Treqna Base Manual

treqna.com

First Edition

1

© All Rights Reserved TreQna 2005

C ontents
DEFINE Step D1 – Map Project Step D2 – Approve Project MEASURE Step M1 – CTQ Characteristics & Standards Step M2 – Measurement System Analysis Step M3 – Data Collection Plan ANALYZE Step A1 – Baseline Process Step A2 – Performance Objective Step A3 – Identify drivers of Variation IMPROVE Improve Step I1 – Screen for Vital Xs Improve Step I2 – Study Interaction between Xs Improve Step I3 – Define Improved Process CONTROL Control Step C1 – MSA on Xs Control Step C2 – Improved Process Capability Control Step C3 – Establish Control Plan Page 4 Page 36 Page 45 Page 82 Page 116

Treqna Base Manual

Page 145 Page 191 Page 197

Page 293 Page 298 Page 303

Page 317 Page 319 Page 321

First Edition

2

© All Rights Reserved TreQna 2005

Treqna Base Manual

DEFINE MEASURE ANALYZE IMPROVE CONTROL

Map project Approve project

CTQ Characteristics and standards Measurement System Analysis Data collection

Baseline process Performance objective Identify drivers of variation

Screen for vital X’s Study interaction between X’s Defined Improve Process

MSA on X’s Improved Process capability Establish control plan

First Edition

3

© All Rights Reserved TreQna 2005

Treqna Base Manual

Define Phase
DEFINE MEASURE ANALYZE IMPROVE CONTROL Step D.2
Approve project Project feasibility CBA FMEA

Step D.1
Step Deliverables Map project Project Charter COPIS Survey, Focus Group, Interview, Charter, COPIS

Tools

First Edition

4

© All Rights Reserved TreQna 2005

Treqna Base Manual

Topics
About Projects Customer, CTQs and VOC COPIS Project Charter

First Edition

5

© All Rights Reserved TreQna 2005

Treqna Base Manual

Characteristics of a good project
Should have clearly defined scope Should have significant impact on customer Should be in line with business objectives Should have synergy with other projects (must not conflict) Should be focused on key CTQ Should generate bottom-line hard dollar benefits, Cost Avoidance or Productivity increase

First Edition

6

© All Rights Reserved TreQna 2005

Treqna Base Manual

Sources for project ideas ?
Project topics usually surface as issues in a product or process. The project ideas may also come from: Customer dashboards Customer Surveys Other related projects Internal issues The tools which can be used to identify project ideas: Brainstorming QFD Root cause analysis

First Edition

7

© All Rights Reserved TreQna 2005

Treqna Base Manual Topics About Projects Customer. CTQs and VoC COPIS Project Charter First Edition 8 © All Rights Reserved TreQna 2005 .

Buyer of the product or services First Edition 9 © All Rights Reserved TreQna 2005 .g. External customer: E.g.Treqna Base Manual Identifying the Customer Who is the customer? Internal Customer or External Customer ? Process Customer Output Input Supplier Customer receives the output of a process Internal customer: E. Marketing person for manufacturing company is an internal customer.

First Edition 10 © All Rights Reserved TreQna 2005 .Treqna Base Manual Critical to Quality (CTQ) CTQ’s from any process are translated from customer requirements External Customer Internal Customer Customer driven CTQ’s Process driven CTQ’s Customer to CTQ Identify Customer Capture customer requirement Derive and Define CTQ’s What is a CTQ? A CTQ (Critical to Quality) also known as a KPI (Key Process Indicator) is a metric that measure some aspect of a product or process which is critical to the customer. The customer defines acceptable levels for CTQs using specification limits.

Observation. Customer specifications. Field reports etc. First Edition 11 © All Rights Reserved TreQna 2005 .Treqna Base Manual Voice of the Customer (VoC) The "voice of the customer" is a process which is used to capture the requirements/feedback from the customer (internal or external). It helps us understand the customers expectations. The voice of the customer can be captured in a variety of ways: Direct discussion or interviews Surveys Focus groups Complaints Others ways like.

Treqna Base Manual VOC Tools Evaluation Survey Advantages: Low administration cost Relatively faster results in case of telephone survey Excellent approach to reach larger end customer base Least trained resources required to execute Disadvantages: Low response rate with Mail surveys Unclear understanding by customer Incomplete responses to mail surveys Potential risk of biased response due to interviewer influence Focus Groups Advantages: Help gather qualitative and in depth information Excellent for understanding and getting CTQ definitions Not prone to understanding and interpretation gaps Disadvantages: Bias due to limited participants Limited data due to qualitative focus Can have a lot of anecdotal information First Edition 12 © All Rights Reserved TreQna 2005 .

Treqna Base Manual VOC Tools Evaluation Interviews Advantages: Complex questions can be asked and a wide range of information gathered Both qualitative and quantitative data can be collected Better communication and interpretation Disadvantages: Higher cost alternate Time consuming Requires trained. experienced resources to execute Customer Complaints Advantages: Specific feedback Provides opportunity to respond appropriately to dissatisfied customer Disadvantages: Sample size issue Situation specific. might be anecdotal Prone to small sample bias First Edition 13 © All Rights Reserved TreQna 2005 .

Double negatives-They may create misunderstanding. Devise measurement scale. •Long surveys-They discourage respondent participation.Treqna Base Manual Survey Development Objectives Why is the survey needed? Who will use the survey results? What specific information is required with the help of this survey? Who needs to be surveyed? Logistics How much time is available to administer the survey? Who will conduct the survey? What is the medium to conduct the survey? What are the financial resources available to conduct the survey? Design Draft questions and validate against objectives. Types of questions: •open questions: Open questions allow the customer to respond to a question in his or her own words. Design report out structure Administration Determine sample size and sampling strategy Pilot survey on selected group (ensure the methodology) Designing survey question: Survey results will be meaningless unless you ask the right questions. Other sources of bias in survey: Question sequence Sample Bias No-response bias Unclear questions Interviewer / facilitator induced bias First Edition 14 © All Rights Reserved TreQna 2005 . •Judging questions-They can lead to guarded or partial responses. •closed questions. Avoid the following: •Leading questions-They inject interviewer bias. •Compound questions-They may generate a partial or no response. The question must contain enough specifics so the respondent can give a meaningful answer. Bias from Design of Questions: The questions you ask your customers must be properly worded in order to achieve good end results. Multiple choice. Closed questions offer the customer a choice of specific responses from which to select. Ambiguous or vague questions-They produce meaningless responses •Acronyms and jargon-These may be unknown to the respondent. rating scale and yes/no questions are examples of closed questions.

Internal Customers: Internal customers have needs associated with job security. Stated Needs: Use questions like “Why do you want…?” until you understand the need Perceived Needs: The same product or service may be perceived differently by customers Intended vs.Treqna Base Manual Understanding Customer Needs Real vs. Effectiveness vs. those relating to the amount of resources allocated or consumed in meeting customer needs. however. The internal customer. First Edition 15 © All Rights Reserved TreQna 2005 . Efficiency Needs: The external customer will generally express effectiveness needs–those relating to the value they receive from the product or service. Actual Usage: Be careful of how customers say they will use a product/service versus how they actually use it. prestige. etc. will generally express efficiency needs.

Treqna Base Manual Topics About Projects Customer. CTQs and VoC COPIS Project Charter First Edition 16 © All Rights Reserved TreQna 2005 .

First Edition 17 © All Rights Reserved TreQna 2005 .Treqna Base Manual COPIS Puts process and customer in perspective Depicts process flow Highlights process boundaries and interdependencies COPIS maps customer and process interaction by defining customer requirements and steps taken to deliver the desired output Thought Process Process Flow C Customer O Output P Process I Input S Supplier COPIS and SIPOC are used interchangeably.

Output: Anything produced for the customer (internal or external). Process: Group of activities required to transform inputs in to customer desired output Input: Material or knowledge required to produce the desired output Supplier: Source that supplies the input Metric: A measure of compliance with customer expectation or established standard First Edition 18 © All Rights Reserved TreQna 2005 . Outcome of the process.Treqna Base Manual COPIS .Components C Customer O Output P Process I Input S Supplier Customer: Recipient of the process output.

Outputs. Suppliers and Process Name C Customer O Output P Process I Input S Supplier Step2: Identify customer requirements Step3: Identify process steps First Edition 19 © All Rights Reserved TreQna 2005 . Inputs.Treqna Base Manual COPIS .Components Step1: Identify Customers.

Treqna Base Manual Topics About Projects Customer. CTQs and VoC COPIS Project Charter First Edition 20 © All Rights Reserved TreQna 2005 .

The critical elements of a project charter are: Business Case Problem and Goal Statement Project Scope and Boundaries Project Team Project Timelines Communication Plan First Edition 21 © All Rights Reserved TreQna 2005 .Treqna Base Manual Project Charter Project charter is a written roadmap that defines key questions or issues to be addressed by the project. It defines the project's purpose and its intended outcomes.

Treqna Base Manual Critical Elements Of A Charter Business Case: The need to do the project Problem and Goal Statement: Project problem and improvement goal in distinct and measurable terms Project Team: Who will be involved in the project and what role will he or she play? Project Scope: Area of influence and resources for the project Project Timelines: Project plan and key milestones Communication Plan: Communication strategy through the project First Edition 22 © All Rights Reserved TreQna 2005 .

NOT doing the project and delaying it Business case also talks about: Brief introduction to the process/business where project is undertaken How does the project help in achieving business goals.Treqna Base Manual The Business Case Project Business Case has three critical elements: It tells us why is the project worth doing It tells us why is it important to do it now It tells us the consequences of both. First Edition 23 © All Rights Reserved TreQna 2005 .

ppm. For instance.Treqna Base Manual The Problem Statement The problem statement: Represents the problem in measurable (quantifiable) terms. Measures the problem against a target (if one exists) Questions that a problem statement must answer What is the problem? Where is the problem? What is the magnitude of the problem? Over what period has the problem been recorded? What is the impact of the problem? First Edition 24 © All Rights Reserved TreQna 2005 . defect rate. yield. sigma etc.

First Edition 25 © All Rights Reserved TreQna 2005 .Treqna Base Manual The Problem Statement Examples Problem Statement: “Our customers are not happy with the delay in monthly credit card account statement mailing” Poorly defined problem statement – generic in nature. It does not highlight the magnitude or the impact of the issue at hand. Clearly brings out the magnitude of the problem and its consequences to the business.000 in late fee charges. Problem Statement: “In the last 3 months 26% of the monthly credit card account statement were delayed by 10 or more days.” Well defined problem statement – specific and direct. due to which business had to waive off $280.

this might happen in business case also) Immeasurable: Challenges in collecting data or data can not be collected (Insight into a problem is limited by measurability of a problem) Scope: Problem statement too narrowly or broadly defined First Edition 26 © All Rights Reserved TreQna 2005 .Treqna Base Manual Things To Watch Out For Potential Pitfalls: Anecdotal: The problem statement is based on assumption (guess) or based on insufficient data? Leading: Problem statement giving hint to a prejudged root cause? (implied solution.

reduce.g. First Edition 27 © All Rights Reserved TreQna 2005 . increase. It must Be time bound NOT suggest or assume a solution Have a specific.Treqna Base Manual The Goal Statement A Goal Statement defines the improvement objective that a project team sets out to achieve. measurable target The goal statement would usually start with the verb. eliminate etc. e.

Goal Statement: “Reduce the percentage of delayed monthly credit card account statement from 26% to 2% by 26th Nov 2004. of delayed credit card statements. measurable and time bound. The target and time frame to achieve it have not been defined.” Poorly defined goal statement – generic in nature. First Edition 28 © All Rights Reserved TreQna 2005 .” Well defined goal statement – is specific.Treqna Base Manual The Goal Statement Example Goal Statement: “We intend to reduce the no.

on the project team You must also define boundaries (start and stop points) for each process under study to avoid ambiguity First Edition 29 © All Rights Reserved TreQna 2005 . While defining the scope of a project. if any.Treqna Base Manual Project Scope and Boundaries Project Scope and Boundaries provide clear demarcation between what the project team can and cannot influence. you must identify Processes under study Resources available to the project team Constraints.

Treqna Base Manual Steps in Scoping a Project Identify customer expectations – Who is the customer? What is the output? What does the customer expect from the output? Translate customer expectations to process CTQs – which process CTQs best represent customer want Map process – detail the process(es) under study to identify areas of specific focus Identify resources – what are the resources available to the project team Understand constraints – what are the other CTQs or parameter that can be influenced by the process but need to remain as is First Edition 30 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Timelines & Project Plans Establishing critical milestones helps ensure focus through the life of the project. Identifies order of execution for critical tasks Identifies interdependencies. yet realistic Project Plan: An activity by activity plan to help meet project timelines. timelines should be aggressive. between tasks Gives an estimate of time requirement Helps identify potential roadblocks and plan contingency measures First Edition 31 © All Rights Reserved TreQna 2005 . if any.

Champion/Sponsor.) Other things to be covered in the plan include the choice of media. weekly. successes.. Cross-functional Teams. plan. Process Owner. Process Teams etc.Treqna Base Manual Communication Plan Communication plan: To prevent sudden surprises: Keep all stakeholders informed about project status To ensure that timely communication of vital information to those who need it To help efficient management of information logistics Good communication plan would address the following: Who would communicate? (Project Owner) What to communicate? (Project Status. failures) When and how frequently? (Tollgates. monthly…) Reason for communication? (importance) Recipients of communication? (Project Manager. challenges. information source and comments regarding details of information First Edition 32 © All Rights Reserved TreQna 2005 . Process Team. schedule. information residence.

Black belt – An employee of an organization who has been trained on Six Sigma and executes/mentors projects. Champion – A member from the leadership team who reviews the progress of projects that he/she champions.An employee of an organization who has been trained on Six Sigma and executes projects as a part of his/her full time job. execution and support of Six Sigma projects.Treqna Base Manual Team Roles Roles which should be defined in the Project Charter: Project Manager – GB/BB or MBB Mentor .BB or MBB Project Champion Project Sponsor Project team members While defining roles: The project champion and the sponsor might be the same person Team members are usually from the process or related to the product for which the project is undertaken.Black belts usually report into a Master Black Belt. Black belts are usually mentored by Master Black Belts Master Black Belt – Master Black Belts are Six Sigma Quality experts that are responsible for the strategic deployment of Six Sigma resources within an organization. Master Black Belt main responsibilities include training and mentoring BBs and GBs. Involvement of support staff is also common. selection. It is not unusual to have team members specific to a particular activity or phase Green belt . he/she provides the necessary resources for execution of the project First Edition 33 © All Rights Reserved TreQna 2005 . Black belts are Six Sigma specialists and mentor Green Belts. The champion is also responsible for supporting the project and help it overcome any challenges Sponsor – Usually the process owner of the process on which a project is being done. Green belts are usually supervised or mentored by a Black belt.

knowledge (subject matter expertise) or competence (statistical or technical skills) Champion’s role is not clearly defined First Edition 34 © All Rights Reserved TreQna 2005 .Treqna Base Manual Potential Pitfalls Common mistakes made while selecting a project team Time commitment from team members is not clearly defined Incorrect mix – lack of influence (hierarchy).

Treqna Base Manual Topics covered About Projects – – Characteristics of a good project Sources for project ideas Customer. CTQs and VoC – – – – Internal or External Customer Critical to Quality Metrics Voice of the Customer VoC tools COPIS Project Charter – – – – – – Business Case Problem and Goal Statement Project Scope and Boundaries Project Team Project Timelines Communication Plan First Edition 35 © All Rights Reserved TreQna 2005 .

COPIS Tools First Edition 36 © All Rights Reserved TreQna 2005 . Interview.1 Step Deliverables Map project Project Charter COPIS Survey.2 Approve project Project feasibility CBA FMEA Step D. Focus Group. Charter.Treqna Base Manual Define Phase DEFINE MEASURE ANALYZE IMPROVE CONTROL Step D.

Treqna Base Manual Topics Project Risks Cost Benefit Analysis Project Go/No Go decision First Edition 37 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Project Risks Brainstorm to identify potential risks associated with each phase of the project – risks that restrict its execution or cause it to fail Assess the likelihood of the risk to occur Evaluate risks associated with external factors influencing the process or product under study Identify contingency measures to mitigate identified risks. Evaluate project feasibility under light of associated risks and contingency plans First Edition 38 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Topics Project Risks Cost Benefit Analysis Project Go/No Go decision First Edition 39 © All Rights Reserved TreQna 2005 .

if possible. of people required to do the job reduce? Understand business budget constraints Are there any budget constraints? Are there other projects with equal or high priority? A Cost Benefits Analysis (CBA) is a calculation which is done in order to evaluate the costs that a project/change may incur and if it will have generate enough benefits to justify the costs.Treqna Base Manual Cost Benefit Analysis Identify costs associated with the project Would new technology or changes to existing technology be required? Is there some new infrastructure that would be required? Translate improvement target. First Edition 40 © All Rights Reserved TreQna 2005 . in to quantifiable dollar benefit Would the raw material costs come down? Would the no.

Treqna Base Manual Topics Project Risks Cost Benefit Analysis Project Go/No Go decision First Edition 41 © All Rights Reserved TreQna 2005 .

the contingency plans to mitigate those risks and the residual exposure to such risks The champion’s commitment to mitigate risks and drive contingency measures. if any require his influence Financial commitment for the completion of the project (allocate budget) First Edition 42 © All Rights Reserved TreQna 2005 .Treqna Base Manual Project Go/ No-go Evaluate the following before undertaking a project Potential risks associated with the project.

its benefits and its importance from the business stand point Project Go/No Go decision – Contract the champion’s commitment to ensure the success of the project First Edition 43 © All Rights Reserved TreQna 2005 .Treqna Base Manual Topics Covered Project Risks – Identify and plan for risks that may delay the project Cost Benefit Analysis – Evaluate costs associated with the project.

Treqna Base Manual DEFINE MEASURE ANALYZE IMPROVE CONTROL Map project Approve project CTQ Characteristics and standards Measurement System Analysis Data collection Baseline process Performance objective Identify drivers of variation Screen for vital X’s Study interaction between X’s Defined Improve Process MSA on X’s Improved Process capability Establish control plan First Edition 44 © All Rights Reserved TreQna 2005 .

1 Step CTQ characteristics and standards Identify project Y metric Establish performance standards Process Map. Test/ Retest Study.2 Measurement system analysis Identify gage error Deliverables Tools Continuous Gage R&R. Short Form Gage.3 Data collection Sampling Strategy Segmentation factors Step M.Treqna Base Manual Measure Phase DEFINE MEASURE Step M. Attribute Gage R&R First Edition 45 © All Rights Reserved TreQna 2005 . FMEA. Pareto ANALYZE IMPROVE CONTROL Step M. QFD.

Treqna Base Manual Topics Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 46 © All Rights Reserved TreQna 2005 .

steps. First Edition 47 © All Rights Reserved TreQna 2005 . Process Map is also called a flowchart.Treqna Base Manual Process mapping Process Mapping: Is a graphical representation of activities. resources and their interactions. information. Process Mapping is a first step in understanding how and why a process behaves the way it does.

methods (including criteria and various documentations used along the process). machines or equipment being used in the process (in process execution or monitoring / measurement. work environment First Edition 48 © All Rights Reserved TreQna 2005 .Treqna Base Manual Understanding a process A process is a series of logical steps that transform inputs (raw materials) in to customer defined outputs. Process Customer Output Input Supplier A process may be largely affected by one or more of the following factors: personnel who operate the processes. materials which are used as inputs (including information).

Treqna Base Manual Components of a process Process Controls ------------------. It helps understand the scope of the process and its constraints. First Edition 49 © All Rights Reserved TreQna 2005 . Process controls help ensure the process is consistent in behavior Metrics and Measurements are the means to measure conformance with requirements placed by customers on outputs and processes on inputs.Process Boundary -------------------Metrics and Measurements Process Boundary defines the process limit.

duplicate or non value add steps Identify data collection points in the process Study interaction and interdependencies between different departments (highlight hand off points) First Edition 50 © All Rights Reserved TreQna 2005 .Treqna Base Manual Why map a process? Mapping a process has many uses: Compare actual vs. assumed (designed) flow Identify redundant.

and their key requirements Identify the inputs.Treqna Base Manual How to map a process? To map a process: Understand the process boundaries and scope Observe the process and identify its steps Arrange the process steps in the order that they are followed Identify the outputs. the suppliers. the customers. and the process’s key requirements First Edition 51 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Frequently used symbols Process step Process Start / End Decision box Input to a process Off page connector Process steps connector Document Multidocument First Edition 52 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Types of process maps Three frequently used process maps are: Basic Process map: used when the process is small and simple Alternate path process map: used for complex and/or large process. Alternate paths replace decision boxes in the map. 20% Cross functional process map: used when the process has many handoffs between different departments. The percentage depiction of alternate paths makes 80% this map more informative. It is also known as deployment map. Human Resource Operations Finance First Edition 53 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Topics Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 54 © All Rights Reserved TreQna 2005 .

g. From customer VOC to project metric Big Y Better Billing Process Good Customer Service Timely Process Curt and Polite Knowledgeable Accurate Billing Less queue time A CTQ drill down tree tool assists in choosing a project metric. A ‘X’ is an independent variable that effects the ‘Y’. First Edition 55 © All Rights Reserved TreQna 2005 . A ‘Y’ is a dependent output variable. The higher the ‘Y’ in a hierarchy of ‘Y’ the bigger it is considered to be. Once established it can be used in projects to help finalize relevant project metrics. Ticket Counter at Movie theatre.Treqna Base Manual CTQ Drill down Tree E. It shows linkage between the project metric and the company goals.

Treqna Base Manual Topics Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 56 © All Rights Reserved TreQna 2005 .

It helps to understand: What customer wants when the customer requirement is not explicit or is indirect.Treqna Base Manual Quality Function Deployment Quality Function Deployment (QFD) is a structured approach to defining customer needs or requirements and translating them into specific plans to produce products to meet those needs. How important it is to the customer What the customer does not want First Edition 57 © All Rights Reserved TreQna 2005 .

First Edition 58 © All Rights Reserved TreQna 2005 . Describe the steps of QFD.Treqna Base Manual QFD -Objectives Understand Quality Function Deployment (QFD) as a tool. Analyze output from QFD.

and service functions. development. Used to identify Critical to Quality Characteristics (CTQ’s) QFD is also known as the “House of Quality” First Edition 59 © All Rights Reserved TreQna 2005 .Treqna Base Manual Definition of QFD QFD is a methodology and tool to identify and translate customer needs and wants into measurable features and requirements – converting the “what’s” in to “how’s” QFD links the needs of the customer (end user) with design. manufacturing.

Typically. The cell is left blank where no correlation exists. This is the output of the QFD exercise. 3 medium correlation. and 9 high correlation. The roof of the house is used to depict correlation between various measures When the matrix is complete. 1 represents low correlation.Treqna Base Manual QFD Matrix Correlation between measures × × × Temperature (cold) Measures that capture customers requirement Customer’s importance rating Customer’s Expectations Customer Needs Quenches thirst Should taste good Should be reasonably priced Good service 4 4 2 3 Time to prepare Ingredient cost Overall CTQ impact on customer needs Strong Positive Ingredient mix 9 36 Quantity 3 9 3 9 9 3 9 39 48 18 42 Impact a CTQ has on the customer’s expectations × Strong Negative × Positive × Negative Elements of QFD matrix: Rows: The rows of the QFD matrix represent the customer needs (VOC) Columns: The columns of the QFD matrix represent the CTQs that measure the customer’s need.3 and 9 depending on the correlation between the customer needs and the CTQs. An empty column indicates that the measure does not fulfill any need and is redundant. It is not necessary to complete every cell of the QFD matrix. The cross section matrix cells are filled with values 1. First Edition 60 © All Rights Reserved TreQna 2005 . Caution: An empty row indicates that the need is not captured by any measure. The column next to needs contains the relative importance given to needs by customer. Add the results and enter the sum for each requirement at the bottom of the matrix. 3 for moderate and 9 for strong) by the priority number (1 to 5) for each corresponding customer expectation. to prioritize the CTQs or requirement: Multiply the strength of each relationship (1 for weak. a QFD matrix would have 3050% of the cells filled.

