”??!!

To help me answer this question, thatby now was burning in my mind, Ilooked up the purpose statement in theChair’s job description. It states, “Theprimary purpose of the chair is to inte-grate the diverse activities of theDivision. The Chair must ensure theseare supportive of the Vision, Mission,Principles, and Strategy. A secondarypurpose of the Chair is to lead the annu-al and long term planning of theDivision.

So I am an integrator (“…to integratethe diverse activities of the Division.”),and an evaluator “…must ensure theseare supportive…), and a planner (…leadthe annual and long term planning…).

I’d like to discuss these roles by firstaddressing what is happening to supportintegration and planning, and then byaddressing the subject of evaluation.

lasting more than a year. They often have sub-plans imbedded in them andbecome complex and difficult to make progress on.

Upgrade: A.1. Split the AQC tactical planning meeting into two sessions. First,a review and evaluation session, where we review progress and evaluate direc-tion. Second, a working session, where tactical plan teams can work on theirspecific plans.

A.2. Dedicate the FTC tactical planning meeting to working sessions, with avery brief wrap-up review at the end of the day.

B.1. We should have a discussion about tactical plans that are generatedfrom the five-year plan (which I consider more strategic and complex) vs. tacti-cal plans that are generated from other sources, such as this annual review (which I consider more operational).

The team working on the tactical plan “Enabling Broad Application ofStatistical Thinking” made good progress this year. They presented at the FTCand are developing a “How-To” book. The FTC presentation is scheduled forpublication in Quality Progress.

The team working on “Integrating Statistical Thinking and Tools intoEducational Curricula” is planning to present the “Virtual Academy” at theStatistics Division booth at this years’ AQC.

4.6 Annual Division EvaluationPurpose: See IntroductionProcedure: See IntroductionEvaluation: This is the sixth annual evaluation. There is still no clear process.Upgrade: See upgrade in Introduction.

4.7 Financial RequestsPurpose: Handle requests for donations in a manner consistent with our mis-

sion.Procedure: Described in the Operating Manual.Evaluation: Working well. Donations done on an annual basis our now

included in the budget.Upgrade: Follow the process.

4.8 Maintenance SystemPurpose: Ensure that action items resulting from meetings are completed.Procedure: Keep a separate flip chart for action items at meetings. Review at

end of meeting and assign responsibilities and timing. Include all action itemsin a single list and include at the front of the minutes. Secretary will send atreminders at set intervals.

Evaluation: Working well. Modifications have been made to the system. Allaction items even those from conference calls are captured on one list. Thus theaction item list changes on a regular basis as new items are added and othersare completed.

Upgrade: Follow the process.

4.9 BudgetsPurpose: Ensure that expenses which the division incurs have been included

in the annual division budget.Procedure: Committee chairs and tactical plan leaders provide a budget to

the treasurer by April 15 of each year.Evaluation: The process works well. We have been running a deficit budget

for several years. We have raised the dues to $8.00 this year to cover increasedoperating expenses.

Upgrade: Make sure we continue to add value to our members.

4.10 Reimbursement PolicyPurpose: Ensure that expenses are reimbursed in a standard manner.Procedure: Section 4.10 of the Statistics Division Operating Manual describes

the procedure for reimbursement. Evaluation: The process works well.Upgrade: Follow the process. ∆

INCOMING CHAIR’S MESSAGE


ANNUAL EVALUATIONContinued from page 4

Continued on page 14


IntroductionCapability indices seem to be the latest fad in judging product quality. As a Statistical Process Control (SPC) coordinator Iam often asked questions such as, what is the Cpk of your process without any regard to process stability or distributionshape. One quick approach to assess the capability of your process is to calculate multiple capability indices (such asCpk) over time and to plot these numbers on an XmR control chart, along with a histogram of the individual values.Although not an exact method, by calculating and charting capability ratios over an extended period of time instead ofproviding a one time value, a clearer picture of your processes’ capability can be established. The following exampleuses 3 statistical tools in order to evaluate and display process capability of a die bond operation.

Basic Tools used to Perform the Capability Study• X and R control chart• XmR control chart• Histogram

Formulas used to Perform Study• Estimated standard deviation = R/d2• Cpk = X - LSL/3 (R/d2)

Description of the Die Bond ProcessAn automatic die bonder picks silicon die from a wafer and places each die onto a thin layer of conductive epoxy whichhas been deposited on a ceramic substrate for bonding. After the die and epoxy have been cured for 1 hour, a sample is

tested for bond strength. A die strength testerrecords the force required (in grams) to shearoff a die. A 3 piece sample is sheared every 2hours and the results are recorded on a X andR control chart (Figure 1). The resulting controlchart displays statistical control and thereforequalifies for a process capability study. Thecapability study compares the natural processlimits to the product test specification. For thedie bond operation, the lower spec limit (LSL)is 400 grams. There is no upper spec limit(USL).

A Graphic Display of Process Capability forthe Die Bond OperationStep 1: Verify that the process displays a rea-sonable degree of statistical process control.Capability studies should only be performed onstable and predictable processes. Although

there is no agreed upon definition among SPCpractitioners as to what constitutes a stable process (statistical process control) because of the various detection rules andthe fact that no process is always stable, for this example statistical process control will be defined as 35 consecutive sub-groups of size 3 without any points outside the control limits (also known as Western Electric Zone Test 1). The last 100subgroups from the die bond process show no values beyond the control limits on either the X or R control chart.

Step 2: For the die bond process each weeks’ data from the X and R control chart is collected and the following statisticsare calculated:

• The estimated process standard deviation (R/d2) • The process mean (X)

Continued on page 7

BASIC TOOLUSING A GRAPHICAL TECHNIQUE FOR DETERMINING

AND DISPLAY PROCESS CAPABILITY

Richard DeRoeck

Figure 1


Step 3: Calculate a Cpk value for each weeks’ data using the following formula:

Cpk = X - LSL/ 3(R/d2)

Cpk values for weeks 1 through 8

Week 1: 1.253 Week 4: 1.201 Week 7: 1.579Week 2: 1.900 Week 5: 1.364 Week 8: 1.454Week 3: 1.294 Week 6: 1.242

Step 4: Plot each weeks Cpk numbers on an Individuals Control chart (also known as a XmR chart). The X axis repre-sents time (weeks) while the Y axis representsCpk values (Figure 2).

Step 5: Calculate control limits and averages linesfor both the X chart and the mR chart. Wait untilyou have collected about 8-10 weeks worth ofdata before setting limits. (Figure 2).

Step 6: Construct a histogram of the individualvalues (Figure 3). Here the last 100 subgroups(300 individual readings) covering approximately2 months worth of data were used to generate thishistogram. Note: When the raw data is skewed,as in this example, there will be a slight skew inthe sample means of the control chart, resulting incontrol limits which are rough approximations.

Interpretation of Cpk Control Chart andHistogramThe XmR control chart of the Cpk values dis-plays a predictable capability. The mean of theX chart is 1.41. This is the best estimate ofprocess capability for the die bond operation.The histogram of the individual values (300readings) show all the values well within thespecification limit (400 grams). It also shows adistribution shape that is skewed to the highside. This suggests that the average Cpk valueof 1.41 is most likely a conservative estimate ofprocess capability.

SummaryAll statistics vary over time, capability indices

such as Cpk are no exception. In order to helpestablish a more realistic picture of your process-

es’ capability, one should always use multiple values. A large sample size is required to ensure a valid estimation. Byusing graphs which show the performance of your process over time (such as an XmR control chart) one can easily gen-erate meaningful capability indices. In addition, knowing the shape of your distribution of data by constructing a his-togram will help you to better interpret these capability values. Finally, one should always keep in mind that only stableand predictable processes can be used when one is performing a capability analysis. ∆

BIBLIOGRAPHYWheeler, Donald J. “Advanced Topics in Statistical Process Control” ( Knoxville, Tenn: SPC Press, 1995)

BASIC TOOLSContinued from page 6

Figure 2

Figure 3


Don is currentlythe ContinuousImprovementManager for ImationCorp, a spin-off of3M’s informationand imaging busi-nesses. Prior to join-ing Imation, he had20 years experience

in 3M Company. Don is located inOakdale, Minnesota.

Don has been active in the StatisticsDivision of ASQ since 1989. He ledthe tactical plan team which devel-oped the Statistics Division “House ofEducation” special edition of thenewsletter and one of the authors ofthe special publication on StatisticalThinking. He has served as theSecretary of the division for the lasttwo years.

Don has a BS in Chemistry from theState University of New York atBrockport and a MS in AppliedStatistics from the Rochester Instituteof Technology (RIT). ∆

Robert (Bob)Mitchell is a QualitySpecialist with 3MCompany’s PersonalCare & RelatedProducts Division.Bob has held a vari-ety of positions withincreasing responsi-bility during his 16

year career with 3M, including QualityAssurance Engineer, ProductDevelopment, Process Development,SPC Coordinator, Statistical Consultant,Supplier Relations Manager, andTechnical Supervisor. Bob is active inhis community as TQM consultant.

Bob, a Senior member of ASQ, is aCertified Quality Engineer and hasbeen active in the Statistics Divisionfor 5 years serving as MembershipChair and Secretary. Bob has led sev-eral tactical plans including develop-ment of a member needs survey andrevision of the Section Liaison &Regional Councilor job descriptions.

Bob has a B.S. in chemistry fromthe University of Minnesota-Morris,and is an inaugural member of 3M’s“Statistical Practitioners’ Forum”. ∆

Janice Shade hasworked for Nabiscofor the past 14 years.She is currently theManager ofContinuousImprovement for theUS Foods Group inParsippany, NewJersey. Janice has

experience in several areas of Quality;including Vendor Quality,Specifications Development,Packaging Design, Quality SystemsAnalysis and Statistical ProcessControl.

Janice has a BS degree in biologyfrom the College of New Jersey, anMBA in Quantitative Analysis fromFairleigh Dickinson University inNew Jersey, and is currently enrolledat Rutgers University in New Jerseyin the Applied Statistics Program. Sheis a Senior Member of the AmericanSociety for Quality, a member ofStrathmore’s Who’s Who for BusinessLeaders. Other Division activitiesinclude Newsletter Editor and is oneof the authors of the SpecialPublication on Statistical Thinking. ∆

Meet Your Officers for 1996-1997

Don EmerlingChair

Bob Mitchell,Secretary

Janice Shade,Treasurer

Don Williams isfounder, president,and senior consul-tant of ProcessImprovementConsultants, amanagement con-sulting firmspecializing inassisting clients in

continuous improvement of theirprocesses, products, and services. Heis a former senior consultant withProcess Management International,and is the founder and formerdirector of the Center for Qualityand Productivity at the University ofNorth Texas.

He has a doctorate in mathematicsand statistics from Oklahoma StateUniversity. Don is a Senior Memberand previously served asCertificationChair of the Statistics Division. ∆

Don Williams,Chair Elect

Second Annual Deming LectureDeming Lecture Title Announced

Dr. Noriaki Kano will present“Attractive Quality Creation” for thesecond annual Deming Lecture at the1997 JSM in Anaheim, CA on Aug. 12.Dr. Kano is professor in theDepartment of Management Science atthe Science University of Tokyo.

Kano first introduced the idea ofAttractive Quality in 1979 and hassince frequently spoken and writtenabout the topic.

The story begins in the 1970s whenthe Konica camera company realizedthat to remain competitive its newcamera must be something completelydifferent from what was available atthe time. However, when the R&D andSales departments began talking tocustomers, it sounded like they wereasking for only minormodifications in the existing model.When they went to a photo processinglab, however, they saw that there weremany failures—such as blurry images,

under and over exposures, and blankrolls. Working to solve these problemsled to many features available in cam-eras today (such as auto focus, built-in-flash, automatic film winding).

As Kano points out, the key to suc-cess was to not just listen to what cus-tomers were saying but to develop adeep understanding of the customers’world and then to address these latentneeds. He will build on this exampleand describe the fundamental notionof “attractive quality” as well as discusshow to go about achieving it.

Kano holds a Doctoral degree inengineering from the University ofTokyo. For many years, he studieddirectly under the supervision ofKaoru Ishikawa. Dr. Kano has beenone of the most active figures in thequality movement for nearly threedecades. He is involved with numer-ous quality organizations, consultswith companies world-wide, and



INTRODUCTIONBeth Propst Chair St. Charles, IL (315) 123-4567Don Emerling Chair Elect Oakdale, MN (612) 704-4606Don Williams Treasurer Denton, TX (940) 243-1147Bob Mitchell Secretary Forest Lake, MN (612) 736-8684Galen Britz Past Chair Maplewood, MN (612) 736-6499Rick Lewis Past Chair St. Louis, MO (314) 694-7735Janice Shade Newsletter

Editor Andover, NJ (201) 682-6236Ed Schilling Standards

Chair Rochester, NY (716) 889-5041Marcey Abate AQC Short Course

Chair Albuquerque, NM (505) 844-9424John Ferrante Member San Jose, CA (510) 353-4846Glen Fondaw Member Wichita, KS (716) 475-6129Joe Pignatiello Member Texas A&M Univ. (409) 862-2081

I. GENERALReview Agenda Beth PropstReview Mission, Vision, Principles, Strategy,

Ground Rules: Beth PropstTreasurer’s Report Don Williams

Total Assets: $320,312Total Liabilities: $ 1,036Fund Balance: $319,274 *

Includes $220,000+ for the Ott Scholarship account.Publication and distribution costs of the Newsletter is ourmajor operating expense, and is the primary communicationvehicle to our members. Expanded use of the electronicwebsite for member communications is being explored.

Membership Report: Bob Mitchell11,011 members as of 3/31/97Membership Satisfaction rating = 89% approvalMail-in membership surveys are conducted quarterlyTelephone interviews of our members will be conducted

semi-annually by the Division CouncilFeedback from surveys, interviews, and telephone logs

will be fed into the Division planning processes.

Announcement of 1997-1998 Officers: Beth PropstChair: Don EmerlingChair-elect: Don WilliamsSecretary: Bob MitchellTreasurer: Janice Shade

Committees: Standards Committee - The Statistics Division“inherited” 3 additional Standards to shepherd: ‘Q3’- Inspection ofIsolated Lots; ‘S2’- Introduction to Attributes Sampling, and ANSIZ1.4 & Z1.9- the ASQC equivalent to the discontinued MIL STDS105E & 414.

Certification Committee - Continued excellentsupport by members to improve, revise the CQE Body ofKnowledge and design of new exam questions. The CQT andCMI exams are also being reviewed.

Conference Update: Marcey Abate, AQC Short Course Chair,reported that the participants in the “Statistical Thinking forHealth Care” Short Course responded very favorably to the mater-ial and that Davis Balestracci is a dynamic speaker. Marcey Abate has volunteered to serve as the 1998 AQC ShortCourse Chair!

Other conference chairs:Susan Albin- 1997 FTC ProgramRick Schleusener- 1997 Rocky Mtn Quality Conf

The 1997 FTC is in Baltimore, MD on Oct 16-17; the FTC CouncilMeeting is scheduled for Oct 15.A Long Range Planning meeting will be scheduled in place of thecustomary FTC Tactical Planning Session.

Booth Activity: Demonstration of our www homepage and theinitial pages of the Virtual Academy. Our intent is to seek inputand suggestions from the participants for future improvements.

Current Tactical Plans - “Becoming More Effective at the LocalLevel”

1. The role of the Regional Councilor within the StatisticsDivision is being modified and responsibilities increased tomake this position a “feeder” group to potential Divisionleadership.

2. A new position, the Section Liaison, is being created to assistthe Regional Councilor. The Section Liaison’s role will be tocommunicate Division products and services to the Sectionmembership and to identify opportunities for the Division atthe Section level.

“Integrate ST2 into Education Curricula”1. Chris Ayers is leaving as Education Chair due to limited avail-

ability2. The Virtual Academy will be demonstrated in the booth.

Volunteers are being sought to work on future modules.

“Improve the How To… Series”1. The Publications Committee needs to be redesigned. Need to

re-think the How To… series.

“Enable Broad Application of Statistical Thinking”1. A “Special Publication” was published in 1996.2. Last year’s Stats Division-sponsored AQC session was pre-

sented by Tom Swails; a follow-up “Special Publication” isdue this summer.

3. This year’s AQC Short Course, “Statistical Thinking in theHealth Care Industry” was presented by Davis Balestracci.

4. A Statistical Thinking “How To…” is being planned.

Tactical/Strategic Planning: Don EmerlingA 3-day Long Range Planning Session has been planned in placeof the usual Tactical Planning associated with the Fall TechnicalConference: October 18, 19, 20 in Baltimore

New Business: None

Benefits and Concerns: (as expressed by Division memberattendees)Benefits:

1. All the officers were present2. Good informal communications vehicle3. The Statistics Division has its act together

Concerns:1. This year’s Annual Open Business meeting conflicted with

the extended AQC exhibit hours and an ASQC meeting.

Beth Propst adjourned the Annual Business Meeting to theHospitality Suite.

Respectfully Submitted,Robert Mitchell

Statistics Division Secretary

STATISTICS DIVISION

AQC ANNUAL BUSINESS MEETINGOmni Rosen Hotel, Orlando, FL

MAY 5, 1997


A very successful two hour session,“The Power of Statistical Thinking”,was held for the Statistics Division inOrlando during the Annual QualityCongress on May 6, 1997. Larry Haugh(University of Vermont) was theSession Manager and Mary Leitnaker(University of Tennessee) organizedthe speakers and moderated the pre-sentation.

Marjorie Green (Allied SignalAerospace) and Tony Cooper (SixSigma Associates) gave a joint presen-tation which revolved around theirpersonal experiences in the dramaticimprovement in productivity thatoccurred at a manufacturing site.Designed experiments played a criticalrole in the overall improvementprocess, but a more general perspec-tive was emphasized. Knowledge ofthe technical or statistical aspects ofthe experimentation alone would nothave been sufficient in implementingthe experimentation process in themanufacturing environment.

Then Wes Anderson (Eli Lilly) dis-cussed the critical role of the measure-ment process in improving “valueadding” manufacturing processes. Thespecific tools and methods used at hispharmaceutical company to evaluateand improve measurement processeswere presented, including the comple-mentary roles of the CPK and the dis-crimination ratio statistics.

This session was very successful insupporting the mission of ourDivision: “Promote statistical thinkingfor quality and productivity improve-ment”. The large audience was able toask a number of questions and tomeet with the speakers after the pre-sentations. I hope you were there; butif not. see you in Philadelphia in 1998!

∆

The AQC Statistics Division spon-sored short course, StatisticalThinking as a Conduit to QualityTransformation in Health Care,Service, and Management, receivedrave reviews from this year’sattendees. The course was taught byDavis Balestracci, an accomplishedinstructor, author, and StatisticalSpecialist at HealthSystem Minnesota.Davis is a dynamic speaker whosehighly interactive course emphasizedthe concept of variation and the useof Statistical Thinking, not merely sta-tistical techniques. Although the shortcourse attendees were diverse in theirpast experience and professions, rang-ing from registered nurses to US AirForce pilots, all left with an expandedawareness of the importance ofbecoming process-oriented in regardto problem identification, becomingchart-oriented in regard to data collec-tion and display, and reacting appro-priately in response to variation.