First Edition 61 © All Rights Reserved TreQna 2005 . A third level QFD can then translate the characteristics to process output measures. A second level of QFD drill down can translate these requirements to product characteristics. Identify processes/measures which have high impact on customer CTQ’s Resolve negative correlations between measures/ processes QFD can be used in iteration to translate customer needs to process CTQs. For instance .Treqna Base Manual Analyzing the QFD matrix Look for blank rows and columnsblank row means that we don’t have any measure to satisfy customer requirement and a blank column means that we have a measure that does cater to any customer requirement.customer needs can be translated to functional requirements using a QFD.

Treqna Base Manual Topics Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 62 © All Rights Reserved TreQna 2005 .

Objectives Understanding what an FMEA does Types of FMEAs When to use Learn steps involved in making of FMEA First Edition 63 © All Rights Reserved TreQna 2005 .Treqna Base Manual FMEA .

Treqna Base Manual Failure Modes and Effects Analysis FMEA is a structured tool to: Identify failure modes for a process Estimate its effect and the risk associated with it Identify and prioritize probable causes of failure Evaluate and develop control plan to contain the risk First Edition 64 © All Rights Reserved TreQna 2005 .

First Edition 65 © All Rights Reserved TreQna 2005 . transactional or other such processes.Treqna Base Manual Types of FMEAs System FMEA: Used to analyze potential failure modes in systems during the design or concept stage. manufacturing. Process FMEA: Used to analyze assembly line. Design FMEA: Used to analyze potential failure modes in products before they are released to production.

FMEA can be used at different phases for different purpose.Treqna Base Manual When is FMEA useful? Evaluate and control potential risk associated with product. Measure – CTQ’s identification Improve – Risk analysis on solution Control – To develop process control plan First Edition 66 © All Rights Reserved TreQna 2005 . processes or products are changed Proactive tool to evaluate and contain risk associated with any process or change in the process. systems in there design phase Evaluate potential risk when existing systems.

Identify input and output of the process Understand the effect of inputs (process variables) on output of the process Rank inputs according to importance List ways in which inputs can vary and the associated failure modes and effects on outputs Assign severity. occurrence and detection to each of the items and calculate risk priority number Prioritize causes with high RPN and develop control plan to minimize risk FMEA: •Is a team exercise •Living document and should be updated regularly to evaluate risk and make control plans First Edition 67 © All Rights Reserved TreQna 2005 .Treqna Base Manual Preparing a FMEA Identify process and map the process.

Treqna Base Manual FMEA: Definition of terms Failure Mode Way in which a process can fail to meet expectation Any fluctuation which can cause break down in process Any fluctuation in process variable which can result in defect or non conformance Potential Cause A deficiency leading to a Failure Mode Any source of variation in inputs or process variables Potential Effect The impact on customer if the failure mode in not contained or avoided Customer could be internal or external. The impact could be on the business or employees also (which is internal customer) First Edition 68 © All Rights Reserved TreQna 2005 .

“What can cause process to fail or what can cause a failure to impact the customer?” First Edition 69 © All Rights Reserved TreQna 2005 .Treqna Base Manual Identifying causes Causes can be identified using: Brainstorming Cause and effect diagram Ask the question .

Treqna Base Manual Rating and RPN Severity How severe is the impact of failure mode on customer? Occurrence How frequent the cause leading to failure mode can occur? Detection What is the possibility of detecting the failure mode if it occurs? Risk Priority Number RPN signifies the amount of risk associated with failure mode RPN = Severance * Occurrence * Detection First Edition 70 © All Rights Reserved TreQna 2005 .

g. not detectable First Edition 71 © All Rights Reserved TreQna 2005 . For simplicity (recommended for GBs) one can use the following scale: Rating 1 Severity Low.g.e.g. remote likelihood Medium- Detection High.e.e.g. Low or no effect customer Medium Occurrence Low – e. effecting customer High.g. almost assured failure Low. occurrence and detection..e.e.Treqna Base Manual Rating Scale Many scales are available for rating severity.g. always detectable Medium 3 9 High.

Treqna Base Manual How does it look? List process steps or product parts List Failure Modes for each process step Failure Modes and Effects analysis (FMEA ) Rate the Severity of the Effect to the customer Rate the detection based on current controls Date S Potential Potential Potential E Process Step Causes failure mode failure effect V. Prepared By: O Current C Controls C D R S O D R E P Recommendations E C E P V. C T N T N List the Effects of each Failure Mode List the causes for each Failure Mode Rate how often a particular cause or Failure Mode occurs Calculate RPN and give recommendations for reducing RPN. Calculate new RPN post implementation of actions First Edition 72 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Topics Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 73 © All Rights Reserved TreQna 2005 .

first call resolution A Performance Standard translates the “Voice of Customer” a measurable metric.Treqna Base Manual Performance Standard A Performance Standard defines The customer want Clearly whether a process is performing well or not E. First Edition 74 © All Rights Reserved TreQna 2005 . loan approval within 24 hours.g.

Treqna Base Manual Elements .CTQ performance standard An Example CTQ Project Y Turn Around Time CTQ Definition Time from application to new account opening CTQ Unit of Measure Days Performance Standard CTQ Target 3 Days Specification Limits Not greater than 5 days Defect Definition Each new account Application which Takes more than 5 days First Edition 75 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Operational Definition A precise description that clarifies: What the CTQ is And how to measure it. Why do we need to have operational definitions? It removes ambiguity in understanding between team members It clearly defines a “standard” way to measure a CTQ Ensure that CTQ representation is correct and independent of different times and operators First Edition 76 © All Rights Reserved TreQna 2005 .

“average time to process a loan should be 3 days but it should not exceed 7 days. however Specification Limits (SL) define the acceptable tolerance in deviation from target. E.Treqna Base Manual Target. VOC .” Would translate into : Target: 3 days Upper SL: 7 days (situation where LSL is not required!) First Edition 77 © All Rights Reserved TreQna 2005 . Target and SL are defined using VOC.g. Specification limit and Defect One can understand Target as “a reference point to shoot at”.

Target Specification Limit Defect First Edition 78 © All Rights Reserved TreQna 2005 . Specification limit and Defect Defect Anything that results in customer dissatisfaction. Nonconformance to the CTQ.Treqna Base Manual Target.

always at least 1 or more. Defect (D) Anything that results in a non-conformance with set quality standards Note: All Opportunities should be independent of each other and must be of significance to the customer First Edition 79 © All Rights Reserved TreQna 2005 . Opportunity (OP) A characteristic of the unit.Treqna Base Manual Definitions Unit (U) The output of a process.

Defects per Opportunity Total Defects divided by total opportunities. First Edition 80 © All Rights Reserved TreQna 2005 . Total Opportunities Total number of units multiplied by opportunity per unit. A strict measure of quality. Defects per Million Opportunities Defects per opportunity multiplied by a million. If there is more than 1 opportunity per unit then the Defects produced may be more than the number of units.Treqna Base Manual Formulas Defects per Unit Total Defects per Unit.

Treqna Base Manual Topics Covered Selecting CTQ Characteristics Process mapping CTQ Drill down tree QFD FMEA Performance Standards First Edition 81 © All Rights Reserved TreQna 2005 .

QFD.2 Measurement system analysis Identify gage error Deliverables Tools Continuous Gage R&R. Pareto ANALYZE IMPROVE CONTROL Step M. FMEA. Short Form Gage.Treqna Base Manual Measure Phase DEFINE MEASURE Step M. Test/ Retest Study.3 Data collection Sampling Strategy Segmentation factors Step M.1 Step CTQ characteristics and standards Identify project Y metric Establish performance standards Process Map. Attribute Gage R&R First Edition 82 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Topics Measurement System Analysis .Concepts Types of MSA First Edition 83 © All Rights Reserved TreQna 2005 .

Evaluated the measurement system and established variation induced by it in process data. Stability. Linearity. Accuracy. Repeatability.Objectives Understand the terms: Precision . Project team’s consensus on measurement system And define calibration standards First Edition 84 © All Rights Reserved TreQna 2005 . Reproducibility.Treqna Base Manual Measurement System Analysis . Resolution .

First Edition 85 © All Rights Reserved TreQna 2005 . The gage study is repeated until the level of acceptance is achieved.Treqna Base Manual Measurement System & Process Variation Process variation consists of the true variation of process and the variation due to measurement system. Until the problem of measurement variation is fixed the data can not be used for analysis and project purpose. If the measurement system is not studied and calibrated: Analysis of data may give misleading results The variation problem may get fixed by fixing measurement system and the project may not be required One might disturb the process operations without realizing that the cause of variation is measurement system A Gage study tells us: •Amount of measurement error •Source of measurement error Once the measurement error is quantified using gage study the next step is to minimize to the level when it becomes acceptable. Total Variation = True Process Variation + Measurement System Variation In this step we try to reduce the variation induced by measurement system.

First Edition 86 © All Rights Reserved TreQna 2005 . it becomes critical to study variation induced by measurement system .Treqna Base Manual Process Variation Process True Variation producing parts different lengths for parts Input Process Output True Process Variation Variation Induced by measurement system Perceived total Variation Because our knowledge about a problem is limited by the data available.the data is only as good as the measurement system that produces it.

Treqna Base Manual Sources of Variation Observed Process Variation Actual Process Variation Measurement Variation Long Term Process Variation Short Term Process Variation Sample Variation Gage Variation Operator Variation (Repeatability) Accuracy (Bias) Precision (Reproducibility) Stability Linearity` First Edition 87 © All Rights Reserved TreQna 2005 .

Example – Observed difference in the length of a pencil measured by different operators using the same measuring scale.Treqna Base Manual Measurement Variation – Key concepts Accuracy / Bias – Difference between the standard and observation. Stability – Variation in measurement of the same unit measured by the same operator using the same equipment over time Example – Observed difference in the length of a pencil measured by different operators using the same measuring scale on different days. Linearity – Consistency of accuracy across the entire range of measurement system. Example – Observed difference in the length of a pencil measured by the same operator using the same measuring scale Reproducibility – Variation in measurements of the same unit measured by different operators using the same equipment. Example – Observed difference in the length of a pencil when measured using different ends of the same measuring scale.A weighing scale measuring greater than “0” in free state Repeatability – Variation in measurements of the same unit measured by the same operator using the same equipment. First Edition 88 © All Rights Reserved TreQna 2005 . Example .

First Edition 89 © All Rights Reserved TreQna 2005 . Accuracy: Minimal or no difference between an observed reading average and a standard. Ideally the gage should be able to resolve the tolerance into approximately ten levels. Resolution: Resolution as defined is the smallest unit in which the gage can measure.Treqna Base Manual Understanding Precision and Accuracy Precision: The variation caused due to gage (noise) should be less than the process variation otherwise it will be difficult to detect process variation.

Treqna Base Manual Understanding Precision and Accuracy Target Precise but away from target Precise but not accurate == At target and precise Precise and accurate At target but not precise Accurate but not precise First Edition 90 © All Rights Reserved TreQna 2005 .

Linearity–The consistency of the accuracy of gage across the entire range of the measurement system.Treqna Base Manual Measurement System Variability Method Material Measurement Machine Environment People Fishbone may be used as a tool to identify Sources of variation in the measurement system Some other terms associated with gage: Repeatability– Variation when one operator repeatedly measures the same unit with the same measuring equipment. First Edition 91 © All Rights Reserved TreQna 2005 . Stability–Variation obtained when the same person measures the same unit with the same equipment over a large gap of time. Reproducibility–Variation when two or more people measure the same unit with the same measuring equipment.

Treqna Base Manual Topics Measurement System Analysis .Concepts Types of MSA First Edition 92 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Types of MSA Attribute Gage: Done when discrete data is involved Gage ANOVA: with continuous data and the most exhaustive gage study. First Edition 93 © All Rights Reserved TreQna 2005 . Destructive Gage: Done when the event is destructive in nature and can not be repeated. Requires resource commitment and time. Continuous data can also be run through short form gage and test-retest.

Treqna Base Manual Improving the Measurement System At target but not precise NOT Precise and NOT Accurate At target but not precise Precise but NOT Accurate At target but not precise Precise and Accurate First Edition 94 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Attribute GAGE Checks gage for: Repeatability Reproducibility Accuracy First Edition 95 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Attribute Gage standards Ideally an attribute gage is treated as adequate if: Repeatability: 90% of repeated measures by an operator match Reproducibility: 90% of repeated measures across all operators match Accuracy: 90% of all individual measures by across all operators match with the standard First Edition 96 © All Rights Reserved TreQna 2005 .

there are 270 data points to calculate accuracy. For instance. Accuracy is calculated by taking each measurement as a data point. whether we have 30 units measured 3 times by 3 operators or 10 units measured 3 times by 9 operators or 10 units measured 9 times by 3 operators. it is better to measure 10 units 9 times. For instance.Treqna Base Manual Attribute Gage standards Repeatability test is more sensitive if there are greater no. First Edition 97 © All Rights Reserved TreQna 2005 . Reproducibility test is more sensitive if the same unit is measured by more operators. than to measure 30 units 3 times. Therefore. it is better to have 10 units measured by 9 operators than to have 30 units measured by 3 operators. of repeated measurements of the same unit.

If the desired resolution is 0. which may not solve the purpose. the gage is good to distinguish a difference of one percentage point only (1/100).5%. If the sample size is hundred.Treqna Base Manual Sample Size for the Attribute Gage Consider the desired resolution to determine the sample size Greater the desired resolution. higher is the sample size requirement.5% means that the gage needs to distinguish a difference of half percentage point. First Edition 98 © All Rights Reserved TreQna 2005 . will a sample size of 100 be enough? A desired resolution of 0.

Reproducibility: % times all operators match on all repeated measures for the same unit.Treqna Base Manual Attribute gage sheet TRUE Sam ple A nswer 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 N Y Y N N Y Y N N N Y Y Y N N N Y N Y Y Y N N N N Y N N N Y First Operator Trial1 N Y Y N N Y N N N N Y N Y N N N Y N Y N Y N N N N Y N N N N Trial2 N Y Y N N Y Y N N N Y N Y N N N Y N Y N Y N N N N Y N N N Y Trial3 N Y N N N Y N N N N Y N Y N N N Y N Y N Y N N N N Y N N N N Second Operator Trial1 N Y Y Y N Y Y Y N N Y Y Y N N N Y N Y Y Y Y Y N N Y N N N Y Trial2 N Y Y Y N Y N N N N Y Y Y N N N Y N Y Y Y Y Y N N Y N N N Y Trial3 N Y Y Y N Y Y Y N N Y Y Y N N N Y N Y Y Y Y Y N N Y N N N Y Third Operator Trial1 Y Y Y N N Y Y N N Y Y Y Y N Y N Y N Y Y Y N N Y N Y N N N Y Trial2 N Y Y N N Y Y N N N Y Y Y N Y N Y N Y Y Y N N Y N Y N N N Y Trial3 N Y Y N N N Y N N Y Y Y Y N Y N Y N Y Y Y N N Y N Y N N N Y Repeatability: % times each operator matches on all three measures for the same unit. there are 270 opportunities First Edition 99 © All Rights Reserved TreQna 2005 . Since there are total 270 measures. Total 90 opportunity exist in the example. Total 30 opportunities exist in the example Accuracy % times each individual measure matches with the standard.

First Edition 100 © All Rights Reserved TreQna 2005 . Test re-test is used when same operator does the measurements. It is good practice to use it before opting for complex gage ANOVA.Treqna Base Manual Test Re-test study When to use it? To check for precision of gage or measurement system. Remember that if a gage is good on precision. the problem of accuracy can be solved by calibration (correction).

For example. Tolerance is defined by the customer.Treqna Base Manual Test Re-test standard Precision: Ideally precision should be less than 10% of tolerance. +10 seconds of specifications would translate into tolerance of 20 seconds. Accuracy: Can be only measured when one knows the standard value. First Edition 101 © All Rights Reserved TreQna 2005 .

100. 104. 103. 95. 95. 100. 98. 105. 98. 94. 103 First Edition 102 © All Rights Reserved TreQna 2005 . The tolerance given is 16 mm (+8mm) and the standard diameter of ball is 98mm . 106. 96. 98. 101. 97. 96.Treqna Base Manual Test re-test gage Let us take an example where we have readings for diameter of a ball. 101. 95. 99.

04762 18 20 Number of runs about median: Expected number of runs: Longest run about median: Approx P-Value for Clustering: Approx P-Value for Mixtures: Number of runs up or down: Expected number of runs: Longest run up or down: A pprox P-Value for Trends: A pprox P-Value for O scillation: First Edition 103 © All Rights Reserved TreQna 2005 .0 97.95238 0.17906 8 10 12 Observation 14 16 16 13.82094 0.0 102.5 105.00000 3 0.00000 2 0.5 95.5 C1 100.Treqna Base Manual Test re-test gage Make Run chart to find out if the data is stable Run Chart of C1 107.0 2 4 6 13 11.

2.895 101.116 0.6 S=3.000 5. First Edition 104 © All Rights Reserved TreQna 2005 .2-98=1.000 96.500 102.622 13.754 Therefore.425 99.36 0. S=3.959324 20 94.200 3.000 98. the measurement device has inadequate precision.235 9 5 % C onfide nce I nte r v a ls Mean Median 96 97 98 99 100 101 0.505 96.2 Summary for C1 A nderson-D arling N ormality T est A -S quared P -V alue M ean S tD ev V ariance S kew ness K urtosis N M inimum 1st Q uartile M edian 3rd Q uartile M aximum 97.6 The inaccuracy or bias can be calculated as 99.Treqna Base Manual Test re-test gage Run Descriptive test to calculate mean and standard deviation Mean=99.367278 -0.000 100.500 106.290 94 96 98 100 102 104 106 95% C onfidence I nterv al for M ean 95% C onfidence Interv al for M edian 95% C onfidence Interv al for S tD ev 2.6 > 10% of 16 = 1.

This is also termed as Reproducibility variation. Appraiser Variation: The variation introduced between different operators. This is also termed as Repeatability variation. within equipment or gage. and different methods. different parts. Dev (R&R))2 = (Std. within the method. This variation comes from the parts of measurement system or process. EV and AV is (Std. (Reproducibility))2 First Edition 105 © All Rights Reserved TreQna 2005 . Dev.Treqna Base Manual EV and AV Total variation (R&R) in a measurement system is further classified as: Equipment Variation: The variation within operator. Dev (Repeatability))2 + (Std. Total variation (R&R) = EV + AV The mathematical equation used for representing the relationship between total variation.

The gage variation might cause this data to fail on the lower specification limit LSL The gage variation might cause this data to fail on the upper specification limit USL The gage variation might cause this data to pass on lower specification limit The gage variation might cause this data to pass on upper specification limit Centre circle is the true data point and the limits on left and right show the measurement variation First Edition 106 © All Rights Reserved TreQna 2005 .Treqna Base Manual How a gage plays with Spec Limits What is the challenge with having high R&R error ? Following scenario explains why..

Objectives Gage ANOVA study: Quantify amount of measurement variability Identify amount of measurement variation from different sources Data Type: Continuous First Edition 107 © All Rights Reserved TreQna 2005 .Treqna Base Manual Gage ANOVA .

0 22.0 17.Data Format The data should be arranged in columns as in the following example. The recommended numbers of operators.5 20.Treqna Base Manual Gage ANOVA .0 18.0 19.5 19.1 19. (3*10*3=90) Operator Part Trial Response 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 1 1 1 1 2 2 2 2 1 1 1 1 2 2 2 2 20.6 18.0 19.1 21. Which would give one 90 data points.0 19.9 First Edition 108 © All Rights Reserved TreQna 2005 . 10 and 3 respectively. parts and trials are 3.2 20.2 21.5 18.0 19.

First Edition 109 © All Rights Reserved TreQna 2005 .Treqna Base Manual Gage ANOVA Minitab Choose Stat> Quality Tools> Gage Study> Gage R&R Study (Crossed).

0 18.0 22.0 19.9 Select method of analysis as ANOVA First Edition 110 © All Rights Reserved TreQna 2005 .1 19.0 17.1 21.2 20.2 21.0 19.5 20.6 18.0 19.5 19.Treqna Base Manual Gage ANOVA – Minitab Operator Part Trial Response 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 1 1 1 1 2 2 2 2 1 1 1 1 2 2 2 2 20.0 19.5 18.

0353 5.627 24. operator and the interaction all are significant factors Source Total Gage R&R Repeatability Reproducibility Operator Operator*Part Part-To-Part Total Variation StdDev (SD) 7.02 16.ANOVA Method Two-Way ANOVA Table With Interaction Source Part Operator Part * Operator Repeatability Total Gage R&R Source Total Gage R&R Repeatability Reproducibility Operator Operator*Part Part-To-Part Total Variation VarComp 50.8 1162. business and the project champion. Project should be evaluated by the mentor and acceptance would depend on the process.70 100.7 MS 2038. Minimum number of distinct categories should be at least 5.192 10.212 35.283 34.012 29.00 Study Var (6 * SD) 42.9144 18.1687 4.47 20.01 2. For example if you have 20 part being evaluated and the number of distinct categories are 2. This means that the most of the parts are not different enough.8 530. However one should consult with BB/MBB if the value is between 10-15%. In the example it is 13.1093 3. 30 4.475 16.72 30.486 108.034 0.9663 P 0.27 25.411 %Contribution (of VarComp) 13.3% which is more than acceptable 10% Number of Distinct Categories = 3 Percentage contribution: It represents the percentage of variation contributed by gage in the process.4785 Look for percentage contribution for Total Gage R&R.00 DF 9 2 18 30 59 SS 18343.20 265.8474 3.37 86.5583 4.936 379.151 328.11 100.29 9. In such a case the precision of the gage is not enough for the process. No. Recommended percentage contribution from Total Gage R&R should ideally be less than 10%.084 19.871 %Study Var (%SV) 36.000 0.Treqna Base Manual Minitab Session Output Gage R&R Study .1366 19. which is addition of Repeatability and Reproducibility percentage contribution.05 says significant variation. If tolerance is beyond 15% it is recommended that the gage be corrected before repeating the gage study. and the data can be divided into two groups.000 P-value < 0.820 116.1046 4. where one might accept the gage error and gets go ahead with the project. In the example Parts.40 64.5 488.041 24.5 20525.23 93. of Distinct Categories: This number represents the number of distinct groups in the data. First Edition 111 © All Rights Reserved TreQna 2005 .65 6.28 F 31.59 16.

7 LCL=0 50 1 Weight by Operator 8 0 Xbar Chart by Operator 1 100 Sample Mean Average 75 _ UCL=89.69 _ X=80.36 100 75 _ R=4. Reported by : Tolerance: M isc: Weight by Part 100 75 50 50 0 Gage R&R Repeat Reprod Part-to-Part 1 2 3 4 5 Part 6 7 8 9 10 R Chart by Operator 16 Sample Range 1 2 3 UCL=15. which says most of the variation is due to parts In by operator graph.Treqna Base Manual Gage ANOVA – Graphical Output Components of variation graph shows percent contribution by category Gage R&R (ANOVA) for Weight G age name: Date of study : Components of Variation 100 Percent % Contribution % Study Var In By part graph. there isn't much difference between operators First Edition 112 © All Rights Reserved TreQna 2005 .01 2 3 100 75 50 1 2 3 4 2 Operator 3 Operator * Part Interaction Operator 1 2 3 50 5 6 Part 7 8 9 10 X bar graph by operator shows most of the points out of control.85 LCL=72. there are large differences between different parts.