Topics covered in the short courseare also provided in a book co-authored by Davis Balestracci, QualityImprovement: Practical Applicationsin Group Medical Practice, 2ndEdition, 1996 (published by the Centerfor Research and Ambulatory HealthCare, Englewood, CO, call 303-397-7888 to order). The authors success-fully demonstrate that despite percep-tions of healthcare personnel, health-care is no “different” from manufactur-ing, service, or administration. Thefirst four chapters of the book developthe implications of Statistical Thinkingthrough process-oriented thinking,understanding variation, and the intel-ligent use of data. Chapter 5 intro-duces some of the human psychologi-cal factors lurking in any organization-al change process and ways to dealwith them. Chapter 6 is an excellentsummary of an approach to under-standing variation in everyday workthrough the use of “tools”. Theemphasis is on the thinking needed to

understand a situation, the tools arealmost presented as afterthoughts.Chapters 7 and 8 present extremelyuseful techniques, run charts, controlcharts, and a valuable innovativeadaptation of these known as “analy-sis of means”. The ends of thesechapters contain extensive and variedexercises (with detailed answers)based on real data. Chapter 9 brings itall together to show how data skillsare indeed a conduit to transforma-tion. Chapter 10 discusses learning asa process along with common trainingand education traps. Chapters 11 and12 illustrate “typical” quality improve-ment efforts with case studies. Thebook concludes with an exhaustivebibliography and detailed curriculumsfor a transformation-based approachto quality education and another fora quality improvement seminar aimedat front-line practitioners. This bookcan benefit anyone involved inquality improvement regardless ofpast experience or industry ofemployment.

The Statistics Division is committedto further enabling the broad applica-tion of Statistical Thinking by inform-ing division members, and others, asto the benefits of applying it to theirwork. In pursuit of this commitment,the 1995 Statistics Division AQC pre-sentation and the 1996 SpecialPublication which defines and discuss-es the philosophy, broad application,and implementation of StatisticalThinking in an organization, is stillavailable from the Quality InformationCenter (1-800-248-1946) at ASQCHeadquarters for a nominal cost tocover materials and handling.Additional division sponsored instruc-tion on Statistical Thinking is plannedfor the future including a pre-confer-ence short course being offered inconjunction with the 41st Annual FallTechnical Conference (October 16-17,1997). Watch the newsletter forfurther details and announcements. ∆

1997 AQC Short Course

Summary on Statistical Thinking

AQC Statistics

Division Session

Summary


Becoming More Effective at the Local Level:

The Creation of Section LiaisonsThe Statistics Division leadership has

struggled with finding ways to be moreeffective at the local level. Our divisionhas approximately 12,000 members,most of which cannot attend annualconferences and meetings across thec o u n t r y .

Our current infrastructure has triedto deal with local representation byassigning Regional Councilors in the 15regions of North America as defined byASQ. The regional councilor positionhas been the focus of dissatisfactionwithin the Statistics Division and indeedthroughout ASQ. At the last severalGeneral Technical Council (GTC) meet-ings there have been breakout groupsdiscussing ways to make the RegionalCouncilor position more effective.Regional Councilors have been chal-lenged with big geographical areas tocover, with many Sections in eachregion and, in many cases, poor expec-tations and follow-up by the divisionleadership. In the Statistics Division, wecreated a tactical plan titled “Becomingmore effective at the local level” toaddress the issues facing the RegionalCouncilors. Bob Mitchell and JLMadrigal were the primary contributorsto this tactical plan and deserve thecredit for the creation of a system thatshould make the Statistics Divisionmuch more effective at the local level.

We have decided to create a newposition in the Statistics Division calledSection Liaison. There are over 240local sections of ASQ, all of which haveregular section meetings and local con-ferences and educational offerings. Weare hoping, by identifying a liaisonwithin these sections, we will have abetter way to spread the message ofstatistical thinking and the tools andtechniques of statistics. As stated in thejob description the purpose of theSection Liaison is “To communicate toSection membership and leadership theproducts and services that the StatisticsDivision has to offer; and to identifyopportunities to deliver those productsand services within the Section.” Theirprimary responsibilities will includeactive participation in Section activities,pursuing joint activities between theStatistics Division and the Section, pre-senting Statistics Division products andservices to the Section, participating incommunication with other SectionLiaisons in your region.

The regional councilor position willcontinue with a significantly modifiedrole. The purpose of the regional coun-cilors will be “To be the voice of theSection Liaisons to the StatisticsDivision.” Regional Councilors willeither be serving as an active sectionliaison or have been a section liaison.Their primary responsibilities willinclude actively recruiting SectionLiaisons in their region, coordinatingand leading section liaison communica-tions in their region, and participatingin communications with other regionalcouncilors. The current leadership ofthe Statistics Division envisions theRegional Councilor position as a prima-ry source of future leaders of theStatistics Division. Depending on thesize of a region and the interest level,we may consider creating more thanone Regional Councilor position in agiven region.

If you are interested in becoming aSection Liaison or a Regional Councilorcomplete the job interest notice in thisn e w s l e t t e r . ∆

In grateful memory of Ellis R. Ott,Honorary and Founding Member ofthe American Society for Quality, theStatistics Division is pleased toannounce the availability of up tothree $5000 scholarships to supportstudents who are enrolled in, or areaccepted into enrollment in, a mastersdegree or higher program with a con-centration in applied statistics and/orquality management. This includes thetheory and application of statisticalinference, statistical decision making,experimental design, analysis andinterpretation of data, statisticalprocess control, quality control, quali-ty assurance, quality improvement,quality management and related fields.The emphasis is on applications asopposed to theory. Funds supportingthese scholarships were donated toASQ by the Ott Foundation with theunderstanding that the scholarshipprogram would be administered bythe Statistics Division.

Qualified applicants must have agrade point average of 3.25 or higheron a 4.0 scale, or equivalent standingon another scale, in any field ofundergraduate study. Scholarshipawards are based on demonstratedability, academic achievement,involvement in student or professionalorganizations, faculty recommenda-tions, and career objectives.

Application instructions and formsmay be downloaded from:

www.asqc.org/about/divisions/stats

Alternatively, they may be request-ed by writing:

Lynne B. HareStatistical Engineering DivisionBuilding 820, Room 353National Institute of Standards andTechnologyGaithersburg, MD 20899-0001

Email: [email protected]

Scholarship applications are due tothe governing board no later thanApril 1, and awards are announced byJune 1. ∆

Ellis R. Ott Scholarships

Sponsored by the

Statistics Division

At this year’s Annual QualityCongress, the Statistics Divisionawarded 4 past chairs for their superi-or service to the Division by providingguidance in strategic planning anddivisional leadership. Awardees are:

Dr. Roger W. Hoerl, GeneralElectric R&D, Chair, 1990-91,Technometrics ManagementCommittee, Technometrics AssociateEditor, Fellow of ASQ and ASA.

Dr. Conrad A. Fung, Consultant,Chair, 1991-92, Hunter AwardCommittee (3 years), FTC ProgramChair (2 years).

Dr. Joseph O. Voelkel, RochesterInstitute of Technology, Chair, 1992-93, AQC Program Chair.

Ms. Nancy Belunis, Merck andCompany, Chair, 1995-96, past BylawsCommittee Chair, NominatingCommittee Chair, Auditing CommitteeChair, Program Committee Chair,Strategic Planning Committee Chair,previous Newsletter Editor.

Heartfelt thanks and congratula-tions to them all! ∆

Statistics Division

Presents

Testimonial Awards

Registration Desk Opens

41st Annual Fall Technical Conference • Mining Data For Quality Improvement

Thursday, October 16, 1997

COFFEE BREAK

LUNCHEONTopic: TBD

Speaker: TBDPresiding: Marvin D. Young, Phillip Morris, USA

W. J. YOUDEN ADDRESSTopic: Contexts of Statistical Practice

Speaker: Gipsie Ranney, Massey Graduate School of Business, Belmont UniversityPresiding: Don Emerling, Imation Corp.

7:30

8:00-9:00

0:00-10:30

0:30-12:00

2:15-1:45

2:00-3:30

4:00-5:00

A. Multiresponse O

Response Surface Mthe Use of Noise Va

Kristi Griffiths, ElRaymond H. Myers,

Multiresponse OTechniques: Role Variance on the Op

Robin Wurl, Rutgers

Moderator: John DafCorning,

A. Process C

Applying TechnologProcess Control

Douglas Wreath, Wre

Simplified Method fControl Chart Desig

Michael D. Mead, NShipbuilding

Moderator: Robert AuHoughtonInternatio

A. Design of Exp

Finding Winning Cwith Hyper-Greco-L

James L. Hansen, UTechnical Center

Running ExperiMultiple Error TermExperiment Is Run I

James M. Lucas, J.MAssociates, MalcoCampbell Soup Co.

Moderator: Jim StuarEastman Company

ASQ - C&PIDr: Marvin D. Young, Phillip Morris USAr-Elect: Kymm K. Hockman, DuPont Co.etary: Dorothy E. Sempolinski, Corning Inc.

ASQ - STATChair: Don Emerling, Imation Corp.C h a i r - E l e c t : Don Williams, Process Improvement

C o n s u l t a n t s

ASA - SPESChair: Greg Piepel, Battelle-NorthwestChair-Elect: John Cornell, University of FloridaSecretary/Treasurer: Kim Erland Vukovinsky, Pratt

General Conference Chair: Kymm K. HockmLocal Conference Chair: Pete Kosmides, NTreasurer: Mary Ann Gorko, DuPont MerckRegistration: Marg Bailey, Patterson-Schw

OFFICERS OF SPONSORING ORGANIZATIONS 1997 FALL

1

2

3

:15-10:00

WELCOME / PLENARY SESSIONTopic: Quality, The Future and You

Speaker: Steven P. Bailey, ASQPresident, DuPont CompanyPresiding: Kymm K. Hockman, DuPont Company

8:00

8:30-10:00

1

10:00-10:30

10:30-12:00

2

12:15-1:45

2:00-3:30

3

A. Case Study DOE

A Case Study in DOE in theManufacturing of Film

Stephen J. Caffrey, EastmanKodak Co.