Treqna Base Manual Destructive Testing Destructive testing is unique in the sense that the measured characteristic is different after the measurement process. Few examples of destructive events are: Crash testing Tensile strength of material Call quality evaluation for call centre (unless it is done on recorded call) Destructive Testing is a unique case since an operator can measure a part multiple times in the earlier case where we used Gage ANOVA First Edition 113 © All Rights Reserved TreQna 2005 .

However if the event is recorded than the measurement can be done multiple times. call quality check done on recorded calls in call centers. The reason it is necessary is otherwise the part to part variation will mask the measurement variation. Each operator measures unique batch.Treqna Base Manual Destructive Testing and Gage Assumptions and restrictions: In such a setup the event happens once and can not be repeated. This means that none of the part is measured by multiple operators First Edition 114 © All Rights Reserved TreQna 2005 . One must be able to assume that all parts in one batch are identical.g. E.

Treqna Base Manual Topics Covered MSA Concepts Understand Accuracy and Precision and their components. Understand sources of variation •Perceived vs. appraiser variation Why gage error is a challenge Gage standards Types of MSAs Attribute gage Test retest Gage R&R First Edition 115 © All Rights Reserved TreQna 2005 . True process variation •Equipment vs.

2 Measurement system analysis Identify gage error Deliverables Tools Continuous Gage R&R. Attribute Gage R&R First Edition 116 © All Rights Reserved TreQna 2005 .1 Objective CTQ characteristics and standards Identify project Y metric Establish performance standards Process Map. QFD.Treqna Base Manual Measure Phase DEFINE MEASURE Step M.3 Data collection Sampling Strategy Segmentation factors Step M. Pareto ANALYZE IMPROVE CONTROL Step M. FMEA. Short Form Gage. Test/ Retest Study.

Treqna Base Manual Topics Data Collection Plan Data Segmentation Sampling First Edition 117 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Data Collection Plan Following are the steps involved in data collection plan 1. Collect data and monitor •Why collect data? •What data to collect? •Establish operational definitions •Establish data collection procedures •Pilot operational definitions and procedures and refine •Establish sampling strategy •Conduct MSA and validate measurement system •Collect data •Monitor for data consistency and make final adjustments First Edition 118 © All Rights Reserved TreQna 2005 . Data Collection Goals 2. Validate measurement system 4. Data definition and procedures 3.

g.Treqna Base Manual Data Collection Plan Till now we have established: Metric for the project . One might have to collect data items like – Touch time per application. No. First Edition 119 © All Rights Reserved TreQna 2005 . of applications reworked etc.What is it that we are trying to measure? E. Also we know that operational definition needs to be piloted so that we can fix loose ends in the definition. Operational Definitions – We know how to define operational definition. Measurement system validation – Depending on the type of data and measurement system setup we know what type of gage to use. as additional data that will help us segment and analyze later. If the metric is ‘No of New Accounts each day’. day of week.

g. First Edition 120 © All Rights Reserved TreQna 2005 . Verify collected data with source. Discrete Data Source – Source/s of data. E.Treqna Base Manual Data Collection Sheet What should one capture while collecting data? Data – Project metric Y or segmentation factor X Data Type – Continuous vs. Such reasons might produce outliers in the data Monitoring Data Collection Question collectors for the understanding of operations definitions. Application. process step Data Range – Possible set of values data can take Segmentation Factors – All the factors which come out of brainstorming Time Period – For what time period data is collected One should also capture if there are any special changes in process during the time period. use sampling or ad hoc QC Ensure that the procedures are followed.

Treqna Base Manual Topics Data Collection Plan Data Segmentation Sampling First Edition 121 © All Rights Reserved TreQna 2005 .

Tools Used for Data Segmentation: Brainstorming.Members of the process and project team SME.Subject matter expert for the process can give valuable inputs First Edition 122 © All Rights Reserved TreQna 2005 .Treqna Base Manual Data Segmentation What is Data Segmentation? Segmentation involves dividing data into logical categories for analyzing data. the project manager may choose to capture the step at which the error occurred. For instance. while recording the errors made by a data entry process. the operator who made the error and so on.

First Edition 123 © All Rights Reserved TreQna 2005 . the possible segmentation factors are Work type Touch time Vintage of processor Accuracy Day of Week No of Processors These segmentation factors are logical groups that may point to be significant x’s during the analyze phase.Treqna Base Manual Data Segmentation Example: A project is being done on Accuracy of Application processing.

Treqna Base Manual Topics Data Collection Plan Data Segmentation Sampling First Edition 124 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Sampling .Objectives Understand sampling and why is it required Advantages of sampling Different sampling techniques First Edition 125 © All Rights Reserved TreQna 2005 .

the universal subset signifies the sample mean X s signifies standard deviation of a sample Sampling tries to answer few critical questions asked while doing a project: How much data is required? How and when to collect data? Sampling also reduces the cost involved in collecting data by reducing the amount of data required.Treqna Base Manual Population vs. the universal set. Symbol used is N The population mean represented by µ Standard deviation is represented by s Sample: The part or subset of a population. First Edition 126 © All Rights Reserved TreQna 2005 . Sample Population: The entire set of data.

Treqna Base Manual Sampling Sampling is the process of collecting subset or portion of data from population and predicting population characteristics Population N=1000 Sample N=50 First Edition 127 © All Rights Reserved TreQna 2005 .

Sampling is an efficient way to collect data for any kind of project or analysis.g.Treqna Base Manual Why Sample ? When to sample The cost associated with data is very high We are measuring a high volume process When not to sample When the subset of data may not be able to predict population characteristics. First Edition 128 © All Rights Reserved TreQna 2005 . If each unit of data is unique Apart from other reasons to do sampling. one of the critical reasons is the time required to collect population data.. E.

First Edition 129 © All Rights Reserved TreQna 2005 .Treqna Base Manual Representative Sample Sample is Representative: When the sample has same statistics as of the population How to guarantee a representative sample: Suitable sampling strategy based on the process Understand the nature of process Understand the characteristics of population Sample must be “representative” of the population. as the data collected otherwise would not deliver any results and might give misleading inferences about the population.

A sample that is biased will not be representative of the population A sample that is biased will lead to incorrect conclusions about the population A biased sample would have incorrect information about the population and will lead to biased conclusions.Treqna Base Manual Sampling Bias Bias occurs when systematic differences are introduced into the sample as a result of the selection process. A bias in the sample can not be eliminated with increasing the sample size. First Edition 130 © All Rights Reserved TreQna 2005 .

First Edition 131 © All Rights Reserved TreQna 2005 .g. E. sampling process on Sunday when only one person works in the process.g. E. It occurs we use the data what is available instead of making as effort to collect data. or following a pattern in collecting data which has interference with underlying process. interviewing the first person you met. Systematic sampling: Collecting data at a particular time.Treqna Base Manual Sampling Bias Types Convenience sampling: This is also called as accidental bias. pick friends and neighbors for survey. Judgmental sampling: When the data collector or surveyor collects data using his judgment or opinion.

Treqna Base Manual Sampling Bias Types Bias by Environment: This type of bias is introduced when the environmental conditions surrounding process have changed since the sample was collected. Measurement Bias: Measurement bias could arise if the operational definitions are not correct or the data collectors have interpreted the operational definition differently. Non-Response Bias: Non-response bias happens when the respondents to a survey tend to differ systematically from respondents to the survey.g. E. First Edition 132 © All Rights Reserved TreQna 2005 . When the call handle time data was collected it was tax season.

Treqna Base Manual Sampling Strategy – Random Sampling Random Sampling: Sample Production Line Samples picked up from assembly line without any order or system. This is the most commonly used method of sampling. In random sampling each unit has equal probability of being selected in the sample. Each item has equal probability of getting selected in the sample. First Edition 133 © All Rights Reserved TreQna 2005 . For practical purpose one can number the data coming out of process and generate random list in excel or on minitab and accordingly sample data. Use random sampling when information about stratification is unknown. Random sampling is a population approach and so care must be taken when the process is cyclical or if the data can be stratified. Using random sampling method avoids bias being introduced in the sampling process.

2 Production Line Assembly data stratified in to two categories and random samples collected for both Stratified random sampling is used when the population has different groups (strata). The two sample sizes for nuts sized A and B would be calculated based on production of size A and size B nuts respectively. First Edition 134 © All Rights Reserved TreQna 2005 . The size of the sample depends on population size of each the group.Treqna Base Manual Stratified Random Sampling Stratified Random Sampling Sample 1. In such situation it is necessary that both groups are represented in the sample and samples are collected from each group (it is like doing random sampling for each group). To sample from a assembly line preparing nuts of two sizes A and B respectively.1 Sample 1.g. E.

g. selecting 2 application every alternate hour for quality check. selecting every fourth call in the call centre for call barging. First Edition 135 © All Rights Reserved TreQna 2005 . When we have real time data coming in systematic sampling is done. E. Unlike random sampling the frequency of collecting data is fixed in systematic sampling. Care must be taken and process must be studied for any underlying structures.Treqna Base Manual Systematic Sampling Systematic Sampling: Sample Production Line Every fourth item selected for sample Random sampling and stratified sampling are done with historical data.

Rational sub-grouping assists us to understand the shift. We will study this in capability analysis under Analyze phase. which is the difference between the long term variability and short term variability of the process. The rational subgroups depend on the nature and type of process from where the data is being selected. In this case sample has subgroup size of three.Treqna Base Manual Rational Sub grouping Random Sampling: 3rd hour 2nd hour 1st hour Sample Production Line Three items selected with a frequency of one hour. First Edition 136 © All Rights Reserved TreQna 2005 . Rational sub-grouping is a process based sampling strategy.

if the data collected is for monitoring process one might want to sample data daily. First Edition 137 © All Rights Reserved TreQna 2005 .Treqna Base Manual Sampling Strategy . Sample more frequently for unstable process and less frequently for stable process Sample more frequently for process with short cycle time and less frequently for process with long cycle time To understand the sampling frequency one should understand the objective of data collection The most important issue to remember when considering sample frequency is the data collection objective. However if the objective is to collect data for the same process to study capability one might want to collect data for few months by sampling few data points each week or month.Tips Use following pointers for sampling strategy for a given process: It is always better to collect small sample spread over longer time period than one large sample over a shorter time period. e. The sampling frequency is driven by the objective of data collection.g.

virtually in all scenarios finite population sampling is conducted without replacement from populations that are of a finite size N. Finite Population Correction The central limit theorem and the standard errors of the mean and of the proportion are based on the premise that the samples selected are chosen with replacement. Δ error or precision required from sample in representing the population is standard deviation N is sample size The value of Zα/2 depends on confidence level we choose. the corresponding value can taken out from Z table. When sample is more than 5% of population size we use Finite population correction factor which is: The sample size is than: N −n N (finite) = n(1 + n/N) FPC = N −1 First Edition 138 © All Rights Reserved TreQna 2005 .Continuous Data Sample represents population characteristics.Treqna Base Manual Sample Size Calc . However.96 n = ⎜ Δ ⎝ σ ⎞ ⎟ ⎠ 2 Where. hence: Solving for n we get: n = ⎛ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎝ ⎞ ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ ⎠ Δ = Z α /2 σ n Z α /2 ∗ σ 2 Δ The equation for 95% confidence becomes (using Z table): x ⎛ 1. We can use confidence interval formula and solve it for n to get sample size: σ n Standard error CI X X ± Z α/ 2 * As standard error is also represented by delta.

Treqna Base Manual Estimating Standard Deviation Use control limits from an existing X chart = (UCL – X) / 3 or (X – LCL) / 3 Collect some data from population and calculate std deviation (recommended – 30 data points) Get realistic estimate from business (based on past experience) First Edition 139 © All Rights Reserved TreQna 2005 .

if the business measures cycle time in days. do not set Δ in minutes or hours Collect some data from population and calculate Δ using the sample First Edition 140 © All Rights Reserved TreQna 2005 . For instance. It may be based on the resolution of measurement.Treqna Base Manual Estimate Δ Take estimate from business.

This uses the confidence interval formula for attribute (discrete) data: P ± Z P α /2 ⎛ ⎜ ⎝ 1− P n ⎞ ⎟ ⎠ Again if the sample is more than 5% of population finite population correction should be used First Edition 141 © All Rights Reserved TreQna 2005 . Za/2 is 1.Treqna Base Manual Sample Size Calc .Discrete Data The formula to calculate sample for proportions data is: ⎛ ⎜ = ⎜ ⎜ ⎝ n Z Δ α /2 ⎞ ⎟ ⎟ ⎟ ⎠ 2 P ⎛ ⎜ ⎝ 1− P ⎞ ⎟ ⎠ Z is the Z score (normal data) and P represents proportionate defective and a represents confidence required from sample.96 for 95% confidence.

Treqna Base Manual Estimating p and Δ Estimate p Calculate proportionate defective using small sample Use historical control charts to estimate Use subject matter expertise to estimate Estimate Δ Use subject matter expertise to estimate Estimate using formula: Δ = Z P B α /2 (1 − P n B ) First Edition 142 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Topics Covered Data Collection Plan Data Segmentation What is segmentation? Why segment? Sampling What is sampling? Why sample? Sampling challenges – Types of Bias Types of Sampling First Edition 143 © All Rights Reserved TreQna 2005 .

Treqna Base Manual DEFINE MEASURE ANALYZE IMPROVE CONTROL Map project Approve project CTQ Characteristics and standards Measurement System Analysis Data collection Baseline process Performance objective Identify drivers of variation Screen for vital Xs Study interaction between Xs Define Improved Process MSA on Xs Improved Process capability Establish control plan First Edition 144 © All Rights Reserved TreQna 2005 .

1 Objective Deliverables Baseline Process Process capability for project Y -Study data for shape. Chi Square–Test of Independence. run chart. Homogeneity of Variance. Common cause variation.3 Performance Objective Improvement goal for project Y -Concept of Benchmarking -Benchmarking Types -Benchmarking as process Identify drivers of variation List all statistically significant X’s -Root Cause Analysis (RCA) -Process map analysis -Graphical analysis -Statistical Analysis Activities Benchmarking 1-Sample T-test. Chi Square–Test for Goodness of Fit. capability pack Tools ANALYZE Step A. Cause and Effect diagram. Simple Linear Regression.2 IMPROVE CONTROL Step A. Long Term sigma -Understanding shift. Correlation/Scatter Diagrams. stability and normality -Capability analysis -Short Term vs. Oneway ANOVA. special cause variation Descriptive test. 2-Sample T-test.Treqna Base Manual Analyze Phase DEFINE MEASURE Step A. Mood’s Median. Pareto. VA/NVA analysis First Edition 145 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Baseline Process . Describe the process by its • • • • • Descriptive statistics Nature of distribution Understand Specification Limits and Centering or Target Values Calculate: probability of a defect and process capability Use Minitab for these purposes First Edition 146 © All Rights Reserved TreQna 2005 . shape • Statistically indicate the nature of the problem • Calculated the baseline capability of the process.Objectives • Study data for stability.

Shape and Normality Stability Tool Used: Run Chart Stability is lack of excessive variation. Normality Test Check if the data follows a normal distributions Shape Tool Used: Descriptive Stats. mixtures or oscillations. There should be no trends.Treqna Base Manual Stability. clusters. Normality Tool Used: Descriptive test. Histogram Study distribution characteristics First Edition 147 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Stability . run chart can also give graphical representation of pre and post process changes comparison as it is plotted on time scale. clusters. If the data is sorted or it does not follow the time order in any other way. STATS > QUALITY TOOLS > RUN CHART Important thing to remember is that data should be in a time order before a run chart is plotted. First Edition 148 © All Rights Reserved TreQna 2005 . The only difference between run chart and control chart comes from the control limits. Indicates any special cause variation in the process.Tests for if there are any trends.Run Chart Tool Utility: To track process over a period of time for stability. Run chart studies the stability of process over a period of time by studying any abnormality is data like trends. clusters and so on. How: Minitab. run chart loses its significance. oscillations in the process. Run chart can be plotted for: Individual values Means or Medians of subgroups (if subgroups are present) Apart from looking for process stability over a period of time.

Treqna Base Manual Run Chart .mtw First Edition 149 © All Rights Reserved TreQna 2005 .Minitab MINITAB FILE: RunChart.

Oscillations: Ho: No more runs observed than expected Ha: More runs observed than expected Oscillation occurs when the data fluctuates up and down rapidly. Run chart is used for stability and helps in identifying if there is any special cause in the process and when does it appear on time scale.` First Edition 150 © All Rights Reserved TreQna 2005 . or two processes operating at different levels. indicating that the process is not steady Clusters: Ho: No fewer runs observed than expected. (Trends and Oscillations) P-values for the following special cases should be read from run chart out put: Trend: Ho: No fewer runs observed than expected Ha: Fewer runs observed than expected A trend is a sustained drift in the data.Treqna Base Manual Run Chart . Run Chart provides two tests for randomness: Based on the number of runs about the median (Mixtures and Clusters) Based on the number of runs up or down. Ha: More runs observed than expected Mixture can be identified by absence of points near the center line. such as measurement problems or sampling from a bad group of parts. Might indicate if the process is about to go out of control or is out of control.Subgroup Size 1 • Choose ‘C2’ for Data • Set Subgroup Size to ‘1’ • Click Ok The graph you see in the output window of Minitab is the run chart. either up or down. Mixtures: Ho: No more runs observed than expected. Ha: Fewer runs observed than expected Clusters are group to group variability which may indicate variation due to special causes. Below the graph you would see hypothesis tests for different measures. Mixtures often indicate combined data from two populations.

special causes present First Edition 151 © All Rights Reserved TreQna 2005 . special causes not present Ha: Data is not random.Treqna Base Manual Studying Stability Ho: Data is random.

Treqna Base Manual Studying Stability First Edition 152 © All Rights Reserved TreQna 2005 .

It can be used for run chart as well. In the input window you can mention subgroup size or column representing subgrouping in the data. The run chart would than plot mean or median depending on your choice.Treqna Base Manual Run Chart – Subgroup Size more than 1 If you have done rational subgrouping in your data. It also plots the variation within the subgroup apart from the stability over a period of time. First Edition 153 © All Rights Reserved TreQna 2005 .

mtw Descriptive stats helps studying the shape of data distribution.Treqna Base Manual Shape . measures of central tendency and dispersion.Descriptive Stats Click on Stat > Basic Statistics > Display descriptive Statistics MINITAB FILE: DescriptiveStats. It also helps to find out what kind of issue we have with the process: whether it is issue with centering or issue with variation ? First Edition 154 © All Rights Reserved TreQna 2005 .

000 45.460 19.424 Variable Maximum Agent A 71.535 8.Treqna Base Manual Tabular Output 1.Double click on “C1 Agent A” 2.000 0.000 56.000 First Edition 155 © All Rights Reserved TreQna 2005 .Click Ok Results for: DescriptiveStats.MTW Descriptive Statistics: Agent A Variable N N* Mean SE Mean StDev Minimum Q1 Median Q3 Agent A 250 0 50.750 51.

variation(spread) and there graphical representation. Following are the key statistic output from graphical summary. First Edition 156 © All Rights Reserved TreQna 2005 .Treqna Base Manual Graphical Output Click on Stat > Basic Statistics > Graphical Summary MINITAB FILE: DescriptiveStats. Double click on “C1 Agent A” 2. Click Ok Graphical summary from minitab lists all the basic statistics of the data which includes measures of central tendency. Normality P value Mean Median Standard Deviation Variance Kurtosis (measure of skew ness of data) Quartiles In graphical representation: Histogram (with o without normal curve) Box plot and Dot plot In the following pages we would cover few basis statistic tools.mtw 1.

defined as Q1/Q3 Skewed First Edition 157 © All Rights Reserved TreQna 2005 .Treqna Base Manual Different Shapes Central Tendency: Mean Normal Dispersion: Standard Deviation Central Tendency: Q1 or Q3 Dispersion: Stability Factor.

First Edition 158 © All Rights Reserved TreQna 2005 .Treqna Base Manual Different Shapes Central Tendency: Median Dispersion: Range or Span Long Tailed Bi-modal Such situation suggests data coming from two processes and not one. separate the data by process and than analyze. One should go back and have a look at the processes.

Treqna Base Manual Normality One can test normality of data using: • Descriptive test • Normality test Hypothesis used for the test is: H0: data follow a normal distribution H1: data do not follow a normal distribution First Edition 159 © All Rights Reserved TreQna 2005 .

1 0 Standard Deviations 1 2 3 The probability of a randomly selected value of being within Deviations from the mean is 95.46% +2 - Standard First Edition 160 © All Rights Reserved TreQna 2005 .26% 95.46% 99.2 .3 .Treqna Base Manual Area under a Normal Curve 68.73% .

Value For any value of X in a distribution there is an equivalent Z – value. Note : Z Value and Z capability are not the same. Statistically. First Edition 161 © All Rights Reserved TreQna 2005 .The Z – value is important as it allows us to compare to dissimilar distributions by using standard deviation as the unit of measure.Treqna Base Manual Z . Z = Xσ μ z μ 1 σ 2 σ 3 σ x The Z-value is used to transform any normal distribution in terms of the standard normal distribution with a Mean of 0 and Standard Deviation of 1. The Z – value is a measure of the number of Standard Deviations that will fit between X and the Mean.

10 4 = =2 2 2 = 10 . First Edition 162 © All Rights Reserved TreQna 2005 . The total area under any distribution is always 1.Value Χ LSL LSL USL USL ZLSL ZUSL Standard Deviations4 2 Units of Measure -3 4 -2 6 -1 8 0 10 1 12 2 14 3 16 18 4 X = 10 s = 2 ZUSL = USL .4 6 = =3 2 2 ZLSL = A Simple Z Capability Calculation For this calculation assume USL = 14 LSL = 4 Nominal = 10 Note : Nominal is also called Target Value How the calculation changes for LSL and USL. The Shaded Region represents the probability of a value being outside specification limits. this is done to ensure that the resulting number is always positive.Treqna Base Manual Z .X s X .LSL s = 14 .

your Z – Bench will be short term capability. then subtract Z shift to obtain Z long term.5 If it is not mentioned whether the data used is for sort or long term.Treqna Base Manual Calculating Z – Bench value P(d) LSL P(d) USL LSL X T USL P(d) Total = P(d) LSL + P(d) USL ZLT = from Z table ZST = ZLT + Z shift The Total Probability of a defect is calculated by adding the probability of defects that will occur at values lesser than the LSL and greater than USL. this yield can directly be converted to Sigma by using a sigma conversion table. always assume it to be long term. this is ZBench and it represent long term process capability. just add Z Shift to the long term capability. Note : Z Shift is empirically set at 1. Using the Z table. These values can be obtained by using the Z capability values you calculated earlier for both ZLSL and ZUSL.. then find the Z-value that corresponds to this total probability of defect. 1 – Z bench is the probability of a conformance also called yield. First Edition 163 © All Rights Reserved TreQna 2005 . If the data is for short term. To calculate process capability for the short term.

70 1.91e-007 0.27e-003 3.21e-007 5.30 3.42e-004 1.67e-005 1.32e-006 1.51e-006 9.90e-001 3.98e-003 2.97e-001 3.89e-004 5.26e-004 8.25e-005 2.72e-003 1.02e-001 2.64e-001 1.02e-003 2.43e-002 1.20 2.07e-005 1.Treqna Base Manual Single-Tail Z Table A (Values of Z from 0.70 3.25e-001 1.55e-007 0.00 5.89e-003 2.74e-005 1.72e-001 4.58e-007 3.19e-003 1.32e-002 1.48e-001 2.34e-001 2.07e-003 2.09e-001 9.26e-003 1.52e-001 1.08e-003 3.16e-005 5.49e-002 6.42e-001 2.92e-001 1.21e-003 4.33e-005 8.50 1.66e-003 3.10 3.50 2.53e-005 5.62e-002 2.68e-003 2.66e-003 6.59e-004 1.30 2.02e-002 1.92e-001 4.20e-005 6.77e-006 6.64e-001 4.28e-002 1.02e-002 7.33e-001 1.74e-001 2.84e-002 3.80 1.34e-006 5.70 0.20 1.12e-006 6.27e-001 1.09e-001 2.93e-006 3.36e-002 3.98e-003 6.24e-006 2.22e-005 3.10e-006 2.96e-001 4.00e-004 1.60 0.43e-006 8.22e-002 1.41e-006 3.29e-002 9.21e-001 3.36e-002 1.91e-005 1.83e-004 3.37e-004 2.70 4.80e-003 2.61e-001 1.55e-003 5.45e-001 1.27e-001 1.90 3.69e-002 7.15e-001 1.50e-002 1.11e-007 0.74e-002 1.58e-002 1.39e-002 1.59e-001 1.79e-005 1.57e-004 3.12e-005 7.67e-002 2.05e-002 4.25e-005 1.33e-002 1.71e-001 1.64e-003 1.21e-002 5.06e-002 4.92e-002 3.49e-001 1.64e-002 6.02 4.39e-003 7.64e-003 1.17e-002 1.36e-004 9.31e-001 1.47e-003 2.95e-003 5.48e-003 1.21e-001 1.49e-007 3.11e-003 7.08e-002 5.12e-004 7.50e-002 1.26e-002 4.22e-005 7.80 3.70 2.30 0.36e-003 2.19e-001 1.72e-004 1.31e-004 8.31e-003 9.68e-007 5.60e-004 1.18e-006 7.09e-002 3.91e-003 2.39e-001 2.94e-002 2.38e-004 3.05e-003 1.17e-001 3.87e-003 1.33e-007 0.44e-002 1.67e-005 3.36e-001 1.28e-005 8.84e-001 2.60 1.91e-005 3.30e-006 7.18e-002 7.56e-001 1.24e-006 7.53e-003 3.83e-002 1.82e-002 4.56e-006 1.04 4.07e-003 7.16e-004 1.02e-006 6.07e-006 6.81e-006 1.10e-002 8.68e-001 2.81e-005 1.00 4.41e-005 4.22e-003 8.42e-004 9.52e-001 3.03 4.30 4.14e-003 8.85e-002 7.43e-005 2.17e-005 2.45e-004 5.73e-006 2.25e-004 2.67e-006 3.56e-005 1.77e-004 4.00e-003 7.24e-004 1.76e-004 2.14e-002 2.76e-003 5.53e-004 1.09e-001 1.39e-003 1.54e-006 5.87e-002 2.55e-002 3.33e-003 1.75e-003 1.08 4.35e-005 1.19e-002 9.48e-001 3.79e-003 2.24e-001 1.33e-004 1.19e-004 2.12e-002 1.14e-003 6.08e-004 7.48e-002 4.11e-004 5.85e-004 1.98e-001 1.35e-002 6.69e-001 1.37e-001 3.40 2.61e-005 3.70e-004 1.05e-001 2.07e-001 9.22e-002 9.45e-001 2.41e-001 3.30 1.67e-001 3.70e-002 1.76e-001 4.68e-002 5.52e-001 4.57e-003 4.07e-005 2.90 2.23e-001 1.98e-005 1.00e-001 4.15e-005 4.83e-006 1.92e-006 1.00 to 4.67e-005 4.50 3.91e-001 2.56e-002 2.18e-002 5.18e-001 1.78e-001 2.81e-002 5.00 2.87e-003 4.29e-001 1.09 4.88e-001 4.07e-002 2.04e-003 4.60 2.81e-001 1.81e-003 1.12e-006 4.39e-002 1.17e-001 9.10 2.29e-002 2.94e-003 3.90e-003 7.56e-001 4.60e-001 4.11e-001 9.64e-003 7.35e-004 6.16e-005 1.18e-003 8.54e-003 1.46e-006 4.23e-003 3.09e-006 1.45e-005 2.38e-002 6.43e-002 5.82e-001 3.68e-002 2.50e-005 5.52e-007 0.95e-002 4.30e-002 5.14e-003 5.51e-004 1.50e-006 4.75e-002 3.93e-006 5.13e-002 8.78e-004 1.02e-004 2.80 4.42e-001 1.76e-003 5.00e-001 8.26e-003 9.71e-006 2.20 3.12e-003 1.71e-002 4.65e-004 1.05 4.36e-001 3.36e-004 5.23e-001 2.03e-001 1.34e-007 5.16e-001 2.42e-003 6.65e-002 3.93e-004 1.91e-006 2.37e-002 4.29e-006 2.68e-004 6.16e-002 4.40 3.54e-001 1.84e-001 1.38e-001 1.81e-001 2.93e-007 4.90e-004 2.62e-004 2.87e-001 1.06 4.53e-002 7.30e-007 3.22e-002 2.78e-002 6.50 0.06e-001 1.07e-002 8.19e-004 4.91e-005 2.66e-001 1.99e-003 1.59e-002 4.13e-004 2.39e-005 8.55e-002 5.45e-005 1.04e-004 6.75e-005 2.68e-001 4.01e-002 3.66e-005 1.50 4.60 4.70e-003 4.12e-001 2.57e-005 6.04e-002 7.80e-006 4.12e-001 1.50e-004 3.35e-007 0.44e-005 3.23e-005 4.89e-003 6.04e-004 6.80e-004 1.09e-001 3.51e-001 2.56e-003 1.49e-004 2.00 1.84e-005 5.08e-004 1.81e-006 4.44e-001 4.40 0.94e-002 4.40e-003 3.11e-006 1.64e-001 2.10 1.60 3.97e-002 1.89e-001 1.01e-002 7.98e-001 2.84e-001 4.04e-005 1.20 0.51e-002 7.01e-001 3.79e-001 1.04e-003 7.96e-005 6.74e-002 2.07e-002 1.21e-006 3.34e-004 3.44e-003 1.75e-002 3.94e-001 3.62e-004 5.20 4.52e-005 9.46e-002 1.66e-006 1.44e-002 2.30e-001 2.40 4.71e-007 0.16e-006 5.54e-002 1.59e-002 2.35e-003 9.47e-001 1.50e-006 2.56e-001 3.47e-004 9.21e-001 1.59e-005 1.80 0.71e-001 3.66e-004 3.19e-001 2.45e-001 3.20e-003 6.80e-003 3.84e-005 5.63e-001 3.98e-004 4.33e-001 2.79e-002 1.18e-007 4.08e-002 6.17e-005 7.39e-003 4.71e-001 2.76e-007 5.26e-003 2.81e-002 2.04e-004 2.95e-001 2.78e-001 3.55e-001 2.34e-002 7.39e-003 4.55e-007 4.74e-004 6.17e-004 7.90 1.59e-005 2.40e-006 2.80e-001 4.10 0.27e-002 3.68e-002 8.16e-002 8.58e-006 9.90 4.68e-006 1.62e-002 1.99) Z 0.21e-004 8.80 2.81e-005 3.58e-001 2.49e-003 1.26e-001 2.74e-001 3.57e-003 2.74e-006 1.41e-004 4.61e-001 2.86e-001 3.40e-001 1.91e-004 2.48e-001 4.59e-003 1.92e-002 1.02e-001 8.90 0.37e-006 8.44e-006 1.25e-001 3.88e-001 2.87e-007 3.64e-004 4.18e-007 0.33e-001 3.83e-007 4.04e-007 3.23e-002 6.30e-005 2.40 1.06e-001 8.01e-002 2.05e-001 3.17e-006 3.93e-002 6.04e-001 8.46e-005 9.54e-003 4.94e-006 3.93e-003 1.02e-007 First Edition 164 © All Rights Reserved TreQna 2005 .13e-001 3.40e-003 1.94e-002 5.13e-001 9.30e-001 2.56e-006 2.01 4.95e-005 4.85e-002 3.79e-007 0.85e-002 8.00 0.01e-006 1.95e-001 1.00e-001 1.69e-003 1.51e-002 2.07e-005 6.66e-002 1.17e-003 2.59e-001 3.76e-001 1.74e-001 1.15e-001 9.34e-003 5.10 4.36e-001 2.95e-006 1.36e-002 2.17e-007 3.22e-006 1.01e-004 3.88e-002 1.01e-005 3.46e-002 3.19e-004 3.87e-004 4.29e-001 3.00 3.89e-005 1.25e-002 9.18e-002 3.16e-004 5.40e-001 4.07 4.15e-003 3.02e-005 6.

02e-020 3.73e-012 4.98e-016 1.43e-009 7.61e-012 2.33e-019 9.61e-009 1.31e-021 3.39e-021 2.74e-016 2.48e-014 6.22e-023 3.06e-011 1.13e-017 4.04e-017 1.86e-022 2.76e-020 1.51e-015 6.63e-021 1.12e-022 4.95e-015 3.99e-012 1.02e-014 9.60e-016 3.79e-007 1.27e-021 4.23e-007 7.07e-013 5.68e-011 8.81e-011 3.02e-009 2.37e-020 9.08 1.31e-010 6.14e-017 1.99e-007 1.23e-019 2.27e-018 5.17e-014 5.21e-009 2.07 1.20e-014 4.48e-022 2.Treqna Base Manual Single-Tail Z Table A (Values of Z from 5.47e-011 7.65e-010 1.02 2.29e-022 1.97e-015 2.70e-018 3.15e-013 5.17e-018 2.75e-016 2.60e-008 9.33e-008 1.20 9.30e-021 1.98e-018 8.59e-011 2.59e-010 8.30e-023 8.47e-021 3.71e-018 2.09e-014 4.45e-007 1.70 6.01e-010 5.65e-017 1.80e-015 3.12e-020 4.14e-008 6.38e-015 2.25e-012 2.24e-010 3.26e-009 6.15e-018 2.70e-023 0.58e-020 3.73e-009 2.82e-021 2.34e-014 3.34e-013 2.40e-020 3.71e-010 4.86e-021 7.53e-014 4.36e-023 3.05e-017 8.25e-008 1.19e-009 6.64e-018 2.10e-007 1.76e-019 4.01e-010 1.85e-020 1.61e-017 3.50 8.12e-009 6.71e-013 3.87e-010 2.10 9.02e-016 4.13e-008 1.13e-010 2.69e-021 6.49e-010 7.54e-023 0.28e-011 3.92e-015 1.48e-021 9.96e-009 2.53e-016 1.43e-017 1.17e-007 6.70e-007 9.40 7.05e-007 6.73e-017 7.35e-018 1.05e-021 7.08e-015 9.24e-011 2.74e-009 2.30e-007 7.35e-020 5.49e-020 5.33e-014 2.28e-012 6.46e-009 1.66e-018 6.60 9.60 5.02e-021 2.10e-023 1.90e-011 2.88e-023 0.03e-008 4.98e-018 3.82e-009 9.18e-009 1.20 5.15e-020 2.02e-011 2.74e-014 1.41e-013 1.45e-007 8.19e-012 5.40e-008 2.00e-019 1.67e-014 2.14e-016 1.20e-010 4.54e-014 1.35e-011 1.49e-013 1.44e-014 1.95e-019 7.72e-008 2.22e-016 2.01e-008 5.08e-013 9.84e-019 2.00e-014 4.20 6.40 5.79e-012 1.60e-012 1.81e-010 3.96e-014 1.80e-013 8.62e-023 1.42e-012 1.92e-017 3.32e-024 0.01e-014 1.32e-009 1.81e-010 9.95e-020 1.51e-022 5.30 5.01e-015 4.23e-017 5.14e-022 2.19e-014 1.72e-014 1.01 2.95e-020 2.00 9.98e-010 2.90 6.50 9.54e-021 5.35e-018 1.82e-016 7.00 2.35e-019 5.20 8.43e-023 1.72e-010 4.29e-011 4.70e-012 8.11e-016 4.14e-009 3.05 2.27e-008 7.68e-016 7.90e-012 3.69e-011 1.23e-022 8.90 7.13e-019 4.81e-016 1.67e-013 7.24e-013 3.90e-008 1.87e-010 5.24e-019 4.01e-009 5.72e-007 1.88e-017 1.12e-008 3.85e-011 4.01e-013 1.60e-015 3.52e-011 1.12e-020 1.69e-008 9.35e-008 7.29e-021 2.32e-018 3.60e-008 4.19e-011 6.79e-020 7.81e-023 1.10e-012 1.80e-013 2.37e-007 8.79e-019 1.70 5.89e-007 1.52e-008 2.45e-011 4.48e-012 7.32e-016 3.95e-008 5.37e-012 6.82e-008 3.02e-022 7.30 8.19e-015 5.60e-013 1.00 7.40 6.09e-012 4.99e-021 1.63e-020 6.40 8.21e-010 1.82e-015 2.40e-010 3.60 6.52e-018 6.17e-018 4.69e-010 8.86e-023 2.79e-017 2.59e-015 2.91e-008 2.54e-016 6.35e-009 3.95e-013 4.61e-007 9.00 8.55e-013 7.00e-022 1.73e-014 7.43e-008 8.08e-010 5.53e-007 8.66e-022 3.34e-017 5.21e-017 2.03e-013 1.48e-018 3.69e-012 1.25e-015 1.70 8.55e-012 2.96e-020 7.99e-022 1.16e-018 8.53e-022 3.18e-014 1.70 9.59e-020 1.28e-010 4.46e-017 6.89e-022 2.66e-019 1.35e-015 1.15e-008 1.58e-007 1.80 9.40e-010 7.35e-022 2.00 6.29e-015 5.10 7.33e-016 1.78e-019 7.63e-022 1.38e-021 1.92e-011 9.23e-020 4.77e-009 1.38e-008 1.92e-010 1.59e-011 2.19e-008 2.00 to 9.11e-012 5.46e-022 9.94e-009 1.13e-019 8.46e-009 2.88e-017 7.51e-016 1.90 0.09 1.49e-016 1.52e-020 2.99e-009 3.68e-010 2.21e-012 1.26e-021 8.39e-010 2.12e-011 5.55e-009 5.25e-015 1.10 5.11e-007 6.71e-021 3.08e-023 0.60 8.83e-012 8.79e-023 1.17e-019 3.02e-022 3.41e-017 1.48e-019 3.43e-024 First Edition 165 © All Rights Reserved TreQna 2005 .75e-013 1.16e-020 8.38e-019 1.24e-016 1.37e-011 6.65e-008 2.33e-009 2.86e-014 2.21e-013 3.01e-010 3.51e-008 8.34e-013 6.77e-015 7.34e-009 7.20e-011 1.90 9.49e-010 1.44e-013 6.81e-018 7.16e-021 4.10e-015 4.87e-014 8.60e-014 7.75e-016 1.05e-010 1.96e-008 5.61e-009 8.28e-011 6.82e-010 1.67e-013 2.65e-009 3.63e-023 2.31e-009 1.52e-009 1.77e-011 4.46e-008 3.61e-013 4.88e-011 1.96e-012 9.96e-012 1.57e-023 2.92e-012 3.80e-011 9.37e-014 6.12e-020 1.38e-011 3.65e-018 1.44e-012 3.01e-008 1.26e-019 2.01e-012 2.40e-013 2.52e-008 1.54e-023 9.24e-013 5.48e-008 4.36e-015 3.15e-010 5.20 7.16e-008 2.29e-011 2.21e-023 2.66e-017 2.87e-016 2.42e-020 1.99e-018 1.83e-020 3.24e-022 4.00 5.30 6.52e-011 1.94e-008 2.80 6.03e-023 0.87e-007 1.00e-021 1.39e-015 6.58e-012 7.07e-008 5.05e-009 5.50 6.82e-021 1.40e-022 1.03e-019 4.26e-023 0.55e-019 1.19e-010 1.64e-015 1.50 7.79e-019 1.57e-018 1.80 7.80 8.41e-010 1.70 7.88e-012 2.04 2.50e-009 4.06e-019 1.04e-011 5.26e-012 1.12e-009 1.64e-010 1.31e-014 2.43e-014 2.02e-009 4.76e-011 1.27e-014 5.10e-015 1.03e-014 2.36e-018 9.33e-010 1.92e-015 8.03e-017 4.98e-008 1.44e-008 5.16e-017 2.51e-009 8.15e-016 9.93e-016 3.48e-008 3.66e-022 6.80 5.60 7.48e-012 4.36e-017 4.71e-022 1.23e-017 9.98e-016 8.30e-015 4.33e-016 4.34e-019 1.23e-012 2.39e-012 3.91e-014 4.37e-022 5.57e-011 7.08e-011 1.40 9.42e-016 6.73e-021 1.33e-007 1.50 5.33e-013 4.29e-016 2.18e-023 1.64e-015 7.93e-019 3.30 9.03 2.75e-013 3.65e-010 3.31e-016 5.29e-015 2.82e-008 1.74e-018 1.20e-008 6.59e-017 6.20e-016 5.66e-008 1.35e-014 1.30 7.14e-023 0.20e-008 4.81e-014 3.24e-013 1.94e-013 9.06 2.07e-018 4.05e-021 4.79e-008 3.39e-023 0.79e-008 1.43e-020 2.40e-021 5.07e-018 1.23e-010 6.10 8.99) Z 5.44e-015 1.44e-012 1.02e-013 2.93e-009 1.70e-017 2.48e-019 5.85e-022 1.03e-012 5.80e-013 1.63e-015 1.28e-017 3.62e-019 6.72e-019 2.39e-018 5.90 8.52e-020 1.21e-007 1.97e-011 3.83e-022 6.57e-009 4.24e-013 1.10 6.29e-011 1.86e-015 1.71e-017 1.91e-014 3.40e-017 1.81e-018 1.05e-009 1.71e-009 9.73e-009 3.30e-010 2.10e-023 1.

Treqna Base Manual Standard Normal Distribution Table B Z This Is Known As The Standard Normal Distribution Which Has A Mean Of Zero (= 0) And A Standard Deviation of One (s = 1) First Edition 166 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Standard Normal Distribution Table B Z This Is Known As The Standard Normal Distribution Which Has A Mean Of Zero (= 0) And A Standard Deviation of One (s = 1) First Edition 167 © All Rights Reserved TreQna 2005 .

Note : Z Shift is empirically set at 1. To calculate process capability for the short term. 1 – Z bench is the probability of a conformance also called yield. always assume it to be long term. If the data is for short term. just add Z Shift to the long term capability. then find the Z-value that corresponds to this total probability of defect. Using the Z table.Objectives • • • • Calculate Process Capability for Discrete Data Calculate Process Capability for Continuous Data Understand rational subgroups and use it for sampling and data analysis Perform the following calculations: • within.5 If it is not mentioned whether the data used is for sort or long term. between and total sum of squares • the long and short-term standard deviation and Z-values • • understand the difference between short-term process capability and long-term process capability establish whether the process has control issue or requires technology improvement The Total Probability of a defect is calculated by adding the probability of defects that will occur at values lesser than the LSL and greater than USL. These values can be obtained by using the Z capability values you calculated earlier for both ZLSL and ZUSL. this is ZBench and it represent long term process capability. then subtract Z shift to obtain Z long term. your Z – Bench will be short term capability.Treqna Base Manual Process Capability . First Edition 168 © All Rights Reserved TreQna 2005 .. this yield can directly be converted to Sigma by using a sigma conversion table.

Look for yield @ USL or LSL in session window 5. Under Variable select column that has data on project Y 3. Calculate Z value using Abridge table First Edition 169 © All Rights Reserved TreQna 2005 .5 to arrive at ZST To convert continuous data to discrete follow the following steps 1.5 (ZShift = 1.Treqna Base Manual Choosing a Capability Tool Start Is data normal ? Is data discrete ? No Yes Run Capability Analysis (normal data) Read ZLT from Capability Analysis report and 1. Select “Count” and “Cumulative percent” 4.5) Yes Run Capability Sixpack (see next page for interpretation) Read ZLT from Capability Analysis report and 1.5 to arrive at ZST No Treat data as discrete for this step and run Discrete Data Process Capability report (see notes) Yes Run Discrete Data Process Capability report (excel spreadsheet) Is target specified? Yes No Is sub group size > 1? No ZST = ZLT + 1. Minitab>Stat>Tables>Tally Individual variables 2.

5 (standard shift) to the long term sigma First Edition 170 © All Rights Reserved TreQna 2005 .Treqna Base Manual Process Capability for Discrete Data ZLT : Long term capability can be calculated using excel formula normsinv(yield) ZST : Short term capability is computed by adding 1.

2 cms • LSL = 9.Treqna Base Manual Process Capability . The ‘length of the pen’ is the measure. • USL = 10. If upper and lower specification limits are provided without a target.mtw • The capability of a Pen manufacturing plant is being calculated.8 cms • The data was collected in batches of 10(Subgroup size = 10) • Stat > Quality Tools > Capability Sixpack > Normal • In ‘Options’ – Enter Target = 10 If subgroup size = 1. Minitab assumes that the target is the mid point of the specification range. First Edition 171 © All Rights Reserved TreQna 2005 . If only one specification limit is entered and the target is left blank.Continuous Data • • Objective: Practice generating a Minitab Capability Sixpack Directions: • Open file Penlength. then Minitab approximates the target with the mean of the data. the short term sigma value reported in the Capability Analysis report is invalid as there are no sub groups and hence no sub group variation.

Continuous Data First Edition 172 © All Rights Reserved TreQna 2005 .Treqna Base Manual Process Capability .

Treqna Base Manual Process Capability .Continuous Data First Edition 173 © All Rights Reserved TreQna 2005 .

Continuous Data First Edition 174 © All Rights Reserved TreQna 2005 .Treqna Base Manual Process Capability .

If R chart is out of control. First Edition 175 © All Rights Reserved TreQna 2005 . Xbar chart interpretation may not be reliable as the control limits on the Xbar chart are arrived at using the Rbar value from the R chart.Treqna Base Manual Interpreting the Capability Sixpack Always check for S Chart stability to judge whether rational sub grouping is acceptable . the rational sub grouping is not correct. If the R chart is not in control. Within subgroup variation is higher than between sub group variation.S chart in control indicates sub groups are acceptable .S chart out of control indicates sampling is not acceptable -If the graph shows an R Chart instead of S then.