Moderator: Wayne Taylor,Baxter Healthcare

A. Reliability & Measurement

Attributes of a SuccessfulReliability Improvement Metric

John A. Conte, DSC C o m m u n i-cations

Measurement Error StudiesUsing Sequential Sampling

Richard W. Andrews, University ofMichigan and Andrew J. Barnett,Ford Motor Company

Moderator: Michael D. Mead,Newport News Ship-building

A. Quality Management

The Fuzzy Front End of the NewProduct Development Process

Ken Scheffel, Thomas Hsiang,Universal Foods Corporation

Polishing the Gems

Mike Donnelly, The DowChemical Company

Moderator: Van Bowen,University ofRichmond

B. Customer Satisfaction

Analyzing Customer - Satis-faction Data Using Dual Scaling

Daniel R. Lawrence, RochesterInstitute of Technology

Moderator: Janice Shade,Nabisco, Inc.

B. Data Mining

Statistical Process Control forMassive Datasets

George C. Runger, Arizona StateUniversity, Thomas R. Willemain,Rensselaer Polytechnic Institute,James M. Grayson, Augusta StateUniversity, William W. Messina,Chrysler Huntsville ElectronicsDivision

Estimate of the Size Distribu-tion of Particles: A Data MiningSuccess

Walter Liggett, Robert Fletcher,NIST

Moderator: Christine Mastangelo,University of Virginia

B. JQT

Design of Multi-Level Frac-tional-Factorial ExperimentsThomas C. Bingham, BoeingCommercial Airplane GroupEstimating Common - CauseSigma in the Presence ofSpecial CausesRussell A. Boyles, Statist icalConsultantModerators: Douglas C.

Montgomery, JQTEditor, Arizona StateUniversity andG. Geoffrey Vining,incoming JQT Editor,University of Florida

C. Outliers

Identifying Multiple Outliers andInfluential Subsets in LinearRegression: A ClusteringApproach

David M. Sebert, Intel, Douglas C.Montgomery, Dwayne A. Rollier,Arizona State University

Moderator: Bob Brill,Monsanto ChemicalGroup

C. Analysis of Experiments

Follow-up Designs That Makethe Most of an Initial ScreeningExperiment

Robert W. Mee, University ofTennessee

Identifying Ridge Behavior inResponse Surfaces

Bruce Ankenman, N o r t h w e s t e r nUniversity, Soren Bisgaard,University of Wisconsin-Madison

Moderator: Dragan Kornicer,J.M. Huber Corp.

C. Sequential Experimentation

Sequential Simplex Design:Modifications and New Applica-tions

George Zeliger, Trilogy ConsultingCorporation, Boris Khurgin,Anvical-Simplex

Sequential Experimental De-signs for Sensitivity Exper-iments

Joseph O. Voelkel, RochesterInstitute of Technology

Moderator: Josef Schmee,Union College

egistration Desk Opens

ni Inner Harbor Hotel, Baltimore, Maryland

y, October 17, 1997

COFFEE BREAK

LUNCHEONTopic: Massive Data Problems

r: Jon Kettenring, ASA President, Bellcoresiding: Greg Piepel, Battelle-Northwest

B. Sources of Variation

Variation Reduction in Multi-Stage Processes

R. Agrawal, GE CorporateResearch and Development, J.F.Lawless, R.J. MacKay, Universityof Waterloo

Identifying Sources of Variationin Sheet Metal Stamping

Karl D. Majeske, Jay Baron,University of Michigan, PatrickHammett, University of Michigan

Moderator: David Rumpf,GE Aircraft Engines

B. Measurement

Implementing GR&R for aSemiconductor Wafer Fab StartUp

Joseph Conklin, Jennifer Tritt-schuh, TwinStar Semiconductor

E S D and Q-Q Plot: Two Pre-requisite Statistical Treatmentsfor Interlaboratory ExchangeData

Alex T. C. Lau, Imper ia l Oi l ,Canada

Moderator: Bryan Olin,The Procter & GambleCompany

B. Health Care

The Role of PerformanceMeasurement in AssuringManaged Care Quality

Randal l K. Spoeri , NYL C a r eHealth Plans, Inc.

Health Care Quality Tools,Techniques and Applications

David Simmons, Health CareEngineering, Inc.

Moderator: Melvin Alexander,GloboMax

C. Optimization

A Data-Analytic Approach toGlobal Optimization

Matthias Schonlau, Will iam J.Welch, University of Waterloo,Donald Jones, General Motors

Simulation-Based Optimizationof Complex Processes Via TheSimultaneous PerturbationMethod

James C. Spall, The JohnsHopkins University

Moderator: Peter Fortini,Cytec Industries

C. Technometrics

A Systematic Approach to theAnalysis of Complex InteractionPatterns in a 2-Level FactorialDesignJames J. Filliben, NIST, Ker-ChauLi, UCLAMonitoring Wafer Map Datafrom Integrated Circuit Fab-rication Processes for SpatialClustered Defects

Mark H. Hansen, Bell Lab-oratories - Lucent Technologies,Vijayan N. Nair, University ofMichigan, David J. Friedman,Integral, Inc.Moderator: Max D. Morris,

Editor Technometrics,Oak Ridge NationalLaboratory

C. Multivariate Process Control

Modeling and Monitoring ofMultivariable Dynamical Sys-tems under Feedback ViaCanonical Variate Analysis

Wallace E. Larimore, Adaptics,Inc., Dale E. Seborg, Yi Wang,University of California

Nonparametric Control Proce-dures for Multivariate Pro-cesses

Carlos D. Paternina, EduardoLerin, Tapas K. Das, University ofSouth Florida

Moderator: Stefan Steiner,University of Waterloo

mpany Lloyd Dixon, Northrop Grumman, Beth Reigel, Envi ronmentalElements, Frank Vojik. SCM Chemicals

Program/ASQ-C&PID: Sharon Fronheiser, Eastman Kodak Co.Program/ASA-SPES: Randy Tobias, SAS Institute, Inc.

ICAL CONFERENCE COMMITTEE

41st Annual Fall Technical ConferenceAttention: M. A. Bailey

P.O. Box 249Hockessin, DE 19707-0249

Please circle the sessions that you will most likely attend:Thursday, October 16: 1A 1B 1C 2A 2B 2C 3A 3B 3CFriday, October 17: 1A 1B 1C 2A 2B 2C 3A 3B 3C

Name:________________________________________________Badge Name:__________________________________________Company:_____________________________________________Address:______________________________________________City/State/Zip:__________________________________________Telephone:____________________________________________

Please circle whichever categories apply:

I am a: Member, Senior, Fellow, of ASQI am a: Member, Senior, Fellow, of ASAI am in Division: C&PID STAT ASA-SPES

I am NOT a member of ASQ or ASA

Registration Fees:Please submit one form for each person attending.

Two Days..........................................................................$190____Thursday, October 16th only.............................................$150____Friday, October 17th only..................................................$150____Student (ID Required).......................................................$ 75____Late Registration Fee (after September 15th)...................$ 10____

Pre-Conference Short Courses, October 15thUsing Designed Experiments to Shrink Health Care Costs .$150____Course Text (ASCQ members).........................................$ 36____Course Text (non-ASQ members) ....................................$ 40____How to Apply Statistical Thinking to Improve Results.......$150____Tour of Baltimore, October 16th........................................$ 45____Habor Dinner Cruise, October 16th...................................$ 40____Tour of Annapolis, October 17th.......................................$ 45____Total ..................................................................................$_______

MAKE CHECKS PAYABLE TO FTC-1997.No credit cards/purchase orders accepted.

All fees payable in U.S. Dollars only.Federal Tax ID #390912502 DETACH HERE

41st Annual Fall Technical ConferenceMail to: Omni Inner Harbor Hotel, Reservation Dept.

101 W. Fayette Street • Baltimore, MD 21201(410) 752-1100

Name: _____________________________________________________Company:___________________________________________________Address:____________________________________________________City/State/Zip: _______________________________________________Telephone:__________________________________________________Please reserve ___________ (# of rooms) for ____________ (# of peopleFor arrival on _____________________Depart on___________________Name(s) of person(s) sharing accommodations:______________________________________________________________________________________________________________________Check or money order enclosed in amount of $ _____________________Credit Card # __________________________- Exp. Date ____________Circle One: AMEX MasterCard VISA Other (specify) _____________I authorize the Omni Inner Harbor Hotel to charge my account for onight’s deposit and all applicable taxes.

Signature______________________________

CONFERENCE REGISTRATION FORM

HOTEL REGISTRATION FORM


The Statistics Division will sponsortwo short courses on Wednesday,October 15th, from 8:30 a.m. to 5:00p.m. The $150 fee for each courseincludes coffee breaks and lunch.Registration is limited, so

Register Early!!

Using Designed Experiments toShrink Health Care Costs

M. Daniel Sloan, Quality HealthSystems of America, Inc. During the first fifteen minutes of

this course, participants will completea factorial experiment, including dataanalysis! No algebra. No computers.No fooling. Sloan’s unique modelmakes it possible for managers,nurses, and doctors to use classical,designed experiments in all aspects ofhealth care. With the required coursetext, “Using Designed Experimentsto Shrink Health Care Costs”,(Quality Press, 1997), participants willspend the day mastering designedexperiment principles for health caremanagement. Case study and successstories will be presented. (Participantsare encouraged to bring a ruler, a pairof scissors, a virus free 3.5” data disk,and a hand held calculator.)