621 69.680 8.023 6.100 6.000 274.7 13.000 158.8 6.200 4.0005 0.20% 81.000 0.8 22.02 0.0 99.8 2.6 3.9 44.048 0.5 4.700 6.46 4.4 21.000 75.930% 98.080 3.200 24.6540% 99.9900% 99.5 18.1 3.50% 86.68 8.390 1.7 1.440 3.000 500.135 0.068 0.9990% 99.320% 3.000 6.860 2.480 621 33 48 68 96 135 186 255 346 466 23 62.80% 75.740 668 66.080 9.000 54.096 0.9992% 99.130% 96.05 0.800 22.2 1.5 3.1810% 98.Treqna Base Manual Abridged Yield to Sigma Conversion Table Long-Term Yield 99.000 90.68 0.007 0.4 4.5 0.400 8.00% 50% 46% 43% 39% 35% ST Process Defects Per Defects Per 1.1 0.700 35.400 21.60% 61.200 808 968 1.2 4.4 38.6 46.680 11.150 1.80% 58.68 11.3 2.9770% 5.9670% 99.2 27.900 690 69 72 75 78 81 84 86 88 90 92 First Edition Note: Subtract 1.3 3.9 4.800 9.500 13.000 344.2 2.3 0.9 0.600 5.50% 84.5 2.60% 65.000 Sigma 100.8 6.000 810.5 1.900 44.0034 0.2 0.0 6.000 135.000 6.000 750.3790% 4.070 1.99966% 99.1 0.6 0.46 5.9960% 99.730% 97.4 5.320% 88.000 540.7 4.4 0.20% 2.700 28.800 330 480 680 960 1.186 0.870 3.8 80.000 420.8 4.9 2.350 1.800 8.66 8.350 1.580 1.35 1.800 80.9520% 99.000 86.1 81.000 242.39 1.5 5.08 0.002 0.07 0.7 3.01 0.000 840.015 0.78 2.003 0.000 88.000 72.000 78.7 0.000 920.34 0.460 4.1 1.000.220% 97.7 1.9995% 99.8 0.000 6.5 Defects Per 1.9 107 139 178 227 287 359 446 548 3.59 4.2 0.200 42.000 880.800 96.400 5.255 0.8 18.01 0.590 4.19 10.100 8.1 4.4 0.000 0.500 7.7 28.000 57.8 1 2 3 4 7 10 15 99.1 Defects Per 10.9 3.8 9.034 0.8 31% 28% 25% 22% 19% 16% 14% 12% 10% 8% 1.4 34.60% 78.0008 0.8 3.7 2.27 2.9850% 99.8 115 135 158 184 212 242 274 344 382 420 460 500 540 570 610 650 720 750 780 810 840 860 880 900 920 0.3 4.3 5.9980% 99.6 5.9970% 99.001 0.000 860.4 1.840 2.000 720.000 900.000 92.820 4.7 35.0 1.460 5.000 610.15 Defects Per 100 0.600 54.520% 91.270 2.2 3.660 230 93.610% 98.3 1.000 650.900 17.9 17.03 0.000 69.800 115.400 38.800 18.000 84.7 5.5 to get long-term Sigma level 176 © All Rights Reserved TreQna 2005 .000 46.6 13.3 4.410% 95.4 3.2 42 46 50 54 57 61 65 308 30.6 4.9 1.33 0.210 8.21 0.540% 94.08 9.80% 72.8140% 99.2 5.000 212.420 2.000 61.5340% 99.1 0.9320% 99.55 3.550 3.000 50.7450% 99.920% 90.000 30.0 66.000 9.0 5.8 96.780 2.000 460.500 7.819 1.9930% 99.0 308.6 2.033 0.004 0.07 1.6 2.5 13.4 2.200 27.3 0.000 184.87 3.000 3.6 25.0 690.1 2.466 0.5 0.700 13.9 5.4 5 8 10 20 30 40 70 100 150 0.04 0.190 10.6 54.000 5.000 570.6 1.500 15.200 7.346 0.23 0.86 2.008 0.00034 0.000 81.2 24.8650% 99.005 0.9040% 99.48 0.000 382.800 34.000 65.8 1.3 0.5 15.48 3.5 34.600 8.400 819 1.00% 54.120 2.000 780.8 5.96 1.

Shift is taken as 1. process performance is more consistent in the short term than the long term. hence short term sigma is always better than the long term sigma.Treqna Base Manual Long Term & Short Term Capability ZLT = Long term process sigma Long term sigma is also the sustained process capability ZST = Short term process sigma The process at entitlement is performing at its best. First Edition 177 © All Rights Reserved TreQna 2005 . This can happen when process short term sigma (Zst) is as high as possible.5 if the data lacks rational sub grouping* * Discussed Later ZST assumes: Theoretically. ZST = ZLT + Shift Where shift is the difference between short term process capability and long term process capability.

If a normal process is centered at 6 standard deviations from each specification limit (or is on target) in the short term.5 sigma comes from manufacturing industry. First Edition 178 © All Rights Reserved TreQna 2005 .5 standard deviations from one specification limit and 7. the process would tend to move 1. six sigma is a measure of the short term capability of a process. Shift compensates for the variation which is non-random in nature and does not get captured while calculating short term sigma. Therefore.5 standard deviations from the other. the process will be 4.5s USL The assumed shift of 1.5 standard deviations towards either specification limit in the long term.Treqna Base Manual The Shift Centered Distribution LSL ± 6s m m T USL Distribution Shifted 1. This means.5σ LSL T 4. It accounts for the changes in process capability over many cycles of that process. assuming standard shift.

the process mean typically shifts by 1.Treqna Base Manual Visual Representation of ST & LT Short-Term capability Time = t Time = t + x Long-Term capability LSL T USL As discussed in the previous slide.5 standard deviations around the target. Short term sigma is a measure of the process at a particular point in time whereas long term sigma is a measure of the process over a period of time. First Edition 179 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Types of Variation Common Cause variation • Random process variation • Also called as inherent variation of any given process • Variation comes from – People – Materials – Methods – Machines – Measurements – Mother Nature Special Cause variation • Non-random process variation • Assignable cause • Can generate outliers Non-Random Variation: Shifts Trends Cycles First Edition 180 © All Rights Reserved TreQna 2005 .

This variation is accounted primarily by special causes (though this might include some common cause variation also) In other words. The attempt is made to collect samples of data in such a way that special cause accounts for variation between different subgroups. To create rational subgroups find: A time period that is short enough such that you do not have This time period will differ from process to process A subgroup should only contain common cause variation. short term capability should have common cause variation only.Within Group Special Cause Variation – between groups Time The objective behind doing Rational Subgrouping is to have subgroups with only common cause variation and minimal special cause variation within them. Long term capability has common cause and special cause variation special causes of variation First Edition 181 © All Rights Reserved TreQna 2005 . There is variation between different subgroups.Treqna Base Manual Rational Subgrouping Rational Subgroups Common Cause Variation .

3 applications – Sample each shift. use the Xbar and R-chart to evaluate stability of short/long-term stability. process measurements typically come “one-at-a-time. 3 applications – Sample each day.” Use the X & RM chart. 3 applications In high-volume situations.Treqna Base Manual Rational Subgroups Characteristics of good rational subgroup: – Samples to be collected over a large period of time so that each sample represents a rational subgroup defined by some logic(which usually is a one process cycle) – Attempt to minimize variation within the subgroup and maximize the variation between subgroups Examples of sampling to create rational subgroups: – Sample each hour. In low-volume situations. First Edition 182 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Components of Variation [ SST SST Total Total SSW } SST SSB [ SSW SSW Within Within ⎞ 2 + ⎟ ⎠ Samples size n=3 at different times SSB SSB Between Between − X ∑ ∑ (X K N j=1 i=1 ij ) 2 = N ∑ K ⎛ ⎜ ⎝ j=1 X j − X ∑ ∑ (X K N j=1 i=1 ij − X j )2 Where K=number of subgroups and N= subgroup size Total variation representing Long term process capability The Shift defined by special cause variation Common cause variation representing Short term process capability SSTotal : deviation of each individual point from the overall mean SSBetween: deviation of each subgroup mean from the overall mean (variation of the means) SSWithin : deviation of each individual point from its corresponding group mean (variation of the individuals within group) Xij : Individual observation Xj X : Subgroup mean : Overall Mean First Edition 183 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Standard Deviation & Sum of Squares One Sample ˆ σ = ∑ (X n i i − X − 1 ) 2 Where “n” is the sample size (n g ) Long-Term ˆ σ LT = ∑ ∑ j= 1 n i= 1 (X gn ij − X ) 2 − 1 Where “n” is the subgroup size and “g” is the number of subgroups. Numerator denotes SSW First Edition 184 © All Rights Reserved TreQna 2005 . X represents overall mean. Numerator = denotes SST. Short-Term ˆ σ = ∑ ∑ (X g n j= 1 i= 1 ij − X j ) 2 ST g (n − 1 ) Where “n” is the subgroup size and “g” is the number of subgroups.

5 sigma. ZLT = SL− μ σLT Z Short-term: Z short term is the best process can deliver if the current distribution is assumed around the Target (and not mean). SL .m is the distance between specification limits given by customer and the true population mean. the population mean.Treqna Base Manual Z Equation The Equation Z = Specification Limit – Central Tendency Standard Deviation Variations Z – Short Term(ZST ) or Long Term(ZLT) Specification Limit – USL or LSL Central Tendency – Mean(m) or Target(T) Standard Deviation – σST or σLT Z Long-term: Which is the actual capability of the process and what is current. For the same reason while calculating Z short term we use Target value and not population mean. so the shift is ZShift=ZST-ZLT Higher the shift bigger the control issue in process.T is the distance the specification is from the target. It is always calculated w. First Edition 185 © All Rights Reserved TreQna 2005 .r. ZST = SL− T σST ZST is bigger than ZLT.t. This means assuming the same variation and if the distribution were to centered around Target what will be process capability. Typical assumed shift when rational subgrouping not present is 1. SL .

5 2. A typical DMAIC project aims at improving controls by eliminating special cause variation. Poor Technology 0. the difference between long term and short term capability • Caused by assignable special causes • Signifies control issues in the process Seeing Short term Sigma and Shift together provides insight into what needs to be changed to improve Long Term process capability. you will be in a position to determine whether your process needs better controls. DFSS projects on the other hand are run to improve technology so that the process can reach higher capability.Treqna Base Manual Relating ST Capability & Process Shift Short Term Sigma • Short Term Sigma signifies the best process can deliver as of current.5 1 2 3 4 Good Control.0 Poor Control. First Edition 186 © All Rights Reserved TreQna 2005 . improved technology or both. • Entitlement of process with current technology • Any attempt to increase the capability would require improvement in current technology Shift • Shift in the distribution over a period of time. 2. Good Technology 5 6 Zst The graph to your right helps identify the issue in your process. Poor Technology Poor Control.5 1.0 Zshift 1. Once you baseline your process and understand it’s short term capability and shift.

5s with assumed shift of 1. If special cause variation is also present along with common cause variation. but consistency among the subgroups or need to revisit sub groups. However.5s.(shift) SHORT-TERM CAPABILITY (ZST) • Best process performance – Entitlement • Way to improve involves improving technology as it is the inherent capability of process at current • 6σ means ZST = 6σ ZST is based on the analysis of subgroups.5. if ZShift is greater than 1. otherwise 6s . this is a definite indication of a control problem. ZST will not reflect the correct (best) picture of present technology. First Edition 187 © All Rights Reserved TreQna 2005 .Treqna Base Manual Summary LT vs. ST LONG-TERM CAPABILITY (ZLT) • Actual current process capability • Way to improve involves improving control (reducing shift) and technology • 6s means 4. A small ZShift does not necessarily indicate good control.

First Edition 188 © All Rights Reserved TreQna 2005 . standard deviation • For capability use Z values from minitab capability reports Normality • Process with only random variation will have such distribution • Can always be related to normal distribution which has mean of zero and standard deviation of one Sample: • Subset of population which has same characteristics as population. median • Variation – variance. Defectives • DPMO • For capability Use ZLT from Z table corresponding to DPMO Continuous data characteristics • Central Tendency – mean.Treqna Base Manual Step Summary – Baseline Process Discrete Data Characteristics • Defect.

box plots. normality graph from minitab to understand process Rational Subgrouping • Attempt is made to collect samples with only random variation(common cause variation) • Between subgroup variation is special cause variation • Derive SSB and SSW to understand ZST.Treqna Base Manual Step Summary – Baseline Process Basic Statistics tools: Use graphical summary. ZLT and Shift Z-value • Signifies number of standard deviation that can be fit between customer specification limit and mean • Corresponds to probability of defect. run charts(for stability). which is area under the curve outside the customer specification limit • Represents process capability First Edition 189 © All Rights Reserved TreQna 2005 .

Short term capability is the inherent capability of process – “Entitlement” First Edition 190 © All Rights Reserved TreQna 2005 .Treqna Base Manual Step Summary – Baseline Process Shift • Difference between ZLT and ZST • Signifies process control issues • Assume shift to be 1.5s to estimate ZST if rational subgrouping not present • Attributed by special and assignable causes What the customer feels is the process Long term capability.

Common cause variation. Simple Linear Regression. Cause and Effect diagram.3 Performance Objective Set Improvement goal for project Y -Concept of Benchmarking -Benchmarking Types -Benchmarking as process Identify drivers of variation List all statistically significant X’s -Root Cause Analysis (RCA) -Process map analysis -Graphical analysis -Statistical Analysis Activities Benchmarking 1-Sample T-test. Long Term sigma -Understanding shift. stability and normality -Capability analysis -Short Term vs. Correlation/Scatter Diagrams.Treqna Base Manual Analyze Phase DEFINE MEASURE Step A. run chart. Chi Square–Test for Goodness of Fit. Homogeneity of Variance. 2-Sample T-test. capability pack Tools ANALYZE Step A. Chi Square–Test of Independence.2 IMPROVE CONTROL Step A.1 Objective Deliverables Baseline Process Process capability for project Y -Study data for shape. Mood’s Median. special cause variation Descriptive test. Oneway ANOVA. Pareto. VA/NVA analysis First Edition 191 © All Rights Reserved TreQna 2005 .

E.reduce DPMO by 50% Six sigma should always have aggressive but achievable target First Edition 192 © All Rights Reserved TreQna 2005 .Treqna Base Manual Setting a goal for the Y metric Following are approaches which one could adopt to define the Goal for Y metric: • Benchmarking: One can target to achieve best in industry • Learning Curve-Based: One could also use multi level goal in terms of sigma for the process metric. • Arbitrary Defect Reduction: This is frequently used with discrete metric.g.

process baseline tells where the process stands against Industry First Edition 193 © All Rights Reserved TreQna 2005 .Treqna Base Manual Benchmark and Baseline Benchmark Z Short Term Z Long Term Benchmark : World Class Performance Z Short Term : The best performance by the process with current technology Baseline Process Sigma : The sustained long term sigma for the process Benchmark is the best in industry.

The subjects that can be benchmarked include strategies. products.Treqna Base Manual Benchmarking Definition: An improvement process in which a company measures its performance against that of best in class companies. It also provides a platform for best practice sharing across industry. Where to use Benchmarking: • Compare process performance against industry’s best • Understand market competition • Establish improvement targets for your process Benchmarking is a continuous process which gives insight into where does your competitor stand and how do they achieve those levels of service excellence. operations. benchmarking is definitely a tool to facilitate service excellence achievement . In the current market scenario where information flows seamless and the access is universal. It also provides target for process capability improvement. The knowledge derived from the process of benchmarking can be used to improve service. processes and procedures. First Edition 194 © All Rights Reserved TreQna 2005 . support functions and systems. determines how those companies achieved their performance levels and uses the information to improve its own performance.

It involves considering high level aspects such as core competencies. It helps in driving short term goals of company. Functional Benchmarking It is used when organizations look to benchmark with market players drawn from different business sectors to find ways of improving similar functions or work processes. First Edition 195 © All Rights Reserved TreQna 2005 . Performance/ Competitive Benchmarking It is used where organizations consider their positions in relation to performance characteristics of key products and services. This sort of benchmarking can lead to innovation and dramatic improvements. Usually focuses is the relevant segment of market. developing new products and services. Process Benchmarking It is used when the focus is on improving specific critical processes and operations. Benchmarking partners are sought from best practice organizations that perform similar work or deliver similar services.Treqna Base Manual Types Of Benchmarking Strategic Benchmarking It is used where organizations seek to improve their overall performance by examining the long-term strategies and approach.

International Benchmarking It is used where partners are sought from other countries because best practitioners are located elsewhere in the world and/or there are too few benchmarking partners within the same country to produce valid results First Edition 196 © All Rights Reserved TreQna 2005 . from business units located in different areas. and.Treqna Base Manual Types Of Benchmarking Internal Benchmarking It involves seeking information from within the same organization. The main advantages of internal benchmarking are that access to sensitive data and information are easier. for example. External benchmarking provides opportunities of learning from those who are at the leading edge. standardized data is often readily available. External Benchmarking External Benchmarking involves seeking outside organizations that are known to be best in class. usually less time and resources are needed.

Treqna Base Manual Benchmarking Plan Primarily following are the things to be considered in your benchmarking plan: • Focus of Study • • • • • • • • Current assessment of area requiring improvement Selection of process for benchmarking Description of current process Purpose of study Outline of areas/issues for questioning rationale for selection initial contact development of detailed questions • Benchmarking partners First Edition 197 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Benchmarking Plan • Site Visit •confirm and arrange site visit •develop briefing package •finalize and communicate plans •conduct visit • Post Visit •points of clarification / follow up •review all data •initial report sent to partners for confirmation First Edition 198 © All Rights Reserved TreQna 2005 .

g.Treqna Base Manual Other sources for setting a goal Following are few more options which drive the goal for process metric: • • • • Corporate mandate Compliance/legal requirement Voice Of Customer Industry Standards (e. ISO) First Edition 199 © All Rights Reserved TreQna 2005 . CMM.

products and systems – Process of knowing how competitor is achieving higher levels of service excellence – Benchmarking is a time consuming and labor intensive process Common Benchmarking Mistakes: – Not establishing the baseline – Confusing benchmarking with participating in a survey. – The process is too large and complex to be manageable – Confusing benchmarking with research – Picking a topic that is too intangible and difficult to measure – Not researching benchmarking partners thoroughly – Not having a code of ethics and contract agreed with partners First Edition 200 © All Rights Reserved TreQna 2005 . – Thinking there are pre-existing "benchmarks" to be found.Treqna Base Manual Summary – Performance Objective What is Benchmarking? – Benchmarking is a continuous process – Valuable tool to gather market knowledge and improve services.

Long Term sigma -Understanding shift. run chart. capability pack Tools ANALYZE Step A. 2-Sample T-test.1 Objective Deliverables Baseline Process Process capability for project Y -Study data for shape. Cause and Effect diagram. VA/NVA analysis First Edition 201 © All Rights Reserved TreQna 2005 . Chi Square–Test of Independence.2 IMPROVE CONTROL Step A. Correlation/Scatter Diagrams. Chi Square–Test for Goodness of Fit.3 Performance Objective Set Improvement goal for project Y -Concept of Benchmarking -Benchmarking Types -Benchmarking as process Identify drivers of variation List all statistically significant X’s -Root Cause Analysis (RCA) -Process map analysis -Graphical analysis -Statistical Analysis Activities Benchmarking 1-Sample T-test. special cause variation Descriptive test. Pareto. Homogeneity of Variance. Mood’s Median. Simple Linear Regression. Oneway ANOVA. Common cause variation.Treqna Base Manual Analyze Phase DEFINE MEASURE Step A. stability and normality -Capability analysis -Short Term vs.

By the end of this section we would have established Critical X’s A Statistical Problem .Treqna Base Manual Bridging the Gap -Practical to Statistical D M Practical Problem • Project Y • Project Charter • MSA • Data Collection • Process Characteristics This section is the beginning of statistical problem solving process.Centering and Variation • Establish Process Capability • Determine Statistical Goal (Benchmarking) • Critical X’s Statistical Solution I C • Transfer function Y=f(X) • Pilot/Implement Solution • MSA on Vital X’s Practical Solution • Control Plan First Edition 202 © All Rights Reserved TreQna 2005 .

It is also used to prioritize X’s which affect project Y First Edition 203 © All Rights Reserved TreQna 2005 . Process map analysis • • VA/NVA analysis Value mapping 8.Treqna Base Manual Tools to be used in this section Graphical Tools 1. Box and whisker plot 7. Fishbone analysis 2. Dot Plot 6. Histograms 5. Pareto 4. Hypothesis Tests Cause and Effect diagram is used in measure for identifying segmentation factors. Matrices 3.

Treqna Base Manual Cause and Effect Diagram (Fishbone) Use the tool: • To get a visual map of all causes • While brainstorming for causes and trying to capture thoughts • Correctly identify root causes and categorize them First Edition 204 © All Rights Reserved TreQna 2005 .

for example “Why is the cycle time high?” Use the basic categories as framework: – People – Materials – Methods – Machines – Measurements – Mother Nature Make your core team understand the problem statement and objective of exercise Cause and Effect diagram is a team tool and so it is important to have members in your team who understand the process well. First Edition 205 © All Rights Reserved TreQna 2005 . It should start with “Why”.Treqna Base Manual How to work with a Fishbone Define your problem statement.

Treqna Base Manual How to work with a Fishbone Building the Diagram – Brainstorm possible causes and attach them to appropriate categories – Relate each cause with the problem statement – Try to bucket small related causes and prioritize The Structure of Diagram: S a m p le F is h b o n e S a m p le F is h b o n e M e a s u r e m e n ts M e a s u r e m e n ts M a t e r ia l M a t e r ia l P e r so n n e l P e r so n n e l P u t in y o u r Q P u t tin n o u r u e s io y Q e h e r u e s t io n h ere E n v ir o n m e n t E n v ir o n m e n t M e th o d s M e th o d s M a c h in e s M a c h in e s The attached diagram is an example to the structure. which once filled gives complete picture of problem and causes. First Edition 206 © All Rights Reserved TreQna 2005 .

t.t.r. to there effect on Y • Quickly surfaces critical X’s from disagreements within the team First Edition 207 © All Rights Reserved TreQna 2005 . its importance and effect on Y How does it help? • Helps six sigma team to focus on X’s on priority • Systemizes the approach of prioritizing as it based on on X’s weighed against each other w.Treqna Base Manual Prioritization Matrix The matrix only prioritizes…the output still needs to be ratified with data and an objective Few prioritization matrices which can be used are: outlook • Criterion v/s criterion • QFD • Control/Impact Why use it ? • To narrow down systematically to critical X’s by weighing each option w.r.

Treqna Base Manual Fishbone to Prioritization Matrix Translate the Cause and Effect diagram to a Prioritization Matrix • Items in Quadrant 1 are ‘Must do’ • Items in Quadrant 2 should be dropped as they complicate processes without significant impact • Items in Quadrant 3 should only be implemented if Project Y does not show necessary improvements after implementing items from Q1 • Items in Quadrant 4 are NOT to be done First Edition 208 © All Rights Reserved TreQna 2005 .

A Pareto is very handy when: – You want to prioritize which causes to eliminate first – You want to display information objectively to others First Edition 209 © All Rights Reserved TreQna 2005 .Treqna Base Manual Pareto Chart Use the tool: To filter vital few from trivial many X’s affecting process Y.

an Italian economist who studied the distribution of wealth in a variety of countries around 1900. Following are few segments: •By Cost •By Financial defects •By risk •By compliance/ legal errors First Edition 210 © All Rights Reserved TreQna 2005 . He discovered a common phenomenon: about 80% of the wealth in most countries was controlled by a consistent minority -.Treqna Base Manual The Pareto Principle The 80:20 rule “ 20% of the causes account for 80% of the effect” The 80:20 rule originated from Vilfredo Pareto.about 20% of the people. While making Pareto the defects grouped by different segments depending on what makes sense to the business. His observation eventually became known as either the "80:20 rule" or "Pareto's Principle".

Data collection by different defect categories. compliance defect and so on.Minitab 1) Collect data P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s P ro c e s s A A B B C C D D E E S o ft S k ills Is s u e S o ft S k ills Is s u e 585 585 821 821 149 149 946 946 415 415 P o o r K n o w le d g e P o o r K n o w le d g e 3949 3949 5408 5408 994 994 6307 6307 3112 3112 C aal l l s M oonni i t or reedd to C ls M N o P r o b le m R e s o lu tio n N o P r o b le m R e s o lu tio n 390 390 548 548 99 99 631 631 277 277 2).Treqna Base Manual Pareto Chart . Total results and arrange data by Segmentation Factors S oo f ft t S kk i il ll lss I Iss ss uu ee S S P oo oo r r K nn oo w l lee dd gg ee P K w N oo P r roo bb l lee m R ee ss oo l luu t ti ioo nn N P m R TT oo t taa l l dd ee f fee cc t tss 22 99 11 77 11 99 22 77 66 22 00 44 55 Following are the steps involved in making of pareto chart: 1. Put sum of frequencies of defects by category First Edition 211 © All Rights Reserved TreQna 2005 . risk. This helps in building an approach towards problem 2. While doing so we capture: Types or categories of defects Frequency of defects For further analysis we might do a drill down on financial loss.

Treqna Base Manual Pareto Tool Utility: Useful tool identify factors/segments which contribute to the majority of the effect How: Minitab > Quality Tools > Pareto Chart First Edition 212 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Pareto Minitab Inputs Choose ‘Chart defects table’ – Labels in ‘Defects’ and Frequencies in ‘Counts’ First Edition 213 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Pareto Minitab Output First Edition 214 © All Rights Reserved TreQna 2005 .

Understand what does customer feel and what customer values. It solves two important purpose: – Identification of problems and improvement opportunity in the process – Sense of urgency to improve as it is seen from customer perspective Types Of Analysis: What does the customer feel?. customer is ready to pay? ”Value to customer” Wait time in process Further on the three key components to the process map analysis: •Moments of truth helps us on focusing on customer and this paradigm change in approach changes the comfort level at which the process is operating. •Value analysis is process of studying the value add activities and non-value add activities in the process.Treqna Base Manual Process Map Analysis Tool Utility: Useful tools to approach process from customer perspective. We also study the touch points of customer with the process and improve the experience. With this approach we try to study the customer – process interaction. First Edition 215 © All Rights Reserved TreQna 2005 . Every activity within the process is seen as whether it adds any value to customer or not. •Wait time is related to the process workflow. “The moment of Truth” Which parts of the services provided.

They aid in delivery of product and services. First Edition 216 © All Rights Reserved TreQna 2005 . Which is a process step repeated.g. E. a step which usually takes you back in the process. Customer Is not willing to pay for process step. Defect caught in downstream step being sent to upstream step in process. Notes on Value enable work: Value enable work is the category which confuses the most. Given the current condition of any given process we might not be able to take out the value enable steps from the process. Answers to this question amount to Value added work which changes the product/ services physically and so is essential to the process. Nonvalue-Added Work This is part of process steps which is Non-Essential for production and delivery product/ service. Indicators of non value add in the process: Approval process involves multiple departments Too many supervisors Multiple reporting Typical non-value add is rework.Treqna Base Manual Value Analysis Value-Added Work For every activity in the process ask the question “Is Customer Willing To Pay For Them?” Value-Enabling Work Process steps which are non essential but enable or support the Value-Adding Tasks to be done better/faster. The only distinguishing factor between non value add and value enable work comes from customer’s perspective where customer is definitely not ready to pay for non-value added work.

g. Rework • External failure are the defects reported by customer. Changing settings of machine in manufacturing plant Inventory Movement: Physical movement of inventory from one point of processing to other point of processing. It is usually due to bottlenecks.g.g. E. Wait/ Delay: • Products/ services waiting or queuing to get processed in front of process step. E. First Edition 217 © All Rights Reserved TreQna 2005 . Internal and External.Treqna Base Manual Types Of Non-Value-Added Work Inspection (Quality Check) Failure (Internal/External) Wait / Delay Setup Initialization Inventory Movement Process failures: • Failures happen two ways. Backlogs Setup / Initialization: • Steps involved in setup or preparation of subsequent step. Internal refers to defect found during processing due to upstream process and being corrected. Inspection(Quality Check): • Internal process step dedicated to check products/ services for defects. It is inventory building in front of a process step. E.

g in call centre the average wait time for call might be 60 seconds but customer might be going through longer wait due to IVR. It is also important to look at this metric from customer point of view. Wait / Delay Time: Total idle time spent by any product or service while going through process steps. First Edition 218 © All Rights Reserved TreQna 2005 . E. Concept of time allocation in process: Cycle Time Total time spent from start point of process to end point of process. (Note: this also includes steps which are non value add).Treqna Base Manual Workflow and Time allocation Wait / Delay Time Processing Time (Product Touch Time) T1 T2+T2 Cycle Time T2 While analyzing workflow for a process one needs to go at product/service unit level and track the flow of that identity. In the process we categorize process step into value add. This is very detailed process map with time allocation with each small step. which we don’t capture. non value add and value enabling. Process Time It is the actual spent on any unit of service or product which goes through the whole process.

Redundancies: When one process step is duplicated amount two or multiple processors. Tricky Handoffs: Work transfer from one person or step to another step or person without any operational definitions in place. Leading to delays or rework. This happens when one processor is unaware of other processor working.Treqna Base Manual Workflow: Disconnects in Process Steps Unclear work responsibility Unclear processing requirements Redundancies Tricky handoffs between process steps Conflict in job goals Process disconnects more often than not lead to delay and wait time in the process. Conflict in Job goals: When the job goals are overlapping or do not cover complete processing involved leading to conflict or incomplete work. Most of the disconnects arise due to following reasons: Unclear work responsibility: Happens when a process step is not owned by a particular person. First Edition 219 © All Rights Reserved TreQna 2005 . It also leads to rework or resource under utilization. The processing such environment happens in a very ad hoc manner leading to delay. Unclear processing Requirements: Operational definitions for process step do not exist. This might lead to rework due to defects caught in later process steps.

The components of cycle time are processing time. First Edition 220 © All Rights Reserved TreQna 2005 .Treqna Base Manual Analyzing Workflow: “Track the unit flow” Agent fills app Agent fills app with customer with customer Agent mails app Agent mails app using USPS using USPS App received in App received in mail room mail room Mail room sorts Mail room sorts by department by department Data entry apps Data entry apps placed in data placed in data entry tub entry tub Apps transferred Apps transferred to data entry to data entry every two hours every two hours Data entry operator Data entry operator screens app for screens app for completeness completeness Is app complete ? Yes No Procure missing Procure missing information from information from agent agent Route to Route to Resolution team Resolution team Order print Order print Complete data entry Complete data entry and forward to audit and forward to audit team team Is information entered correct ? Yes Sort issues and Sort issues and declines declines No Correct Correct information information Create policy package for Create policy package for Issued policy and letter for Issued policy and letter for declined policies declined policies Route to Route to underwriter underwriter Can we issue policy ? Yes Apply postage and mail to Apply postage and mail to agent using SPS agent using SPS No Decline policy Decline policy Agent receives letter / / Agent receives letter package and notifies client package and notifies client Issue policy Issue policy Cycle time is the total time taken from the point at which the customer requests a good or service until the good or service is delivered to the customer. But when the customer might think otherwise as you might have another process step having lot of delays. Lot of times we might just look at one part of process which might be working at its best. The customer always looks at total cycle time. inventory movement time. inspection time. Cycle time metric should always be seen with caution.wait time.

Treqna Base Manual Value Analysis Process Step 11 22 33 44 55 66 77 88 99 10 11 12 13 14 15 16 17 18 Total Time % Total Process Step 10 11 12 13 14 15 16 17 18 Total Time % Total Time Taken (min) 22 22 10 21 120 33 44 55 33 77 99 12 11 22 33 21 240 10 475 100% Time Taken (min) 10 21 120 12 21 240 10 475 100% Value Add 24 5% Value Add 24 5% Value enabling Value enabling Rework Rework Transportation Transportation Inventory / /queue Inventory queue Audit / /Inspection Audit Inspection Prep / /Set up Prep Set up 99 66 261 261 147 147 24 24 44 2% 2% 1% 1% 55% 55% 31% 31% 5% 5% 1% 1% Usually. and non-value add. How to Interpret the Matrix: Look for the longest cycle time component. can you still reduce time? Look for frequent category of non-value add steps in your process?. Conclusion: Example 4 steps (<25% of total steps) provide 100% value and take 24 minutes or 5% of total cycle time. When we put our cycle time analysis against value analysis it gives tremendous insight into improvement opportunities. Is it a value-added step? If not. we might come the data where non-value added work adds to maximum in cycle time Note: We might come across a situation where for a process step we may not have any data available. and consume 95% of cycle time. value enable. Look for overall breakup of cycle time into value add. In such a scenario it is best to estimate time than skip the step. 14 steps (>75% of steps) are non-value-added. This might give overview of improvement opportunities. It is an eye opener exercise to understand the process. It tells exactly where we spent maximum time in process and whether or not it is value added work. They might be driven due to things outside process. First Edition 221 © All Rights Reserved TreQna 2005 . a majority of the time is spent in Non Value add or Value Enabling work Value analysis breaks the process into small steps categorized into value add and non value add.

and how is the variation against the customer specification limits.Treqna Base Manual Histogram Tool Utility: Study variation in process. or without classes 10. How: Minitab. 11 etc. Histogram can also be made using excel sheet. In order to do so: . age 10-20. Gives visual display of data distribution as bar graph (frequency graph).Plot the frequency against individual X values or Classes on X-axis.create a frequency table with or without defined classes(e. 1 0 1 0 # of Students # of Students 5 5 0 0 5 5 5 5 6 0 6 0 6 5 6 5 7 0 7 0 T e s t G ra d e s T e s t G ra d e s 7 5 7 5 8 0 8 0 8 5 8 5 9 0 9 0 9 5 9 5 1 0 0 1 0 0 First Edition 222 © All Rights Reserved TreQna 2005 .) . Useful tool to determine if distribution is centered around target.g.

Click on ‘Simple’ 2.mtw 1.Treqna Base Manual Histogram – Minitab Click on Graph > Histograms MINITAB FILE: DescriptiveStats.Click Ok First Edition 223 © All Rights Reserved TreQna 2005 .

Is it off the process target?. • • The options at the bottom of histogram input window gives option to put specification limits. Spread: How is the spread of bars in histogram?. Note that spread has got nothing to do with Normality of data. Is the spread large compare to requirement?(look against the specification limits). Center: Where is the distribution centered?. First Edition 224 © All Rights Reserved TreQna 2005 . as you might have normal data with very large spread when compared against specification limits from customer. Are there multiple modes in the distribution?.Treqna Base Manual Histogram Output Histogram is the graphical representation of the distribution of population plotted. This might give us idea if the given data is mix of two or more different population sets. It helps understanding following for any given population: • Shape: Look for the distribution of bars.

As a part of descriptive statistical analysis used in initial stages. Also used for quick comparison of two or more groups for variation. Click ok First Edition 225 © All Rights Reserved TreQna 2005 .Treqna Base Manual Dot Plot Tool Utility: Another tool to study variation in process. How: Use Minitab GRAPH > DOTPLOT Click on ‘Simple’. Click Ok Double Click on C1.

mtw Dot plot can be plotted from GRAPHS> DOTPLOT In Minitab Release 14 the Dotplot from Graphs menu gives multiple option for Single Y. First Edition 226 © All Rights Reserved TreQna 2005 . Multiple Y. Multiple group and there combination.Minitab MINITAB FILE: DescriptiveStats. Single group.Treqna Base Manual Dot Plot .

Treqna Base Manual Dot Plot Output When Dotplot plotted for multiple operators (groups) gives us comparative visual representation of their: •Variation •Center • and spread First Edition 227 © All Rights Reserved TreQna 2005 .

Quartiles . How: Minitab > Graph > Boxplot First Edition 228 © All Rights Reserved TreQna 2005 . The tool plots: .Outliers. if any The box tells about the spread of data and is a visual used for comparing two or more data groups.Median .Treqna Base Manual Box And Whisker Plot Tool Utility: Useful tool to compare two or more groups on variation and median.

How to Read ? * Outlier any point outside the lower or upper limit Maximum Observation that falls within the upper limit = Q3 + 1.5 (Q3 .Treqna Base Manual Box plot .Q1) 75th Percentile (Q3) Median (50th Percentile) 25th Percentile (Q1) Minimum Observation that falls within the lower limit = Q1 .1.Q1) First Edition 229 © All Rights Reserved TreQna 2005 .5 (Q3 .

Treqna Base Manual Box plot . Click Ok.mtw Minitab > Graph > Boxplot • Choose ‘One Y’ ‘With Groups’.Minitab MINITAB FILE: BoxPlots. • Use C6 for Data. • C7 for Subscripts • Click ok First Edition 230 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Statistical Hypothesis Testing First Edition 231 © All Rights Reserved TreQna 2005 .

First Edition 232 © All Rights Reserved TreQna 2005 . In Analyze phase we run statistical tests to determine which X’s have an effect on the Y. These tests may be testing for variance or for mean depending on what is our Y metric. more often than not we would target only one problem at a time. Focus on reducing Variation Desired Process Current Process T LSL USL Poor Accuracy. Focus on Centering the process While working on Six Sigma Projects we come across two types of problems with data.Treqna Base Manual The Nature Of Statistical Problems Desired Process Current Process LSL T USL Poor Precision. •Variation (Spread) •Centering Though working on one problem has some impact on other also.

Treqna Base Manual A Statistical Hypothesis Definition of Hypothesis: Statement about the parameters of the Population In hypothesis testing there are two hypotheses of interest. – The null hypothesis (H0) – The alternative hypothesis (HA) First Edition 233 © All Rights Reserved TreQna 2005 .

First Edition 234 © All Rights Reserved TreQna 2005 . 2. Are these Samples from the same distribution or are they truly from different processes. Different Samples from the same process may vary 3.Treqna Base Manual Concept Of Hypothesis Testing 1. Hypothesis testing differentiates between sampling error (Precision) and true process differences. All processes have variation.

Validate improvement in process Statistically validate our statement about the characteristics of population Helps in decision making process. Enabler for data driven approach. First Edition 235 © All Rights Reserved TreQna 2005 .Treqna Base Manual Why Do Hypothesis Testing? Identification and validation of factors which impact process performance.

Treqna Base Manual Hypothesis Testing In Hypothesis Testing one has to make a decision: – – to either to accept null hypothesis or to reject null hypothesis The hypotheses are always statements about the population parameters.) • The Critical Region (The range of values of the test statistic that indicate the Null Hypothesis is false.) First Edition 236 © All Rights Reserved TreQna 2005 . To define a statistical Test we – – – Choose a statistic (called the test statistic) State your null and alternate hypothesis Divide the range of possible values for the test statistic into two parts • • The Acceptance Region The Critical Region •The Acceptance Region (The range of values of the test statistic that indicate the Null Hypothesis is true.

No difference True unless proven otherwise Represented by = or > or < It is the conclusion we are trying to make from data Signs used in Minitab: ≠ or < or > First Edition 237 © All Rights Reserved TreQna 2005 .Treqna Base Manual Nature of Ho and Ha Null Hypothesis (Ho ): Alternative Hypothesis (Ha ): Null hypothesis always represents Status Quo.

First Edition 238 © All Rights Reserved TreQna 2005 . Relating to the previous example about the height of people from two different countries. • If there is not enough evidence we accept null hypothesis (defendant is innocent) Hypotheses are statements about population parameters. • The hypothesis than is: • Ho: Person is not guilty. not the sample means.Treqna Base Manual Criminal Trial Analogy • In the court defendant is always treated as innocent until proven guilty. • There after once all the evidence(data) is collected we make a decision based on it: • If there is sufficient evidence(statistically significant) we reject null hypothesis and accept alternate(defendant is guilty). we could state: Ho: μB ≤ μA Ha: μB > μA The statement is about the population means. • Ha: Person is guilty.

Jailed 239 © All Rights Reserved TreQna 2005 . Jailed Guilty. Set Free Verdict Verdict Ho Ha First Edition Jailed Innocent.Treqna Base Manual Decision Error Based on the evidence there are four possible outcomes of decision we make: Truth Truth Ho Innocent Set Free Innocent. Set Free Ha Guilty Guilty.

Usually we have α set at .α = Confidence Ha Guilty Type II Error β Jailed Ha Type I Error α Correct Decision 1 .β = Power • 1 . • It is not possible to commit a Type I and Type II decision error simultaneously For any statistical testing procedure define 1. α = P[Rejecting the null hypothesis when it is true] = P[ type I error] 2.β = Probability of correctly accepting the alternative hypothesis (detecting a change when there is one).α = Probability of correctly accepting the null hypothesis (detecting no change when there is none). β = P[accepting the null hypothesis when it is false] = P[ type II error] The symbol α is called the level of significance.Treqna Base Manual Decision Error .05. First Edition 240 © All Rights Reserved TreQna 2005 . The Acceptance Region and The Critical Region are chosen form underneath the sampling distribution of the test statistic when H0 is true. The Critical Region lies in the tails of sampling distribution of the test statistic when H0 is true.α) confidence in accepting null hypothesis. • 1 . which means we have 95%(1.Types Truth Truth Ho Innocent Set Free Verdict Verdict Ho Correct Decision 1 .

Ho : μ = constant = T Ha : μ ≠ constant = T Ho Ha 2.Treqna Base Manual Types of Hypothesis One Sample Hypothesis 1. Ho : σ2 = constant Ha : σ2 ≠ constant T In this and following slides read: μ as mean and σ as standard deviation First Edition 241 © All Rights Reserved TreQna 2005 .

Ho : μ1 = μ2 Ha : μ1≠ μ2 4.Treqna Base Manual Types of Hypothesis Two Sample Hypotheses 3. Ho : σ12 = σ22 Ha : σ12 ≠ σ22 σ1 σ2 First Edition 242 © All Rights Reserved TreQna 2005 . Ho : μ1 ≤ μ2 Ha : μ1 > μ2 μ1 μ2 5.

= σn2 Ha : at least one not equal First Edition 243 © All Rights Reserved TreQna 2005 . Ho : σ12 = σ22 = . . .Treqna Base Manual Types of Hypothesis Multi Sample Hypotheses 6. . . Ho : μ1 = μ2 = . = μn Ha : at least one not equal 7.

The symbols used for one-sided tests are ≥. ≤. Ho : μ1 = 10 Ha : μ1 ≠ 10 4 5 Acceptance Region 1-α α/2 6 7 8 9 α/2 μ =101011 12 13 14 15 16 The symbols used for two-sided tests are = and ≠. < and > First Edition 244 © All Rights Reserved TreQna 2005 . Ho : μ1 < 10 Ha : μ1 > 10 4 5 6 Acceptance Region 1-α 7 8 9 α/2 12 13 14 15 16 μ = 10 10 11 Two-Sided Test Ex.Treqna Base Manual One-sided & Two-sided Hypothesis Tests One-Sided Test Ex.

05 means we reject the null hypothesis and accept the alternate hypothesis p < a: Reject Hoo p < a: Reject H p > a: Accept Hoo p > a: Accept H First Edition 245 © All Rights Reserved TreQna 2005 . Any p-value less than 0.Treqna Base Manual The P-Value Alpha denotes the probability of making a Type I error P-value is the probability value based on which we make judgment on hypothesis.

Moods Median Test – tests if medians from 2 or more sample means are equal Y = Normal First Edition Y = Non Normal .One way ANOVA – tests if 2 or more samples have equal means .HOV (Levene) – Compares 2 or more samples 246 © All Rights Reserved TreQna 2005 .1-sample T test – tests if a sample means is equal to a known mean or target .F test – Compares 2 samples .HOV (Levene) – Compares 2 or more samples .Two way ANOVA – tests if means from samples classified by 2 categories are equal Test for MEDIAN .χ2 Test – Compares a sample variance to a known population variance Central Tendency Test for MEAN .Treqna Base Manual Hypothesis Tests Summary Continuous Y and Discrete X Variation .2-sample T test – tests if 2 sample means are equal .

hence we can use Normal distribution for predictions.Treqna Base Manual Normal (z) Distribution When sample size is infinite uncertainty prediction becomes very small. Sample size infinite Z = X σ− μ μ df = ∞ First Edition 247 © All Rights Reserved TreQna 2005 .

we might be interested to know if factor X has an impact on variance in Y. First Edition 248 © All Rights Reserved TreQna 2005 . Some tools which are used to test means of groups use pooled deviation and so HOV is a tool which used before we decide which tool we will use before testing for means.Treqna Base Manual Homogeneity of Variance (HOV) To compare variance of two or multiple data samples we use Homogeneity of Variance (HOV) Test. The hypothesis for the test is: HO = Variance for all the groups are equal Ha = At least one group variance is different HOV is also prerequisite to testing for group means. While working on project if our metric is related to controlling or decreasing spread.

mtw First Edition 249 © All Rights Reserved TreQna 2005 .Treqna Base Manual HOV . Perform the Homogeneity of Variance Test.Minitab If you wish to study variation across 2 or more Groups and you have established that each groups stable and normally distributed. Minitab File: HOV.

Input H H o a : σ : σ 2 1 2 1 = σ ≠ σ α = .05 2 2 2 2 First Edition 250 © All Rights Reserved TreQna 2005 .Treqna Base Manual HOV .

First Edition 251 © All Rights Reserved TreQna 2005 . Levene’s Test: Use this test result when data is continuous and not necessary normal. Bartlett’s Test: When your data is normal and have more than two groups under evaluation. Do not use this test result when your data is non. The test results which we can read from left side window are: F-Test: We get this result only when there are two groups and normal data. For more than two groups and normal data F-test is replaced by Bartlett’s test.Treqna Base Manual HOV Test .normal as Bartlett’s test is not robust to nonnormality.05 p < a: Reject Ho p > a: Accept Ho In the output window the graph on left side represent confidence interval around standard deviation of each group.Output H H o a : σ : σ 2 1 2 1 = σ ≠ σ 2 2 2 2 α = . On the right hand side we get P-values for the test.

Treqna Base Manual 2 Sample t test HO (Null Hypothesis) : μ1 = μ2 HA (Alternate Hypothesis) : μ1 = μ2 First Edition 252 © All Rights Reserved TreQna 2005 .

Treqna Base Manual 2 Sample t test .Minitab Minitab File: HOV.mtw First Edition 253 © All Rights Reserved TreQna 2005 .

Treqna Base Manual 2 sample t .Input Check this box if HOV Indicates that there is no statistically significant difference in Variation among the groups being tested Click ‘Options’ to change your Alternate Hypothesis Run Homogeneity of Variance test before you indicate to assume equal variances in the input window. First Edition 254 © All Rights Reserved TreQna 2005 .

28 P-Value = 0.mu (Agent B) Estimate for difference: 20.Treqna Base Manual 2 Sample t -Output Two-Sample T-Test and CI: Agent A.45 Difference = mu (Agent A) .4920.7580) T-Test of difference = 0 (vs not =): T-Value = 36.000 DF = 73 Accept HA (Alternate Hypothesis) : μ1 = μ2 The p value indicates that the two agents are statically different in their performance.85 0.6250 95% CI for difference: (19.35 0. First Edition 255 © All Rights Reserved TreQna 2005 . 21.58 2.19 Agent B 40 49. Agent B Two-sample T for Agent A vs Agent B N Mean StDev SE Mean Agent A 40 70.20 2.

The hypothesis used is: Ho: μ1 = μ2 = μ3 = μ4 Ha: At least one m different from the others Since ANOVA uses pooled standard deviation it is essential to run HOV before this test.Treqna Base Manual One Way Analysis of Variance (ANOVA) ANOVA stand for Analysis of Variance. however the difference comes from the fact that Two Sample T-test can analyze only two sample at a time and so running it for multiple groups is a logistical nightmare. First Edition 256 © All Rights Reserved TreQna 2005 . There are two ways ANOVA is different from Regression Analysis: •The independent variables are categorical •It doesn’t make any assumption about the the nature of relationship between Y and independent variable. Though the name is misleading One Way ANOVA is used to compare means of two or more groups. Two Sample T-test and One Way ANOVA test the same thing. In case HOV fails than one may use Two Sample T-Test. Though Two Sample T-test gives us flexibility of not assuming equal variances ANOVA v/s Regression: Analysis of variance (ANOVA) is similar to regression in that it is used to investigate and model the relationship between a response variable and one or more independent variables.

mtw First Edition 257 © All Rights Reserved TreQna 2005 .Treqna Base Manual One Way ANOVA .Minitab A Test to determine if the means of the number of Queries Resolved by 4 agents is statistically different Ho: μ1 = μ2 = μ3 = μ4 Ha: at least one m different from the others Minitab File: Anova.