How to Apply Statistical ThinkingEffectively

Lynne Hare, NIST, Roger Hoerl,General Electric Company,and Ron Snee, Bell AtlanticCorporation.The purpose of this course is to

better prepare attendees to applyStatistical Thinking effectively withintheir own organizations. It will beginby discussing the need for widespreadunderstanding of Statistical Thinkingas a precursor to utilizing formalstatistical methods properly, in asystematic way. The philosophy willbe formally defined, and contrastedwith statistical methods per se.Attendees will work in teams toaddress problems, and suggestspecific ways in which StatisticalThinking could be applied morebroadly in business, government oracademia. Suggestions for over-coming typical barriers in itsapplication will also be provided.Lastly, each attendee will develop aplan to increase the applicationof Statistical Thinking in their ownorganization. ∆

41st Annual Fall Technical Conference

PRE-CONFERENCE

SHORT COURSES

INCOMING CHAIR’S MESSAGE


A lot of the activity of Division lead-ership is focused on integrating andplanning. We just completed a two dayoperational planning meeting where welaid out all of the work that needs to beaccomplished to support the FTC andAQC and to publish our newsletters. Wealso started some detailed planning for along range planning meeting this fall,were we will revisit the mission, vision,principles, and strategy of the Divisionand align our strategic and tactical plansto our mission and vision. We have acurrent five year strategic plan, whichwill be reviewed, revised, and updatedduring this long range planning meeting.At every FTC and AQC we have a tacti-cal planning meeting to review the workaccomplished and identify new opportu-nities and opportunities for improve-ment. I really think the past leadershipof the Division has done a great job atsetting up the systems, processes, andimprovement cycle to deliver on the“integration and planning” part of thejob. All I have to do is not screw it up!

Finally, I’d like to address the subjectof evaluation. I believe evaluation is the“job” of all of us, but mostly it’s the“job” of the members of the Division.The leaders of the Division can evaluatehow proposed strategies, tactics, prod-ucts and services fit with the missionand vision of the Division. But the mem-bers of the Division must evaluate howwell they believe that the selected strate-gies, tactics, products, and services filltheir individual need to be more effec-tive in their lives. The Statistics Divisionof ASQ is a big organization, about12,000 of you have chosen to becomemembers of the division. At a very basiclevel, the purpose of any volunteerorganization is to add value to everyperson who chooses to become a mem-ber. Otherwise, why be a member. Youhave made a decision to spend some ofyour money on membership with thetacit assumption that you will receivesufficient benefit for that money. This isa definition of value. One of our mem-bers, Bert Gunter, sent the officers an e-mail last year that challenged our focus.A phrase from that letter sticks in mymind as I think about the purpose of adivision and the “job” of the Chair. Hewrote, “What we seem to lack is the fog-giest idea of why people join, stay, orleave the division…” That seems to be areal important statement. Those of usinvolved in the leadership of theDivision all would respond (with theappropriate amount of indignation!) that

we have spent a great deal of time andenergy trying to understand membersreasons for joining, staying, or leavingbut I wonder if Bert isn’t right, that wedon’t have the foggiest idea of why. Andmaybe the “why” can be discovered byprobing the purpose of a division. SoI’m asking you, the current members ofthe Statistics Division; What is the pur-pose of a division? Why do you chose tostay as a member of the StatisticsDivision? Is it inertia? – I just fill out theASQ membership form the same everyyear. Or do you make a value decision?– I choose to spend my $8.00 on theStatistics Division because it deliversmore than $8.00 in benefit to me.Assuming you are in the second group,What benefits do you receive? What doyou think of our current mission and

vision? If you were the Chair of thisDivision, what is the one change youwould make in the first week to betterserve the members? What other ques-tions come to your mind?

I encourage you to send me answers,questions, or comments to any of thesequestions or call me to discuss them.Send me your answers, questions, com-ments, etc. I can be reached by e-mail –[email protected], by surfacemail – 1 Imation Place, Oakdale, MN55128, or by phone – 612-704-4606.

I’ve decided! My job is to work withas many other members (yes, I’m amember too) as possible to make thisDivision a valuable resource for all of usand, through our work, make thisDivision a valuable resource to oursociety. Help me do “this job.” ∆

Part I1. Introduction

It is well known in the world of statistics that theExponentially Weighted Moving Averages (EWMA) was ini-tially developed as an empirical model for forecastingbased on real-life data. It can also be shown that EWMA isa very special case of the Auto Regressive Moving Average(ARMA) model. The relationship between ARMA andEWMA was addressed by many and discussions on this canbe found in a plethora of articles and books (see forinstance, Box and Jenkins 1970, 176, Pandit and Wu 1983,Chatfield 1983).

EWMA has been found to be a powerful and sensitivemethod which can be used as an effective processmonitoring technique, as well as a useful prediction toolfor short term forecasting. It is based on exponentialsmoothing.

In order to use exponential smoothing, (sometimes alsocalled Geometric smoothing), we assume that the data setis free of any trends but there are short oscillations present.Time-series found in real life often contain componentssuch as seasonal patterns and trends, which can be mea-sured. Established filtering techniques are available toreduce or to remove these components from the data thatproduces a trend free and stationary time series. TheEWMA, an asymmetric filter, is computed by taking aweighted average of past observations with progressivelysmaller weights over time. This procedure is effective,when the zone of acceptance for processes is very narrow,and we need a more sensitive method to detect smallershifts. On the other hand, if the acceptance zone of theprocess is not that narrow, using EWMA to detect smallshifts will not be an efficient or an economical method.The most important advantage of EWMA comes from itsflexibility in using data on the history of the process. Thisis done by selecting a weight factor and effectively using itto achieve the desired balance between the older data andmore recent observations.

Ordinarily, the weight factor or the prediction parameteris selected arbitrarily or on the basis of some strong intu-ition or “gut” feelings. In order to come up with a reason-ably good prediction, the weight factor needs to be select-ed more precisely. Different approaches have been devel-oped to do this. In this short note, the prediction parame-ter is selected by minimizing error in the predicted values,which seems to be a simple approach for practicepurposes.2. EWMA

The EWMA, in its basic form, is computed by choosinga number ν, where 0 < ν < 1 such that for νj, we can havethe decaying or decreasing weights by adjusting the valueof ν carefully. However, if we multiplied χ

t–j, by theseweights νj and added them to get the forecast, it is no

longer an average, since the weights do not add to one(∑νj ≠ 1). In fact, they add to 1/1 – ν). So, if we want aweighted average, we should take the weights (1 – ν)νj,

which add to one. We can then write a recursive relation-ship between an observation χ

t+1, at time instant t + 1 as afunction of previous observations χ

t–j for all j,

χt+1 = ∑

or

χt+1 = ∑

By substituting γ = 1 – ν, 0 < γ < 1This simple mathematical manipulation allows us to

expand the power series. It therefore ensures exponentiallydecaying weights assigned to the observations and the sumof weights equal to unity. This exercise can be found inany time-series book with a discussion on EWMA.Prediction of a new forecast can be easily generated byusing the recursive nature of EWMA updating the old fore-cast and it is not necessary to use the entire past data setevery time. The time series model for which exponentialsmoothing yields optimal predictions was pointed out byMuth (1960).

The statistic Zt, is a weighted average of the currentobservation and all the past observations. This Zt is plottedin the EWMA chart, and is defined by the recursive rela-tionship:

Zt = (1 – γ)Zt–1 + γXt, 0 < γ < 1where Xt is the sample observation at time t. It has been

assigned a weight factor, γ. The weighted Zt defined abovehas the auto regressive nature and is a normally distributedvariable (by central limit theorem). An EWMA chart withtwo-sided control limits can then be constructed to monitorZt.

In plotting the EWMA control chart, the central line rep-resents the target value µ0 for the process average µ in acontrolled process with no fluctuations. Also, E(Zt) = µ.

The upper and the lower control limits are plotted at thelevels µ ± κσxt, where κ denotes a suitable constant and σx,

the standard deviation of the process. The variance σ2z t of

Zt can be derived to be as

Var(Z t) = σ2zt = σ2

x (–––)γ2–γ .(1–(1–γ)2t),

and the two-sided EWMA 3-sigma control limits aregiven by

µ±3σχ √ in the limiting form as t → ∞ (see referenceRoberts 1959, Montgomery 1996).

Remark 1 It might be worthwhile to mention, that in adata set from a given process, the observation might be anindividual observation at time t or an average of a sample


Mini PaperA Note On EWMA

Sumedha Sengupta, Consulting Statistician

5566 Felicia Ave., Livermore, CA 94550-8126

γ2−γ


(1 – γ) j γ χt–j

(1 – ν) ν j χt–j

∝

∝

i = 0

j = 0


or a subgroup at time t. In case, Xt represents the average ofa subgroup, the standard deviation used in the controllimits have to be modified based on the subgroup size. Thissituation will occur, when the EWMA chart is being con-structed by modifying a Shewhart control chart already inuse and the initial data was collected in the form of sub-groups.

2.1 An Application in ProductionWe now describe an easy application of EWMA for short

prediction in an industrial setting, in the line of Hunter(1995).

In this application, we need to make a forecast for theaverage weekly production of a particular consumer item.A Shewhart chart was already in use to monitor this aver-age. By the nature of this average, older observations areless important to the forecast than the recent ones. Toexpress this mathematically, let us assume that the nextweek’s forecast values for the average production is givenby

• Yt +1 = Yt + γe t = Yt + γ (Yt – Yt )

Where,

• Yt = current week’s actual average production

• Yt = current week’s predicted average production

• Yt +1 = next week’s predicted average prediction

• et = current week’s error in prediction,

et = Yt – Yt

In this simple application, we assume that a weightingfactor of γ = 0.5 is selected to weight the error. We alsoassume that a predicted value (which can also be a targetvalue) for the current week is given as 40 units. We set theinitial predicted or target value as Y1 = 40. Suppose theactual observed value on the current average production isYt = 42 units. So the first predicted value is given by therelationship as

• Y2 = 40 + 0.5(42 – 40) = 41.Suppose that the next observation is given by Y2 = 45,

then• Y3 = 41 + 0.5(45 – 41) = 43.These observed and predicted values can be plotted on

a graduated chart to show the relationship between them.In this application, the 3σ limits (Shewhart) for the

observed values were 40±12, where the average weeklyproduction, Yt had a standard deviation of 4. The modified3σ limits for the EWMA would be 40±4.63 with the weightfactor γ = 0.5. We find that these are much tighter limits. Inusing the EWMA, the Shewhart limits become less and lessimportant.