1 Total 199 12274.908 2.0 Pooled StDev = 3.234 (*-) Agent B 50 29.6 F P 3 9998.0 Agent A 50 19.01 0.46% R-Sq(adj) = 81.6 3332.596 4.0 42. Confidence band to understand which ones are different First Edition 258 © All Rights Reserved TreQna 2005 .940 2.0 36.408 --------+---------+---------+---------+- An ANOVA output gives us: 1.6 S = 3.9 287. P-value for significance 2.17% Individual 95% CIs For Mean Based on Pooled StDev Level N Mean StDev --------+---------+---------+---------+(-*) (*-) (-*) 30.874 4.049 Agent D 50 28.Treqna Base Manual One-Way ANOVA – Input/ Output One-way ANOVA: Average Queries Resolved versus Agent Source DF Agent SS MS 11.000 Error 196 2276.164 24.779 Agent C 50 39.408 R-Sq = 81.

Normality Test Possible reasons for not normal data Understand interpretation of Mood’s Median Test Understand HOV Test First Edition 259 © All Rights Reserved TreQna 2005 .Objectives Understand distribution .Treqna Base Manual Non–Parametric Test .

First Edition 260 © All Rights Reserved TreQna 2005 . Tools used for analysis are: Stability : Run Chart Centre tendency : Mood’s Median Test Spread (Variance) Capability : HOV – Use Levine's test : Use DPMO to calculate sigma It is common to have data which is not not normal.Treqna Base Manual Non-parametric Tests What if I don’t have normal data? When we don’t have Normal data we use non-parametric tests. Distribution should be check for multiple population Check for outliers Normal data gives flexibility in terms of number of tools available for analysis. Nonparametric means that no assumption for distribution of data exist. Care should be taken before reaching that conclusion. However sufficient tools are available for not normal data.

Metric data where only one sided specifications limits are defined. First Edition 261 © All Rights Reserved TreQna 2005 .Treqna Base Manual Non-parametric Tests The term nonparametric means that no assumption is made about the distribution of data. Examples of non-normal data: Cycle time data is usually non normal data. – Mood’s Median Test is a nonparametric test which is frequently used to compare medians of two or more groups. One way ANOVA is the counterpart of this test for Normal data. Usually a skewed distribution. – Non-parametric tests are used when we have no assumption for distribution of the data to work with.

limit your removal of data to only a few data points). Common transforms include: – finding the square root of all data points – finding the log of all data points – finding the square of all data points – Box-cox Transformation Care is required before concluding that data is non-normal. this is not recommended. However it should not be used as a compulsion. The measurement System should be able to capture the unit of data in an adequate fashion. Primarily because more often than not data is supposed to be normal. – Check your sample size.Treqna Base Manual Dealing With Non-Normal Data Do some checks before concluding that data is non normal: – Normality Test – Look for resolution of data. Consult with MBB/BB before you use transformation. – Attempt to transform the data. – Investigate outliers(A word of caution: sometimes while taking out outliers we end up taking out a considerable amount of data from sample. Use the non-normal tests! First Edition 262 © All Rights Reserved TreQna 2005 . Small sample size (<30 points) might not be able to predict population characteristics correctly. However we still encounter situations when we come across nonnormal data.

Are the medians different? (interpret the p-value) 2. File: MoodsMedian Stat > Nonparametrics > Mood’s Median Test – Response = Average Call Handle Time – Factor = Team Answer the following questions: 1. Which medians are different? (interpret the confidence intervals) To perform a Mood’s Median Test on collected data: Enter Y data in one column in Minitab Second column should define subgroups Choose test from Stat > Nonparametric > Mood’s Median Test First Edition 263 © All Rights Reserved TreQna 2005 .g.Treqna Base Manual Mood’s Median Test Hypothesis for the test: HO = Medians for all the groups are equal HA = At least one median is different E.

Treqna Base Manual Mood’s Median Test Select Stat > Nonparametrics > Moods Median Test First Edition 264 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Mood’s Median Test – Input Response data or Y Factors or X’s First Edition 265 © All Rights Reserved TreQna 2005 .

0 Mood’s Median test output gives a table where it defines by group number of data items below and above group’s median.0 (---*--) 1 14 296.0% CIs Team N<= N> Median Q3-Q1 -----+---------+---------+---------+A B C 7 8 280. you would get the same result.0 (-----*--) (----*----) Confidence interval -----+---------+---------+---------+264 276 288 300 Overall median = 279.Output Mood Median Test: AHT versus Team Mood median test for AHT Chi-Square = 26.0 14. On the right side of table we get a visual of confidence band around group medians.0 10.Treqna Base Manual Mood’s Median Test . First Edition 266 © All Rights Reserved TreQna 2005 .0 13. It then performs a simple Chi-square test on this summary table.32 DF = 2 P = 0. If you would manually run Chi-square test on the table. Based on the test it gives P value.0 15 0 263. When the test is significant you would see that one of the group confidence band far from others.000 P value Individual 95.

. oscillations or mixtures Normality Test Ho : Data is normally distributed Ha : Data is not normally distributed Mood’s Median Test Ho : X1 = X2 = X3 =…. cluster oscillations or mixtures Ha : There exist trends.= Xn Homogeneity of variance Ho : σ12 = σ22 =….. = σn2 Ha : σ12 ≠ σ22 ≠ …. clusters.. ≠ σn2 ∼ ∼ ∼ ∼ Ha : At least one median is different from the others First Edition 267 © All Rights Reserved TreQna 2005 .Treqna Base Manual Hypothesis for various tests Run Charts Ho : There are no trends.

Treqna Base Manual Hypothesis for various tests 1-sample t test Ho : μ = T Ha : μ ≠ T 2-sample t test Ho : μ1 = μ2 Ha : μ1 ≠ μ2 Paired t test Ho : Difference between the two means = 0 Ha : Difference between the two means ≠ 0 ANOVA Ho : μ1 = μ2 = μ3 =….= μn Ha : At least one mean is different from the others First Edition 268 © All Rights Reserved TreQna 2005 .

for whatever reason. Though theoretically different. First Edition 269 © All Rights Reserved TreQna 2005 . A significant correlation means that 2 variables tend to go up or down together (or change in opposite directions). again. for whatever reason.Treqna Base Manual Continuous Y and X Continuous Y and Continuous X • Correlation : Tests Linear relationship between two variables • Regression : Defines linear relationship between a dependent and an independent variable Correlation assumes that both variables can change while regression (typically) assumes that the independent variable is measured without error. A significant regression means that the dependent variable can be predicted by the independent variable. both tests usually give the same answer (significant or not). Causality is not tested by either method.

File: Grades.Minitab E.Treqna Base Manual Correlation .g.MTW Stat > Basic Stats > Correlation First Edition 270 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Correlation – Inputs/Outputs First Edition 271 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Regression Minitab First Edition 272 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Regression Minitab Inputs First Edition 273 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Regression Minitab Inputs Click the ‘Graphs’ button First Edition 274 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Regression Minitab Inputs Click the ‘Results’ button First Edition 275 © All Rights Reserved TreQna 2005 .

40 Score2 Predictor Coef SE Coef T P Constant -4.296 0.56 0.3514 12.328 Total 8 53.1% Analysis of Variance Source DF SS MS F P Regression 1 51.353 156.649 First Edition 276 © All Rights Reserved TreQna 2005 .67 + 4.4.001 Score2 4.8572 -5.Treqna Base Manual Regression Minitab Outputs Regression Analysis: Score1 versus Score2 The regression equation is Score1 = .3975 0.44 0.572711 R-Sq = 95.000 S = 0.7% R-Sq(adj) = 95.6674 0.51 0.353 51.000 Residual Error 7 2.

the confidence o df – degrees of freedom for data First Edition 277 © All Rights Reserved TreQna 2005 .Treqna Base Manual Discrete Y & Discrete X Two Chi-Square Tests: o Goodness-of-Fit Test to test if observed data has expected distribution o Test of Independence to test if two or more proportions are different • Chi-Square statistic is calculated using formula: χ • 2 = Σ1 i= n ⎛ ⎜ ⎝ f o − f e e ⎞ ⎟ ⎠ 2 f Critical Value depends on two factors: o a .

Goodness-of-Fit Test: To test if an observed set of data fits an expected distribution of data. .Test of Independence: To test if two or more proportions are associated (Belong to similar distribution) First Edition 278 © All Rights Reserved TreQna 2005 . There are two types of Chi-square χ2 tests: .Treqna Base Manual Chi-Square Test–χ2 Tool Utility: Chi-square tests enables us to test whether two or more population can be considered equal.

Treqna Base Manual Chi-Square Statistic Calculation for Chi-square statistic is: χ 2 = Σ1 i= n ⎛ ⎜ ⎝ f o − f e e ⎞ ⎟ ⎠ 2 f Where: n = number of categories fo = observed frequency of data fe = expected frequency of data First Edition 279 © All Rights Reserved TreQna 2005 .

For any statistic being computed. First Edition 280 © All Rights Reserved TreQna 2005 . the degrees of freedom of a sample are the number of values in the sample which are free to vary without changing the statistic being measured.Treqna Base Manual Degrees Of Freedom The chi-square uses “degrees of freedom” of data under scrutiny as a parameter while calculating Chi-Square critical value.

the alternative hypothesis (that the data are not typical of the expected distribution) is accepted. Steps involved in doing manual calculation are: • Create frequency table and see if any of the frequency is less than 5(as we have mentioned earlier that avoid small frequency) • Calculate Chi-square statistic using the formula • Calculate critical value using the table. the null hypothesis (good fit) is accepted.Treqna Base Manual Chi Square Stat Test The statistical test can be done both manually or using minitab. Critical value depends on two things.df) First Edition 281 © All Rights Reserved TreQna 2005 . Chi-Square Distribution Critical Value A B Area of Rejection P-value can also be calculated using formula in excel: CHIDIST(statistic. • If the chi-square statistic is less than the critical value in the table. If not. a the confidence and degrees of freedom(df).

9 2 4 3 5 .9 5 5 2 3 .0 1 7 1 3 .4 0 1 4 2 .0 7 0 1 2 .4 0 2 8 6 .7 6 6 7 9 .4 8 8 1 1 .3 0 7 1 9 .9 8 0 4 4 .2 9 0 4 9 .7 4 1 3 7 .2 6 6 1 8 .3 3 6 5 3 .6 2 6 7 .3 8 9 1 2 .4 8 8 2 8 .7 0 3 7 3 .5 2 8 3 6 .6 7 2 6 6 .6 7 1 3 3 .2 5 2 4 0 .6 3 8 4 2 .3 3 9 3 0 .0 2 6 2 2 .8 1 2 2 1 .7 1 1 3 4 .4 8 9 2 1 .0 3 7 1 0 .0 0 1 1 0 .5 8 9 2 5 .8 1 9 3 1 .4 0 9 3 4 .2 0 4 2 8 .3 3 7 4 2 .0 0 0 3 3 .7 5 0 1 8 .1 4 1 3 0 .3 6 4 4 0 .2 1 0 1 1 .8 3 8 1 4 .0 7 6 3 9 .8 4 1 5 .1 9 6 3 4 .7 0 6 4 .2 1 7 2 7 .3 6 2 2 3 .6 1 6 5 6 .0 0 7 3 3 .5 1 5 2 2 .0 2 3 2 0 .7 1 8 2 3 .1 9 1 3 7 .5 6 6 3 8 .1 2 3 3 7 .3 2 2 2 6 .1 4 4 3 1 .6 2 9 1 1 8 .5 0 7 1 6 .3 3 7 2 4 .9 7 9 5 9 .2 6 4 3 2 .3 4 2 A lp h a .1 8 8 2 6 .1 7 0 3 5 .2 9 8 9 5 .0 8 6 1 6 .6 3 5 9 .8 7 7 2 9 .8 4 1 9 .8 5 2 3 4 .5 6 1 .9 8 9 2 7 .5 2 6 3 2 .5 0 5 7 9 .5 3 1 1 0 1 .1 7 2 3 6 .1 8 1 4 5 .9 8 1 7 7 .2 8 9 4 1 .5 7 8 1 0 7 .8 1 3 3 2 .9 0 9 3 4 .8 1 6 1 6 .9 2 3 4 3 .3 0 7 2 3 .7 8 1 3 8 .4 9 0 9 1 .6 9 1 7 6 .2 7 7 1 5 .5 9 2 1 4 .1 6 7 7 4 .8 6 9 3 0 .7 1 8 3 7 .8 8 5 4 0 .3 4 5 1 3 .0 8 7 4 0 .2 9 6 2 7 .3 7 8 9 .8 2 8 1 3 .3 1 2 4 3 .7 9 0 4 3 .5 4 9 1 3 .6 7 5 2 1 .0 5 0 3 .1 3 0 9 8 .4 7 9 3 6 .1 6 9 .6 4 5 5 0 .6 4 6 4 1 .9 3 2 4 0 .4 7 5 2 0 .0 3 9 2 7 .3 8 5 6 .5 5 8 4 6 .4 1 2 2 9 .1 9 1 3 1 .2 0 9 2 4 .5 5 7 4 3 .2 3 6 1 0 .1 1 9 2 7 .9 9 7 4 1 .0 5 2 5 5 .7 2 5 2 6 .4 6 1 4 5 .9 1 6 3 9 .8 7 9 1 1 3 .6 6 1 9 9 .3 0 2 5 9 .2 7 5 1 8 .3 2 1 1 2 8 .1 2 5 2 7 .0 1 3 1 7 .3 2 9 1 2 4 .6 0 5 2 2 .1 3 6 1 2 9 .2 0 8 1 4 9 .9 1 9 1 8 .5 4 9 1 9 .2 5 1 7 .2 7 8 2 1 .0 2 4 7 .3 1 7 1 2 4 .9 9 6 2 6 .2 4 1 2 9 .4 9 8 .1 4 5 1 2 4 .8 4 5 3 0 .1 4 3 1 2 .4 7 6 5 6 .7 2 8 5 1 .9 5 2 1 0 4 .7 0 1 1 4 .5 2 8 3 2 .2 1 9 1 1 .2 4 5 1 9 .6 1 5 3 0 .1 9 4 4 4 .9 3 5 2 6 .1 1 6 1 3 5 .8 0 5 3 6 .1 0 8 5 .0 9 0 2 1 .3 1 5 4 6 .6 0 7 1 1 2 .6 0 5 6 .1 1 3 4 1 .8 3 2 1 4 .5 3 5 1 9 .7 2 2 4 6 .4 2 5 1 1 2 .2 6 7 3 5 .4 4 9 First Edition 282 © All Rights Reserved TreQna 2005 .8 1 2 1 8 .6 4 5 1 2 .3 1 9 3 2 .8 9 2 6 3 .8 0 5 6 3 .8 6 0 1 6 .4 1 0 3 2 .9 8 4 1 7 .5 6 3 3 6 .8 3 9 1 3 7 .8 2 8 2 4 .6 2 0 3 3 .6 5 0 1 0 9 .7 7 3 5 5 .7 7 9 9 .3 6 2 1 4 .1 5 6 3 8 .2 1 5 1 1 6 .6 6 6 2 3 .1 1 7 1 8 .5 4 8 2 0 .4 8 3 2 1 .6 8 8 2 9 .0 2 5 5 .1 4 1 2 8 .8 1 5 9 .6 9 7 3 9 .3 2 3 2 .7 3 6 2 6 .7 6 9 2 5 .8 0 7 .4 6 7 2 0 .8 4 5 1 5 .5 7 8 3 2 .3 9 7 8 5 .3 8 2 3 5 .4 4 9 1 6 .1 7 9 5 2 .3 4 8 1 1 .6 2 0 5 4 .8 7 9 1 0 .8 0 1 3 4 .0 1 0 6 .9 9 1 7 .7 7 3 4 .4 3 4 3 1 .7 5 7 2 8 .6 5 2 3 8 .0 6 4 2 2 .2 7 8 4 9 .Treqna Base Manual Chi-Square Distribution df 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 40 50 60 70 80 90 100 .9 2 8 4 8 .3 4 2 7 1 .1 4 1 .0 0 5 7 .5 8 7 2 8 .2 6 8 4 9 .5 8 2 3 9 .5 8 8 3 1 .9 6 3 4 8 .4 2 0 8 3 .5 8 8 5 0 .0 8 2 9 0 .8 0 0 4 5 .2 5 6 5 1 .0 2 3 1 0 6 .3 1 4 4 5 .6 8 4 1 5 .9 2 0 2 3 .5 2 7 9 6 .9 9 3 5 2 .1 5 4 8 8 .7 5 8 6 7 .3 6 9 2 0 .9 8 7 1 7 .3 0 0 2 9 .5 4 2 2 4 .8 9 2 5 8 .7 9 6 4 4 .4 5 8 2 4 .4 1 5 3 7 .6 8 5 2 4 .0 6 7 1 5 .3 7 9 1 0 0 .1 0 0 2 .2 5 0 1 .2 9 9 1 4 0 .7 9 7 4 8 .5 7 7 8 8 .6 4 2 4 6 .5 6 5 1 1 8 .5 9 7 1 2 .3 3 4 6 6 .8 2 0 4 5 .

3 .8 .6 .4 8 5 4 5 6 7 5 3 1 0 .2 .3 .5 .5 .6 5 1 1 2 .9 9 5 .0 3 6 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 1 2 3 4 5 6 7 8 9 9 0 1 2 3 4 .3 .4 3 3 3 2 .9 0 0 .8 .7 .7 5 0 .8 .8 .9 9 1 3 5 .1 .5 3 9 8 0 4 1 0 0 2 6 0 6 3 0 9 7 0 0 0 7 6 4 9 9 2 8 4 1 7 1 3 2 8 1 0 4 3 8 7 1 A lp h a .8 .9 9 7 6 5 5 8 9 5 6 3 0 0 1 2 3 3 4 5 6 6 .0 0 0 0 3 9 0 .6 .2 5 5 4 2 8 3 7 2 2 6 .5 5 4 0 1 1 2 2 3 3 4 4 5 5 6 7 7 8 .8 .8 .6 .4 .2 .3 .0 3 4 8 .5 2 0 1 1 1 1 1 1 1 2 3 3 .0 0 0 1 6 0 0 .3 5 7 4 .5 0 3 9 6 6 7 0 4 9 4 4 3 0 8 5 8 4 2 0 7 .5 .3 .3 8 6 2 .8 .3 4 4 4 4 4 8 6 4 3 2 1 0 0 9 9 8 8 8 8 7 7 7 7 7 7 6 6 6 6 6 7 .5 .3 .6 .0 0 0 9 8 0 0 .5 8 4 8 .6 4 3 9 .3 .2 6 0 9 .6 .6 6 0 4 2 .7 .4 5 5 0 0 0 1 .3 5 7 4 0 .8 .6 .3 5 1 5 6 7 8 9 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 .1 9 4 2 3 8 5 5 3 3 9 .1 .3 .4 5 9 5 6 7 8 5 4 3 2 .3 .3 .2 0 7 0 .8 3 1 1 1 2 2 3 3 4 5 5 6 6 7 8 8 9 1 1 1 1 1 1 1 1 1 1 .5 .1 .1 .8 6 5 1 1 .0 .1 .2 .7 .8 .2 .0 .4 .6 .4 .1 .5 .5 .0 .1 .5 .5 0 9 3 4 .2 .7 8 8 8 0 2 4 7 0 4 9 40 50 60 70 80 90 100 2 0 .3 9 0 1 0 .5 .4 1 9 7 6 5 4 4 3 3 3 4 4 5 6 7 2 2 5 2 2 4 0 7 7 9 3 9 7 7 8 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 .7 .7 6 0 6 2 8 0 8 1 1 7 8 .2 9 7 0 .9 4 2 5 2 .7 0 7 3 7 .6 .0 .2 1 3 1 .3 .6 7 5 3 4 5 5 6 .1 .9 9 0 .2 .3 3 5 4 9 .3 4 4 4 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 7 .1 .8 .3 3 5 5 9 .3 .1 6 4 2 9 .3 7 7 9 2 5 2 6 8 2 2 .1 0 1 5 0 0 0 .2 9 4 6 7 8 9 1 1 0 0 .4 .7 .9 .7 .3 .3 .1 .9 3 9 2 2 9 1 6 9 2 9 .1 .9 .2 7 3 4 8 5 5 7 0 6 2 1 0 1 3 6 2 9 6 8 8 3 1 7 0 9 2 8 5 3 0 .2 1 1 0 .5 2 4 1 1 1 1 1 2 2 3 4 4 .6 .7 .5 .4 8 2 4 5 6 7 8 7 5 4 .3 .3 3 3 3 3 4 4 4 4 First Edition 283 © All Rights Reserved TreQna 2005 .5 .5 0 0 .2 .3 .9 2 3 2 .0 5 1 3 7 .8 8 6 1 0 .4 .6 8 9 4 6 .3 .3 2 7 9 5 9 8 1 8 3 3 .2 .3 .3 5 2 0 .7 1 1 1 .1 1 7 1 0 .1 9 6 1 0 .9 5 0 .2 .3 .8 .4 .2 1 6 0 .2 .0 .4 .0 6 4 1 .0 4 4 5 6 2 0 4 5 2 4 .2 .7 0 7 2 7 .9 .1 6 5 1 1 .7 .8 9 7 9 .6 .0 .0 .3 .1 .9 .3 .6 .Treqna Base Manual Chi-Square Distribution d f 1 2 3 4 5 6 7 8 9 10 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 .8 .2 .9 6 2 8 .5 .8 5 1 1 1 1 1 1 1 1 1 1 1 1 2 3 3 4 5 6 6 7 8 .0 .0 .0 .2 .3 .7 .3 .0 0 3 9 3 0 0 .4 3 8 9 .4 7 8 4 3 5 0 7 6 7 0 4 9 6 4 3 6 9 4 5 6 3 4 5 5 1 2 7 5 4 4 .1 0 3 0 .9 .2 .6 .9 .9 .2 .0 5 1 0 .0 8 5 1 0 .4 4 3 1 1 1 1 1 1 1 1 1 2 3 4 4 5 6 7 8 8 9 0 .3 .3 .5 7 5 1 .3 .3 .0 1 0 0 .7 6 4 4 3 .0 2 0 0 .3 1 2 1 0 .5 4 2 1 0 .5 3 5 4 5 5 6 3 1 9 7 .2 3 6 3 2 4 7 2 9 7 6 5 7 3 5 0 5 6 2 1 1 .2 .4 9 3 9 4 1 7 5 2 0 9 1 8 1 8 1 9 1 8 8 3 9 .0 .3 .2 9 9 1 0 .3 .5 4 4 4 5 7 9 1 3 6 9 0 1 8 9 3 2 4 9 8 9 8 .6 1 0 2 2 3 4 4 5 6 7 7 8 .4 1 2 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 .0 1 5 8 0 0 0 .6 .6 .3 .9 .1 .0 7 2 0 .1 4 5 1 2 2 3 3 4 5 5 6 7 .1 .5 8 4 1 .4 .8 .3 .5 .6 .3 .6 7 2 9 .3 3 5 6 7 8 9 9 9 9 9 .7 .1 .1 1 5 0 .3 .6 .1 .6 .6 .9 .4 .3 .2 .9 7 5 .7 .4 8 4 0 .1 8 8 5 6 6 7 1 0 9 7 .8 5 6 1 1 .7 5 5 7 9 3 5 5 1 9 7 .1 .0 .1 .3 6 6 3 .3 .2 .

The hypothesis used is: H0 = all proportions are equal Ha = at least one proportion is different from rest Again as a caution frequency less than 5 not to be used for the test. The rows and the columns define the classification of data. Following is an example of contingency table: Tele sales Team A Leads Dead Ends fo = 37 fo = 63 Total = 100 Tele sales Team A fo = 113 fo = 167 Total = 280 Total = 150 Total = 230 Degrees of Freedom: Degrees of freedom for the test are calculated using formula: (M-1)*(N-1) Where M and N are number of columns and rows respectively. First Edition 284 © All Rights Reserved TreQna 2005 .Treqna Base Manual Test of Independence This Test is used to check independence of two or more proportions. To do the test we use Contingency tables.