In this application, the initial purpose was to use theEWMA to get a forecast value and by choosing γ = 0.5 mid-way between 0 and 1, ensures a fairly acceptable predic-tion. [This also happens to be consistent with the fact thatEWMA lies between the CUSUM and the Shewhart charts(see Part II)]. However, the selection is still arbitrary. Inorder to select γ, one looks for a simple and practicalmethod and how to make this selection.

In Part II of this article, we discuss this with an applica-tion, and include a list of references cited in the text of thisnote.

Part II1. How to Select the Weight Factor?

In selecting the weight factor, one makes a decision onhow much of the previous data influences the presentdecision. The larger the value of γ, the smaller is the effectof the past history of the process and the more recent val-ues become more important for the prediction process. Ifwe set the weight factor, γ, equal to 1, then the propertiesof the EWMA control chart resemble more and more thatof the Shewhart chart. The closer γ is to 0, the more theproperties resemble the Cumulative Sum Control Chart(CUSUM) chart. From this, we can safely conclude that theEWMA chart lies in between the Shewhart chart and theCUSUM chart, and one can choose its position between thetwo charts, by adjusting the weighting factor, γ. Values ofthe weight factor γ depend on the properties of a giventime-series. Lower values, say between 0.1 and 0.3 areused in a prediction or forecast where past observationsare given more importance than the recent observations. Itis a common practice to select a high value of γ closer toone, when the most recent observation will have the maxi-mum weight.

One practical approach is to focus on the predictionerror generated in the prediction process and to select avalue of γ, which produces the minimum mean squareerror, ∑e2 (MSE) (see Box and Jenkins, 1970, 1976,Chatfield, 1983). This is similar to the least square methodwhich is commonly used to estimate the parameters in aregression fitting. In the following section, summary of anapplication is described in which the value of γ has beenselected by minimizing the MSE.

1.1 A Second ApplicationOne area in which we wanted to use the EWMA tech-

nique for short term prediction was on a set of data fromthe NASDAQ stock volumes, a well accepted example of atime-series with Auto Regression properties. These twocomponents are the time-sequenced average daily values(D) and the Closing Values (C). The data set showed notrends and very few erratic oscillations. We first tabulated(Table 1.1) and plotted (Fig. 1.1) a sample of the data fromthese two series to study their nature, and found that theyhave slightly different patterns as was intuitively expected.

MINI PAPERContinued from page 15


^

^

^

^

^

^

^ ^ ^ ^


Table 1.1 Data on Daily (D) and Closing (C)# Daily-D Close-C1 361.90 360.702 363.0 365.93 359.5 367.74 360.7 368.55 365.9 366.16 360.7 365.77 365.9 363.88 367.9 365.19 362.7 368.610 367.7 367.411 368.9 367.312 366.0 370.213 368.5 371.214 368.50 371.2015 367.0 355.016 368.0 359.917 369.4 355.518 364.5 354.619 366.1 356.120 366.1 352.221 363.2 351.422 365.7 351.423 367.4 353.824 363.4 352.425 363.8 352.7

A short program (in C language) was written to computethe predicted values for C and D. The program first gener-ates a series of values for γ, and then computes the predict-ed values for each of the observed values for C and D, andthen interactively generates individual plots for these pre-dicted values for all γ = 0.01 to 0.91 at an interval of 0.1.These plots demonstrated different patterns for the predict-ed values and showed how the oscillations slowlysmoothed out with the change in the γ values. The pro-gram also computed the MSE for the C–values and theD–value MSEs, which were stored in a data file and thenplotted with the help of an interactive graphical softwarepackage.

The plots and the analysis clearly displayed the patternsof the mean square error (MSE). The arrows in Table 1.2,indicate the values which appear as the minimum points inthe plot (Fig. 1.2). The corresponding values of γ, indicatedby the asterisk, *, are the values selected as the predictionparameters.

Table 1.2 MSE for C and Dγ MSE–C MSE–D0.01 48.53 19.580.11 38.37 08.610.21 27.15 06.520.31 20.97 06.520.41 17.64 05.770.51* 15.75 05.57 ←0.61 14.67 05.970.71 14.09 06.430.81* 13.91 ← 07.020.91 14.05 07.74

2. Concluding RemarksThe minimum error points in Table 1.2 corresponded to

γ = 0.81 for C–values and γ = 0.51 for the D–values, showthat these values correspond to the 9th and the 6th valuesof γ in the MSE plot given in Fig. 1.2. This concludes theC–value has more weight to the recent observations thanthe D–value does. Such conclusion can also be intuitivelyagreed upon. A closing value ought to have more weighttoward the next prediction than the daily average value.

Main interest in this note has been in using EWMA forshort term prediction, where large amount of data satisfy-ing basic assumptions was available. In recent years, addi-tional techniques based on neural network (generalizednon-linear models) are being explored by many for time-series data form the financial world.

Simulation was not used in this method. A fairly largenumber of data points were available from the sameprocess. Plotting control charts was not one of the objec-tives, but we do mention in subsection 2.1 of Part I , howthe weight factor relates to the control limits, in case one isinterested in doing so.

Over the years, numerous papers based on EWMA havebeen published in statistical journals and it is a formidabletask to include an exhaustive bibliography here. Books andarticles cited in the body of the text are listed in the refer-ences.

A short talk based on this topic was presented at theSanta Clara ASQC section in 1994. The author may be con-tacted for further details on this study in which other meth-ods were compared to the EWMA predictions.

The author wishes to acknowledge helpful discussionswith her husband Sailes K. Sengupta, Ph.D. ∆

3. References1. Box, G.E.P. and Jenkins, G.M. (1970, 1976) Time-Series Analysis,

Forecasting and Control, Holden Day Publications, San Francisco, CA.2. Chatfield, C. (1984) The Analysis of Time-Series-An Introduction. Third

Edition, Chapman and Hall, New York, NY.3. Hunter, Stuart (1995) “Just What Does An EWMA Do?” Mini-paper, Part

1 & 2, ASQC Statistics Division Newsletter, N.J.4. Montgomery, Douglas (1996), Introduction to Statistical Quality Control.

Third Edition, John Wiley & Sons, New York, NY.5. Muth, J.F. (1960), “Optimal Properties of Exponentially Weighted

Forecasts.” J. Am. Stat. Assoc., V.55, P. 299.6. Pandit, S.M. and Wu S.M. (1983), Time-Series and Systems Analysis with

Applications. John Wiley & Sons, New York.7. Roberts, S.W. (1959), “Control Chart Tests Based on Geometric Moving

Averages.” Technometrics, pp. 239-250.

MINI PAPERContinued from page 16


The Quality and Productivity sec-tion of the American StatisticalAssociation will offer seven continuingeducation courses at the 1997 JointStatistical Meetings in Anaheim thisAugust. A brief description of thecourses follows. Please visit the Q&Ppage on the internet for more com-plete information and full abstracts athttp://funnelweb.utcc.utk.edu/~asaqp

Designs and Models forExperiments with MixturesAugust 10, 8:30 AM - 5:00 PMOptional evening session 6:30 PM -8:00 PM

A one day course by John Cornell,Univ. of Florida and Greg Piepel,Battelle. This course will provide anoverview of various approaches andtechniques used in designing mixtureexperiments and analyzing the result-ing data. The course will concludewith an evening consulting/discussionsession using real examples submittedby attendees.

Analytical Tools for Assessing andUsing Customer SatisfactionSurveysAugust 13, 8:30 AM - 12:30 PM

A half day course by Susan Devlin,Bellcore. This course will explore avariety of analytical methods forextracting information from customerquality assessments. Using examplesdrawn primarily from communicationsservice companies, the fundamentalsof these methods will be reviewedwith an emphasis on their strengthsand weaknesses for supporting busi-ness decisions.

How to Create a Course Homepagethat Makes Classroom InstructionMore EffectiveAugust 12, 9:00 AM - 1:00 PM

A half day workshop by RAMonLeon and Lance Milner, Univ. ofTennessee. The many advantages ofhaving a course homepage will be dis-cussed on the basis of real examples.Participants will learn how to create acourse homepage using new easy-to-use authoring tools. Techniques forworking with images, mathematicalnotation, and computer screen shotswill be presented.

Creative Methods for Teaching andExplaining Statistics and QualityAugust 11, 9:00 AM - 5:00 PM

A one day workshop by CarlWetzstein and Sharon Fronheiser,Eastman Kodak Company. Anyonewho teaches or needs to explainstatistics and quality tools has strug-gled with the fact people are lessthan enthusiastic about these topics.What can we do? Let people discoverthat statistics and quality methodsdo apply to their work, are usefuland that learning them can be fun!Statistics can be fun! In this workshopyou will participate in a large varietyof activities that you can use to teachor explain SPC methods, basicstatistics and design of experiments.

Statistical Models for the Reliabilityof Repairable SystemsAugust 12, 1:00 PM - 5:00 PM

A half day course by Steve Rigdon,Southern Illinois Univ. at Edwardsville.This course will cover the differencesin notation and terminology betweenrepairable and nonrepairable systems.Several models for repairable systemswill be discussed. In addition, statisti-cal inference for the power lawprocess will be examined in detail.Inference for other distributions willalso be covered. Graphical methodsof displaying data from repairablesystems will be emphasized.

Hotelling’s T-Square: A ControlStatistic for Multivariate ProcessesAugust 9, 9:00 AM - 5:00 PM

A one day course by Robert Mason,Southwest Research Institute, JohnYoung and M. Pam Langley, McNeeseState Univ. This course focuses onusing industrial examples to demon-strate the versatility of the T-Squareas a control statistic for multivariateprocesses. Recent developments inthe use of the T-Square chartingstatistic will be explored, and thevarious aspects of this statistic willbe examined using computer soft-ware. Students will be shown how toapply and interpret the T-Squareprocedure.

A Problem Resolution Toolkit:Invention Machine Lab (IMLab)Software Tools to Aid ConceptSelection for Robust DesignAugust 14, 8:30 AM - 12:30 PM

A half day computer technologyworkshop by Val Tsourikov, InventionMachine Lab, and Stephen A. Zayac,Ford motor Co. The emphasis of thiscourse will be on application to practi-cal problems and all concepts will beillustrated using real world examples.The workshop is intended for theapplied statistician wanting to masterthe toolkit’s functional analysisprocess. ∆

Joint Statistical Meeting

Continuing Education Courses

Fall Technical

Conference

ScholarshipsThis year, the Statistics Division will

offer up to 5 scholarships to cover thecost of attending the Fall TechnicalConference in Baltimore. The confer-ence will be held on October 16-17,1997. The conference is sponsoredjointly by the Chemical and ProcessIndustries Division and the StatisticsDivision of ASQ, and the Section onPhysical and Engineering Sciences ofthe American Statistical Association.

Scholarship covered expensesinclude the cost of conference regis-tration, meals (up to $50), lodging fortwo nights and transportation to andfrom the conference location. To qual-ify, applicants must be currentlyenrolled in undergraduate or graduateprogram in statistics or quality man-agement. Recipients may be asked toserve as room monitors for a sessionat the conference and will be asked towrite a brief article about their confer-ence experience for this newsletter.

Applicants should send a letter ofinterest together with a letter of rec-ommendation from a major professorby August 1 to:

Lynne HareNISTBuilding 820, Room 353Gaithersburg, MD 20899 Notifications will be mailed by

September 1. ∆


works as both an academic and apractitioner to further enhance ourunderstanding of quality and how toachieve it.

Kano Named Second DemingLecturer

Dr. Noriaki Kano has been selectedto give the second annual DemingLecture at the JSM on 12 August 1997in Anaheim, CA. This selection is par-ticularly appropriate in light of Dr.Deming’s long association with theJapanese quality movement.

Over the course of nearly threedecades, Dr. Kano has established areputation well beyond his nativeJapan as a premiere practitioner andacademician in the quality and produc-tivity arena. Indeed, he is a muchsought-after management counselorthroughout the world. He has workedclosely with leading Japanese corpora-tions, including such household namesas Honda Motors, Komatsu, andMatsushita. In the U.S., he was one ofthe lead consultants for Florida Power& Light when it became the first non-Japanese company to win the covetedDeming Prize. Kano has also consulted

with Hewlett-Packard, Goodyear,Xerox, 3M, and a number of otherleading American firms. In addition,he has worked with other firmsthroughout Asia, Europe, and LatinAmerica.

A living example of Dr. Deming’sprinciple of balancing theory and prac-tice, Kano has maintained an activeacademic career. He received hisDoctoral degree in Engineering fromthe University of Tokyo in 1970. From1970 until 1982 he was with the facultyof the University of Electro-Communications. He then moved toDepartment of Management Science atthe Science University of Tokyo wherehe became a full professor and thenhead of the department in 1988. Six ofhis papers and one book havereceived Japan’s prestigious NikkeiQuality Control Literature Prize.

Kano has been a member of Japan’shighly esteemed Deming PrizeCommittee since 1973. He is perhapsbest known internationally for hiswork on “Attractive Quality Creation”,developed in the 1970s and publishedin 1984. This work has had a profoundeffect on how organizations define andunderstand quality from a customer’sviewpoint. As part of this work, hedefines three levels of customer per-ceptions, including aspects that “must

be” present to avoid disappointment,those where “more is better”, andthose that result in delight if they arethere at all.

He is currently a councilor for theJapanese Union of Scientists andEngineers (JUSE), where he has alsoserved as a Board member and editorof Hinshitsu. He is on the InternationalEditorial Board for the EuropeanOrganization for Quality and has sup-ported numerous organizations inCentral and South America. In 1995, hebecame a fellow of the AmericanSociety for Quality Control.

Kano’s influence around the worldtakes on other forms as well. He is afrequent keynote speaker and invitedlecturer for quality and productivityorganizations on four continents. Hehas played an active role in manyinternational agencies including theUnited Nations Industrial DevelopmentOrganization, the Agency for OverseasTechnical Scholarship, and theInternational Standards Organization.

“Dr. Kano is one the most importantmanagement consultants in the worldtoday,” says Dr. Brian Joiner, lastyear’s Deming Lecturer and now chairof the Deming Lecture Committee.“We are honored to recognize hiscontributions to both the theory andpractice of management.” ∆

SECOND ANNUALDEMING LECTUREContinued from page 8

MembershipReport

For the first time in the last five years,the total number of members of our divi-sion (July 1996 - June 1997) did notdecrease. Statistical analyses of the datasuggest the number of members is level-ing off. We believe that this is a result ofDivision efforts to provide better servicesto our members.

Beginning on July 1, we will have localsection representatives. This new positionwill allow us to further increase commu-nication with all of you, our members.We need your support. If you are inter-ested in becoming one of the local sec-tion representatives, please complete theopen position request form and forwardto Beth Propst (Past Chair).

Finally, I would like to take this oppor-tunity to thank all the members (and non-member friends) that stopped by ourbooth to experience our Virtual Academy.Your comments and suggestions weregreatly appreciated. ∆


SPAIG Workshop A SuccessThe Statistics Partnerships among Academe, Industry, & Government (SPAIG) Workshop was held in Raleigh, NC on 30-31 May.

Eighty senior-level statisticians from academe, industry, and government attended, with an approximately equal number attending fromacademe and industry/government organizations.

Bob Starbuck (Wyeth-Ayerst) opened the workshop by mentioning the opportunities that could be achieved by better and morewidespread partnering relationships, including expanding and improving the use of statistics; increasing the value of statistics tosociety; improving the educational experience of students; improving the career decision process & outcome; increasing communica-tions among all statisticians; enabling greater awareness of each other’s needs, issues, & concerns; improving the self-image of statisti-cians; making statistics a more rewarding profession; and ensuring that statistics is a growth field.

Ron Iman (Southwest Technology Consultants) provided an overview of partnering models and the great success that partneringhas had and is having in the semiconductor industry. Ron presented several partnering models and also provided data that show thatover 90% of new PhD statisticians are employed in industry and government rather than in academe, thereby emphasizing the impor-tance of training statisticians to achieve the skills required to be successful in industry and government.

G. Rex Bryce (Brigham Young Univ.), Dean Isaacson (Iowa State Univ.), John Spurrier (Univ. of South Carolina), and Bob Hogg(Univ. of Iowa) followed with success stories of partnering with industry through internships, partnering with other departments oncampus, a capstone course in statistics, and partnering with engineering, respectively. The needs of industry, government, and acad-eme were then expressed by Gene Meieran (Intel Corp.), Cynthia Clark (US Bureau of Census), and Dan Solomon (NC State Univ.).

Ron Snee (NYNEX) presented the keynote speech He pointed out that global competition and advances in computer technologyare forcing changes in how US corporations are managed. The result is an expanding role for statistics and statisticians. This expansionalso results in changing roles for statisticians because the needs of industry have changed. In short, the needs are managerial as well astechnical. The managerial needs are less well defined and are challenging to satisfy.

Ron concluded by saying that partnering with industry is needed to effectively and efficiently identify how to align statistical educa-tion and research with these new needs. Personal change is required to take advantage of the expanding role. We are reminded thatthose who do not respond to their changing world will have decreasing influence in their world. George Bernard Shaw noted that “Ifyou can’t change your mind, you can’t change anything.”

Workshop participants were asked to identify the consequences of maintaining the status quo, i.e., doing nothing to improve thepartnering relationships between academe and industry/government. Their responses are summarized in the following Pareto chart:

Consequences of Maintaining Status Quo of Statistical Partnering

A Statistics as a discipline viewed as irrelevant; decline in influence I Reduced $, resources, supportB Non statisticians will do statistics J Good applied statisticians not available to hireC Decline or elimination of statistics departments & professional societies K Fewer employment opportunitiesD Students not prepared to solve future problems L Continued alienation among A/GE Slow growth in technical advances in statistics M Outside forces will determine the direction of theF Industry/government will do in-house statistics training disciplineG Reduced contribution of statistics to society N US industry will be less competitiveH Failure to attract good students O Miscellaneous

There was clearly a strong belief among those present that the role of statistics and statisticians will diminish if the status quo ismaintained. Negative consequences would include fewer students choosing statistics as a career, decline or elimination of statisticsdepartments, and fewer employment opportunities for statisticians.

Participants were then divided up into four groups to address the following four topics:1. Short-term visits between statisticians in academe and industry/government2. Long-term collaboration between academe & industry/government statisticians3. Partnering with other disciplines4. Industry/government input into academic statistical programs & curriculaAndy Kirsch (3M Co.), Lynne Hare (Nat’l. Inst. of Standards & Technology), Roger Hoerl (General Electric), Dean Isaacson, and

Susan Schall (Dupont Engineering Polymers) assisted individual groups in the use of affinity mapping, interrelationship digraph, andmulti-voting TQM tools to:

1. identify why partnering was not occurring or not occurring as much as it could be2. group identified causes into related clusters (root causes)3. determine which root causes exerted the most influence on other root causes4. identify solutions for addressing the most influential root causes5. determine which solutions would be focused on to develop specific action plans.


This process led to the development of the following specific action plans:

WHAT WHOCreate Contact Lists to Facilitate Partnering and Ron Randles (Univ. of Florida)Make Available As Appropriate Ron Bosecker (USDA Nat’l. Ag. Stat. Service)* Academic consulting list Rita Patwardhan (SmithKline Beecham)* I/G internships Bruce Rodda (Schering-Plough)* Speakers bureau Bill Rolfes (3M Co.)* I/G contacts Jim Rosenberger (Penn State Univ.)* Academic department heads Bob Starbuck (Wyeth-Ayerst)* Include web sites in every Amstat News issue Bill Wilson (Univ. of North Florida)Conduct I/G Salary Survey Ray Waller (ASA)* Improve understanding of value of statistics & statisticians in I/G Rich Allen (USDA Nat’l. Ag. Stat. Service)* Improve recruitment of students into statistics field Alan Hopkins (Genentech)* Provide evidence to academic administrators of value of statistics Bob Starbuck (Wyeth-Ayerst)Regional SPAIG Workshops Designed and Implemented Ray Waller (ASA)* A/I/G Statisticians Robert Ahlbrandt (Hoechst Marion Roussel)* Other Disciplines William Caldwell (Univ. of North Florida)* Determine regional I/G continuing education needs Anne Parkhurst (Univ. of Nebraska-Lincoln)* Identify and publicize successful models of A=>I/G courses Jim Rosenberger (Penn State Univ.)

Nancy Smith (US FDA)Ron Snee (NYNEX)Kathleen Stralka (Science Appl. Int’l. Corp.)

Include I/G Statisticians on Academic Advisory Boards and John Spurrier (Univ. of South Carolina)and Review Teams William Caldwell (Univ. of North Florida)

Tom Gerig (NC State Univ.)Doug Kelly (Univ. of North Carolina)Dick Kryscio (Univ. of Kentucky)Andy Kirsch (3M Co.)Anne Parkhurst (Univ. of Nebraska-Lincoln)Kathleen Stralka (Science Appl. Int’l. Corp.)

“How To” Manual on Short-term Visits Between A/I/G Statisticians Paul Tobias (Sematech)James Colaianne (Merck)Geetha Ramachandran (CA State Univ.-Sacramento)

Incorporate Collaboration into A/I/G Job Descriptions and Job Goals Tom Gerig (NC State Univ.)Yahia Ahmed (US IRS)

ASA Award for Partnership Alan Hopkins (Genentech)James Colaianne, Merck

Assess Past and Current Long-term Collaboration Between A/I/G and Robert Kushler (Oakland Univ.)Develop Process for Long-term Collaboration Robert Ahlbrandt (Hoechst Marion Roussel)

Ron Iman (Southwest Technology Consultants)Andy Kirsch (3M Co.)Jon Lemke (Univ. of Iowa)Ron Snee (NYNEX)Mark West (Auburn Univ.)

Survey Collaboration with Other Disciplines Bob Hogg (Univ. of Iowa)• How Measured? Dipak Dey (Univ. of Connecticut)• Incentives Robert Jernigan (American Univ.)

Geetha Ramachandran (CA State Univ.-Sacramento)Robert Rathmacher (Eli Lilly)

Collect and Edit Partnering Success Stories David Fitts (SmithKline Beecham)Manage SPAIG Web Site Bill Parr (Univ. of Tennessee)Chairperson for action item indicated in bold text.

SPAIG SUCCESSContinued from page 20

Participants who expressed at the conclusion of the workshop an interest in working on one or more of the action items are identi-fied in the above table. Anyone who would like to participate in or contribute to an action item (including workshop participants notlisted above) should contact the appropriate chairperson for the action item.

Volunteers participating in these initiatives will be developing detailed action plans with milestones, responsibilities, and measuresof success. The action plans will then be implemented, and progress reports will be provided at the JSM and in Amstat News articles.

Additional information on the results of this SPAIG Workshop, including full text of many of the presentations and content of over-heads summarizing the results from work sessions held during the workshop, can be found on the SPAIG web site at http://funnel-web.utcc.utk.edu/~wparr/spaig.html. Information on detailed action plans, milestones, responsibilities, measures of success, andprogress as it occurs will be regularly posted to the web site.

We greatly appreciate the financial support of the NSF for supporting the planning of this workshop (cf. Amstat News, 235, pp. 19-20) and of the following professional societies to cover the costs of coffee breaks and continental breakfasts: ASA Section on Physical& Engineering Sciences, American Society for Quality Control, ASA Biopharmaceutical Section, and ASA Section on Quality &Productivity. Thanks go also to Lynne Hare and Susan Schall for developing the workshop agenda, to Tom Gerig for his excellent jobof doing the local arrangements for the workshop, and to all those who facilitated and participated in the workshop. ∆


1997-1998 Officers and Committee Chairs

Beth Propst

Phone: (630) 443-8132Fax: (630) [email protected]

Past Chair

ChairBeth Propst

ChairBeth Propst

Division ChairDon Emerling

Chair ElectDon Williams

SecretaryBob Mitchell

TreasurerJanice Shade

BylawsCommittee

NominatingCommittee

NominatingCommittee

ExaminingCommittee

AwardsCommittee

CertificationCommittee

Hunter Award

PublicationsCommittee

StandardsCommittee

EducationCommittee

ASA Q&PLiaison

ChairDon Emerling

ChairDon Emerling

AQC ProgramRichard Kleinknecht

Phone:Fax:

AQC Short CoursesMarcey Abate


FTC ProgramSusan Albin

Phone: (908) 445-2238Fax: (908) 445-5467

[email protected]

FTC Short CourseOpen Position

Applied Stat. Conf.Frank Alt

Phone: (301) 405-2231

ChairBob Perry

Phone: (612) 330-8144Fax: (612) 330-1893

[email protected]

ChairLynne Hare


AuditingCommittee

ChairDon Williams

ChairDon Strickert

Phone: (214) 334-4305Fax: (214) 334-4444

[email protected]

How-To EditorWalter Liggett

[email protected]

Glossary EditorJacob Van Bowen

Phone: (804) 289-8081Fax: (804) 287-6444

[email protected]

Acquisitions Coord.Open Position

New Prod. Coord.Open Position

Briefings EditorRick Lewis

Phone: (314) 694-7735Fax: (314) 694-5614

r a l e w i @ c c m a i l . m o n s a n t o . c o m

ChairNick Martino

Phone: (508) 534-2556Fax: (508) 840-0112

[email protected]. com

C h a i rGalen Britz


Strategic PlanningC o m m i t t e e

Newsletter EditorJanice Shade


Chair Ed Schilling

Phone: (716) 475-6129Fax: (716) 475-5959

[email protected]

Chair Marilyn Hwan


NewsletterCommittee

ChairJL Madrigal


BBS AdministratorMark Kiel

Phone: (708) 849-2500Fax: (708) 841-0661

[email protected]

Joseph O. Voelkel

Phone: (716) 475-2231Fax: (716) 475-5959

[email protected]

MembershipCommittee

Don Emerling

Phone: (612) 704-4606Fax: (612) 704-3892

[email protected]

ChairDon Williams

Phone: (940) 243-1147Fax: (940) 387-2251

[email protected]

Chair-ElectBob Mitchell


SecretaryJanice Shade


Treasurer

Division Officers

↓ ↓ ↓ ↓


Job Openings

We currently need volunteers to serve in the following positions:

■ Regional Councilor■ Section Liaison

If you are interested in volunteering for one of these positions, please fill out the form below and send it to:

Past Chair Beth Propst41 West 202 Whitney Road

St. Charles, IL 60175

In addition, proposals for newsletter Mini-Papers and Basic Tools articles are always welcome.

Member Interest Form

Name: ______________________________________________________________________Date: ______________________

Title: _________________________________________________________________Member No.:______________________

Address: _______________________________________________________________________________________________

_______________________________________________________________________________________________________

_______________________________________________________________________________________________________

Phone: _____________________________________________________Fax:________________________________________

E-mail: ________________________________________________________ Membership: ____ Reg. ____ Sr. ____ Fellow

Education/Certifications/Experience: _______________________________________________________________________

Time Availability/Company Support: _______________________________________________________________________

Please check or circle all areas of interest:___ Education Committee ___ Awards Committee ___ Standards Committee___ Examining Committee ___ Certification Committee ___ Newsletter Committee___ Membership Committee ___ Program Committee

Positions include: Annual Quality Congress (AQC) Division Session Manager, AQC Short Course Chair,AQC Technical Paper Reviewers, AQC Topic Session Manager, Fall Technical Conference (FTC) TechnicalProgram Chair and FTC Short Course Chair.

___ Publications Committee

Positions include: Acquisitions Coordinator, Glossary Editor, Briefings Editor, How-To Series Editors andNew Products Coordinator.

___ Other: ___________________________________________________________________________________

___________________________________________________________________________________

ASQ Statistics Division

Job Openings and Member Interest Form


The ASQ Statistics Division Newsletterispublished quarterly by the StatisticsDivision of the American Society forQuality.

All communications regarding this pub-lication, EXCLUDING CHANGE OFADDRESS, should be addressed to:

Janice Shade, EditorASQ Statistics Division NewsletterNabisco, Inc.7 Sylvan WayP.O. Box 304Parsippany, NJ 07054-0304Phone: (973) 682-6236Fax: (973) [email protected]

Other communications relating to theStatistics Division of ASQ should beaddressed to:

Don EmerlingImation Corp.1 Imation PlaceEndeavor 2E-2OOakdale, MN 55128-3412

Communications regarding change ofaddress should be sent to ASQ at:

American Society for QualityP.O. Box 3005Milwaukee, WI 53201-3005

This will change the address for allpublications you receive from ASQincluding the newsletter. You can alsohandle this by phone (414) 272-8575 or(800) 248-1946.

Printed on Recycled Paper

STATISTICS DIVISIONAMERICAN SOCIETY FORQUALITYc/o Janice Shade7 Sylvan WayP.O. Box 304Parsippany, NJ 07054-0304

Non-Profit Org.U.S. Postage

PAIDCedarburg, WIPermit No. 199

UPCOMING NEWSLETTER

DEADLINES

Issue Vol. No. Due Date

Fall ’97 16 7 Aug. 29, 1997

Winter ’98 16 8 Nov. 14, 1997

Summer ’98 16 8 May 15, 1998

outgoing chair’s message incoming chair’s...

Documents