Treqna Base Manual Test of Independence – Minitab Input The test of Independence includes calculating Chi-Square statistic and Critical value from chi-square table using degrees of freedom and confidence.However. doing the test on minitab is easier: First Edition 285 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Test of Independence Minitab Input And Results Ho: Ha: No correlation between team and leads generated. Correlation between team and leads generated First Edition 286 © All Rights Reserved TreQna 2005 .

Treqna Base Manual

Tool picking guide

First Edition

287

© All Rights Reserved TreQna 2005

Treqna Base Manual

Statistical Test Choices

First Edition

288

© All Rights Reserved TreQna 2005

Treqna Base Manual

DEFINE MEASURE ANALYZE IMPROVE CONTROL

Map project Approve project

CTQ Characteristics and standards Measurement System Analysis Data collection

Baseline process Performance objective Identify drivers of variation

Screen for vital X’s Study interaction between X’s Defined Improve Process

MSA on X’s Improved Process capability Establish control plan

First Edition

289

© All Rights Reserved TreQna 2005

Treqna Base Manual

Improve Phase Objectives
• Develop and propose a solution
• • •

Identify Vital X’s Develop transfer function Experiment with possibilities and develop solution

• Confirm that proposed solution will improve the process
• •

Pilot solution with small scale tests in real business environment Run statistical tests on the results to confirm the solution

• Identify and deploy resources to implement the solution

• Deploy solution on the process

First Edition

290

© All Rights Reserved TreQna 2005

Treqna Base Manual

Improve Phase
DEFINE MEASURE
Step I.1 Objective Screen for Vital Xs Vital Xs Deliverables Identify Vital Xs Activities

ANALYZE

IMPROVE

CONTROL
Step I.3 Define Improved Process Pilot Solution

Step I.2 Study Interaction between Xs Interaction between Xs, Transfer function Develop a transfer functions

Experiment with various solutions

Tools

FMEA, Process Map, Pareto, Fishbone analysis

Control Impact matrix, Regression

First Edition

291

© All Rights Reserved TreQna 2005

Treqna Base Manual

Characterization of Xs
Vital X’s or key elements which impact project Y have following characteristics: • Independent variables • Can be set at different levels to manipulate the response variable or dependent variable Y. • Variation in independent variable contributes to variation in dependent variable Y • Different X’s will have different impact on Y depending on the transfer function • May be continuous and/or discrete • Independent variables might not be measurable (E.g. alternative work-flow sequences)

First Edition

292

© All Rights Reserved TreQna 2005

Treqna Base Manual

Screening of Xs
Screening of X’s is done for various reasons: • Focus on few factors which make most impact • Ensure best utilization of resources available for the project Tools Used: • FMEA • Pareto • Fishbone analysis • Screening DOE if advanced cases when X’s have interaction

First Edition

293

© All Rights Reserved TreQna 2005

So working on rest of the factors might not be as beneficial. First Edition 294 © All Rights Reserved TreQna 2005 .Treqna Base Manual Using Pareto Project Metric: Errors in Application form Analyze result gave us seven factors which are significant. Now the pareto chart can be used to prioritize or screen further these factors. The output clearly says that Name. DOB and Country field contribute around 80% errors.

On insurance application form where premium calculation is based on age risk associated with error in DOB field would be more than error associated with wrong address as it might result in financial loss to the insurance company. First Edition 295 © All Rights Reserved TreQna 2005 . E.Treqna Base Manual Using FMEA When all the significant factors are available from Analyze step A3. Risk priority number.g. One might use FMEA to evaluate corresponding risk associated i.e.

Transfer function Develop a transfer functions Experiment with various solutions Tools FMEA. Pareto. Regression First Edition 296 © All Rights Reserved TreQna 2005 .Treqna Base Manual Improve Phase DEFINE MEASURE Step I.2 Study Interaction between Xs Interaction between Xs. Process Map.3 Define Improved Process Pilot Solution Step I. Fishbone analysis Control Impact matrix.1 Objective Screen for Vital Xs Vital Xs Deliverables Identify Vital Xs Activities ANALYZE IMPROVE CONTROL Step I.

Treqna Base Manual Problem Solving Process Practical Problem Statistical Problem Statistical Solution Transfer Function Y=F(X) Practical Solution First Edition 297 © All Rights Reserved TreQna 2005 .

X2.Treqna Base Manual Transfer Function The goal of improve is to develop a solution using the transfer function Y = {X1. X3…Xn} The transfer function: • Relates the vital X’s to the project Y • Predicts the effect and direction of changes on Y due to changes in X’s • Helps in fitting a solution which maximizes the effect on Y using all or few of the X’s First Edition 298 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Transfer Function The goal of improve is to develop a solution using the transfer function Y = {X1. X3…Xn} Tools that give mathematical transfer function: • DOE • Regression • GLM (General Linear Model) First Edition 299 © All Rights Reserved TreQna 2005 . X2.

while length of the call has low but negative effect on the call quality. In absence of a transfer function it is a very handy tool to get insight into development of possible solution Influence on Call Quality (Project Y) + Subject Knowledge Training - High Finance Background Length of the call Low First Edition 300 © All Rights Reserved TreQna 2005 .Treqna Base Manual Interaction Matrix Using the Interaction Matrix – In the following example training has high and positive impact on call quality.

Fishbone analysis Control Impact matrix. Pareto. Process Map.3 Define Improved Process Pilot Solution Step I.1 Objective Screen for Vital Xs Vital Xs Deliverables Identify Vital Xs Activities ANALYZE IMPROVE CONTROL Step I.2 Study Interaction between Xs Interaction between Xs.Treqna Base Manual Improve Phase DEFINE MEASURE Step I. Transfer function Develop a transfer functions Experiment with various solutions Tools FMEA. Regression First Edition 301 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Tolerancing Tolerance means the allowed variation in a variable Once we have a a transfer function: Y = F (X) We would like to know for given allowed variation in Y what is the variation one would expect for X’s This process of identifying variation limits for X’s using LSL and USL for Y is called Statistical Tolerancing First Edition 302 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Concept of Tolerancing Assume given transfer function Y= aX + b Y USL LSL Y = aX + b Tolerance for Y X The concept can be extrapolated to transfer function involving two X’s by imagining a three dimensions (three axes) Tolerance for X First Edition 303 © All Rights Reserved TreQna 2005 .

E. Specification limits defined by customer • When multiple X’s are involved. Set Tolerance for all simultaneously. specially when the interaction between X’s are involved • Adjust tolerance for measurement variation First Edition 304 © All Rights Reserved TreQna 2005 .Treqna Base Manual More on Tolerancing • Set Tolerance based on customer VOC.g.

Treqna Base Manual Solution Identification This step may be broken into three steps: • Ideas generation • Solution design • Pilot solution First Edition 305 © All Rights Reserved TreQna 2005 .

in its absence one has fair idea about what needs to be done with X’s in order to influence Y positively. i.e.Treqna Base Manual Ideas Generation By this step we have established: • Vital X’s – prioritized • Transfer function – Either we have a mathematical transfer function or we know how are X’s related to Y. what should we do to better Y – increase X or decrease X and will it make more impact on Y than any other X ? • Tolerancing – Based on the transfer function we know how much variation we can allow for each of the X’s While transfer function can exactly tells us the optimum setting for X’s. First Edition 306 © All Rights Reserved TreQna 2005 .

Following may be used for generating ideas: • • • Brainstorming – team tool for generating lots of ideas Problem Analogy – use an appropriate analogy for the current problem and try to generate solution for the same Best Practices – team might also look for similar processes and make a list of best practices which might be useful First Edition 307 © All Rights Reserved TreQna 2005 .Treqna Base Manual Ideas Generation The project team and process experts can together participate in a team effort to generate ideas for solution.

addressing different X’s or supplementing each other First Edition 308 © All Rights Reserved TreQna 2005 .Treqna Base Manual Solution Design Selecting the best Ideas All the ideas and best practices should be than ranked based on following factors in the same priority: • Impact on Y • Resource Requirement for implementation Select ideas which would have maximum impact on Y and minimum resource requirement A solution could be: • One solution • Set of solution.

Helps surface challenges in full scale implementation The objective of running a pilot is to: • Analyze the effects of your solution on the process and plan for a successful fullscale implementation • Establish that the solution would achieve what is desired. • It is the proof of concept and can help in getting buy in from project champion for resource mobilization required for full scale implementation.Treqna Base Manual The Pilot – Proof of Concept Pilot is a process in which a solution is tested on existing process with real business environment. The proof of concept – it is important to prove that a hypothetical solution actually delivers what is expected. • To identify if there would be any challenges in full scale implementation of solution. First Edition 309 © All Rights Reserved TreQna 2005 .

Create data collection plan • Pilot solution and collect data • Analyze data and establish if the solution meets expectation using statistical test • Identify resources required for full scale implementation of the solution First Edition 310 © All Rights Reserved TreQna 2005 . duration and timing.Treqna Base Manual Running a Pilot • Prepare a plan to execute pilot • • • Do risk assessment of doing the pilot. location. resources. use FMEA Discuss and establish test population.

Treqna Base Manual Improve Summary • Vital X’s – select few which make big impact on Y • Tools used are Pareto. prioritizing ideas • Pilot solution – steps in running of a pilot First Edition 311 © All Rights Reserved TreQna 2005 . Fishbone Analysis • Transfer function – what can one do in absence of mathematical transfer function • Statistical Tolerancing – translating allowed variation in Y to X’s • Solution Design • Generating Ideas. FMEA.

Treqna Base Manual DEFINE MEASURE ANALYZE IMPROVE CONTROL Map project Approve project CTQ Characteristics and standards Measurement System Analysis Data collection Baseline process Performance objective Identify drivers of variation Screen for vital X’s Study interaction between X’s Defined Improve Process MSA on X’s Improved Process capability Establish control plan First Edition 312 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Control Phase Objectives • Ensure that the process stays in control after the solution is implemented • Detect and identify any potential deviation from control state and ensure that a corresponding correcting procedure is established • Establish plan to ensure sustenance of improved process performance First Edition 313 © All Rights Reserved TreQna 2005 .

Treqna Base Manual What Does Control Mean • • • Monitor X’s using control charts for variation Identify potential variation causes and prepare an action plan Establish plan to ensure sustenance of improved process performance First Edition 314 © All Rights Reserved TreQna 2005 .

Treqna Base Manual

Control Phase
DEFINE MEASURE
Step C.1 MSA on Xs Objective Acceptable gage Deliverables

ANALYZE

IMPROVE

CONTROL
Step C.3 Establish Control Plan

Step C.2 Improved Process Capability Process capability, PrePost data comparison

Control Plan

Activities

Mistake proofing, Risk management, Statistical process control Refer to step M.3 Refer to Step A.1 and A.3 FMEA & Control Charts

Tools

First Edition

315

© All Rights Reserved TreQna 2005

Treqna Base Manual

MSA on Xs
Deliverable for this step: • • Validate measurement system for Xs Refer to step M3

First Edition

316

© All Rights Reserved TreQna 2005

Treqna Base Manual

Control Phase
DEFINE MEASURE
Step C.1 MSA on Xs Objective Acceptable gage Deliverables

ANALYZE

IMPROVE

CONTROL
Step C.3 Establish Control Plan

Step C.2 Improved Process Capability Process capability, PrePost data comparison

Control Plan

Activities

Mistake proofing, Risk management, Statistical process control Refer to step M.3 Refer to Step A.1 and A.3 FMEA & Control Charts

Tools

First Edition

317

© All Rights Reserved TreQna 2005

Treqna Base Manual

Improved Process Capability
Deliverable for this step: • Calculate capability for new process • Compare pre and post data for project Y using tools described in step A.3

First Edition

318

© All Rights Reserved TreQna 2005

Treqna Base Manual

Control Phase
DEFINE MEASURE
Step C.1 MSA on Xs Objective Acceptable gage Deliverables

ANALYZE

IMPROVE

CONTROL
Step C.3 Establish Control Plan

Step C.2 Improved Process Capability Process capability, PrePost data comparison

Control Plan

Activities

Mistake proofing, Risk management, Statistical process control Refer to step M.3 Refer to Step A.1 and A.3 FMEA & Control Charts

Tools

First Edition

319

© All Rights Reserved TreQna 2005

Treqna Base Manual

Statistical Process Control (SPC)
• Control charts are used to monitor variation on X’s and process performance over time. Control charts help in identifying variation caused due to special cause.

Control charts are not proactive tool as mistake proofing is. They are useful when X’s can not be mistake proofed and controlled within tolerance.

First Edition

320

© All Rights Reserved TreQna 2005

Treqna Base Manual

Statistical Process Control
• The purpose of using statistical purpose control is to monitor X’s and process performance over time for variation. • Post improvements a stable process working at its new capability would have common cause variation. • If the process has changed and has special cause variation than action needs to be taken.

Control charts are used as tools for statistical process control. They tell us when X’s go out of control and surface the special cause performance.

First Edition

321

© All Rights Reserved TreQna 2005

Treqna Base Manual Variation • Common Cause Variation – is the natural or true variation of process and is inherent in the process • Special Cause Variation. It is usually due to special causes and happens once in a while.is usually of larger magnitude compared to common cause variation. • Precautionary measure should be taken to avoid recurrence of special causes First Edition 322 © All Rights Reserved TreQna 2005 .

Only common cause • Out of Control. • Special Cause • Unusual • Specific and once in a while (sporadic) • Captured in control charts • Process stability may be explained as: • In Control. Unstable . Stable .Common cause with special cause is present First Edition 323 © All Rights Reserved TreQna 2005 .Treqna Base Manual Common cause. Random • Inherent to any process • Variation caused by common cause can be reduced by making changes to the system. This would result in improved capability of the system. Special cause • Common cause • Natural.

Precaution should be taken while studying variation. taking precautions.Treqna Base Manual Differentiating Variation Perceived Variation Common Causes Special Causes Common Causes True Variation Special Causes Focus on process change Tampering of process Under-reacting Opportunity missed for taking precautions Focus on mistake proofing. the mistakes in treating common cause as special cause or special cause as common cause causes problem for the process. First Edition 324 © All Rights Reserved TreQna 2005 . We usually end up in either creating more variation in the process or not able to rectify the cause of variation.

First Edition 325 © All Rights Reserved TreQna 2005 . One should calculate the control limits when the process is stable and freeze them. Control charts in the future should be used using the same limits. this would keep an eye on special cause as it should and would also tell if the capability of the process is going down.Treqna Base Manual Control Charts A control chart: • • • Is a time ordered plot of the data Plots the variation of the data in expected range also called control limits Identifies when there is a special cause present Upper Control Limit Lower Control Limit Example of Control Chart Control charts use +/.3 sigma for control limits.

but you should select all eight tests. First Edition 326 © All Rights Reserved TreQna 2005 . Note: Minitab rules only should be used in the exam to evaluate special causes of variation. You can select the rules you want Minitab to use. Each segment becomes a zone in separated from centre line by multiples of s. The eight tests defined in Minitab look for position of data points on in the following zones of control charts. +3σ Zone A +2σ +1σ -1σ -2σ -3σ Zone B Zone C Zone C Zone B Zone A For any control chart. you can find the Minitab Rules under the “test” Button.Treqna Base Manual Control Charts – Minitab Tests Area of any control chart can be divided into different segments in terms of distance from centre line in multiples of standard deviation.

First Edition 327 © All Rights Reserved TreQna 2005 . 1 A B C C B A UCL Average LCL Test 2: 9 data points in a row on the same side of center line. process breakdown etc •Test 2 signals a shift in the process mean.Treqna Base Manual Minitab Tests Test 1: One point or more points more than 3s from the center line. or if there is a “single aberration in the process such as a mistake in calculation. A B C C B A UCL Average 2 LCL •Test 1 is positive if there is a shift in the process mean if there is an increase in the process standard deviation. an error in measurement.

Treqna Base Manual

Minitab Tests
Test 3: Six points in a row, either increasing or decreasing
A B C C B A UCL

Average

3

LCL

Test 4: Fourteen points in row, alternating up and down
A B C C B A

4

UCL

Average

LCL

•Test 3 signals a drift in the process mean. The causes can include improvement in skill, any kind of deterioration in the process. •Test 4 signals a systematic effect produced by two different population. The causes could be due to two different operator, two different process etc.

First Edition

328

© All Rights Reserved TreQna 2005

Treqna Base Manual

Minitab Tests
Test 5: Two out of three points greater than 2s on the same side
A B C C B A

5

UCL

Average

LCL

Test 6: Four out of five points greater than 1s on the same side
A B C C B A UCL

Average

6
LCL

•In the case of charts for variables, the first four tests should be augmented by Tests 5 and 6 when earlier warning is desired.

First Edition

329

© All Rights Reserved TreQna 2005

Treqna Base Manual

Minitab Tests
Test 7: Fifteen points in a row within 1s from center line on either side.
A B C C B A UCL

7
Average

LCL

Test 8: Eight points in a row greater than 1s from centre line on either side
A B C C B A UCL

Average

8

LCL

•Tests 7 and 8 indicate stratification (observations in a subgroup have multiple sources with different means). Test 7 is positive when the observations in the subgroup always have multiple sources. Test 8 is positive when the subgroups are taken from one source at a time.

First Edition

330

© All Rights Reserved TreQna 2005

Treqna Base Manual

Control Limits vs. Specification Limits
Control Limits: Control limits are calculated by Minitab based on the data input. They are internal to the process and are calculated using standard deviation. Control limits refer to stability of the process at a given capability.

Specification Limits: Specification limits are defined by the customer. It defines the desired level of performance by customer. Specification limits are not based on any statistic, but is based on what customer expects. Specification limits are used to derive the defect rate of process and hence the capability of the process.
PROCESS A
Lower SL Upper SL

PROCESS B
Lower SL Upper SL

Process A under control limit and under customer specification limit

Process B under control limit but has unacceptable variation when evaluated against customer specification limits.

First Edition

331

© All Rights Reserved TreQna 2005

Treqna Base Manual

Process States
The Four possibilities of any Process The State of Chaos • • • • • • • • • • • • Non conforming and non predictable Non conformity is also unstable Random changes to process would not fix the process Need to eliminate effect of assignable causes

The Brink of Chaos Confirming yet unpredictable Process output is influenced by assignable causes

The Threshold State Non Confirming and predictable Periodic instability Stable but not able to deliver on customer standards fully

The Ideal State Confirming and predictable Process control limits fall within specification limits Process stays inherently stable over time

First Edition

332

© All Rights Reserved TreQna 2005

Treqna Base Manual

Types of Control Charts
Variable Control Chart • Uses continuous metric like Cycle Time, call handle time, Lengths, Diameters, Drops, etc. • Generally one characteristic per chart

Attribute Control Chart • Uses attribute or ordinal data like Pass/Fail, Good/Bad, Go/No-Go • Could be many characteristics per chart

Variable = Continuous Attribute = Discrete

First Edition

333

© All Rights Reserved TreQna 2005

Treqna Base Manual

Selecting Control Charts
Get Data

Variable or Attribute

Continuous Data (cycle time, length etc

Attribute Data (Yes / No, Go / No-go, Good/ Bad

Rational Subgroups? Constant Lot Size? Yes X-bar and R for sample of size <8 X-bar and S sample of larger size c I-MR Defect or Defective np u Defect or Defective

Yes

No

No

p

First Edition

334

© All Rights Reserved TreQna 2005

it can be changed in chart options) • I-MR Chart: plots individual chart and moving range chart in single window Variable Control Chart with rational Subgroups: • S Chart: a plot of sample standard deviations over time • X-bar Chart: plots of the sample means over time • R-Chart: a plot of the sample range of a sample over time • X-bar and R: plots x-bar chart and r chart in the same chart window • X-bar and S: plots x-bar chart and s chart in the same process window Variable = Continuous Attribute = Discrete First Edition 335 © All Rights Reserved TreQna 2005 .Treqna Base Manual Variable Control Charts Variable Control Chart without rational Subgroups: • Individuals Chart: a plot of the individual values over time • Moving-Range Chart: a plot of the moving range over time (Minitab takes the range as two consecutive data points.

each data point is not influenced by past data point value. they give reliable results for non normal data also. Minitab provides Box-Cox transformation model for transforming data for control charts First Edition 336 © All Rights Reserved TreQna 2005 . • The data has normal probability distribution function • Though control charts assume normality of data.Treqna Base Manual Assumptions for Variable Charts • There are two assumptions about the plotted statistic which control charts make: • Data is random. However if the data is extremely skewed one might like to transform data.

Treqna Base Manual Data Collection • Data should be in time order as used for run charts • Use rational sub grouping as discussed in step M.3 • Use team and subject matter experts for rational sub grouping First Edition 337 © All Rights Reserved TreQna 2005 .

Treqna Base Manual X Bar R Chart • Used for data with subgroup size less than 10.5 U C L=600. which plots subgroup standard deviations instead of range Xbar-R Chart of Weight 600.0 LC L=598.835 Sample Range 2 _ R=1.548 Central line at Mean 599.341 1 0 1 3 5 7 9 11 Sample 13 15 17 19 LC L=0 First Edition 338 © All Rights Reserved TreQna 2005 .775 1 3 5 7 9 11 Sample 13 15 17 19 Control Limits 3 U C L=2. for subgroup size more than 10 use X bar S.5 599.0 _ _ X=599.321 Sample M ean 600.

548 599 LCL=598.0 3 5 7 9 11 13 15 17 19 UCL=2.457 0.0 1 3 5 7 9 11 Sample 13 15 17 19 LCL=0 First Edition 339 © All Rights Reserved TreQna 2005 .8 __ MR=0.5 _ R=1.341 R chart or S chart showing within sample variation 0.0 1 3.493 0.6 3 5 7 9 11 13 15 17 19 Individual charts for sample means.Treqna Base Manual I-MR-R/S (Between/Within) Three charts and three perspectives for the process I-MR-R/S (Between/Within) Chart of Weight 601 Subgroup Mean UCL=600.333 1 1. Refers to between sample variation Sample Range 1.835 LCL=0 Moving range charts subgroup means.763 600 _ X=599. Refers to stability of process location MR of Subgroup Mean UCL=1. control limits are based on moving range of means.

016 0. i.012 0.10268 0.11782 0.115 0.120 U C L=0.11025 Individual values plotted with mean as centre line 13 17 21 O bser vation 25 29 33 37 0.Treqna Base Manual I-MR Chart Used for data with no sub-grouping I-MR Chart of Data 1 Individual V alue 0.110 0.100 1 5 9 1 _ X=0.105 LC L=0.000 1 1 M oving Range 1 U C L=0. length of 2 __ M R=0.00930 Moving range using alternate data values.00285 LC L=0 5 9 13 17 21 O bser vation 25 29 33 37 First Edition 340 © All Rights Reserved TreQna 2005 .008 0.e.004 0.

Treqna Base Manual

Discrete Control Charts

Variable = Continuous Attribute = Discrete

First Edition

341

© All Rights Reserved TreQna 2005

Treqna Base Manual

C Chart
• Charts defects when sub group size is constant and based on Poisson distribution • C-bar is calculated as (Total defects / Total number of units)

C Chart of Data
8 7 6 Sample Count 5 4 3 2 1 0 1 3 5 7 9 11 Sample 13 15 17 19 LCL=0 _ C=2.55 UCL=7.341

First Edition

342

© All Rights Reserved TreQna 2005

Treqna Base Manual

u Chart
• Charts defects when sub group size is not constant and based on Poisson distribution • U-bar is calculated as (Total defects / Total number of units)

U Chart of Defects
0.14 0.12 Sample Count Per Unit 0.10 0.08 0.06 0.04 0.02 0.00 1 2 3 4 5 6 7 8 9 Sample 10 11 12 13 14 15 LCL=0 _ U=0.0478
1

UCL=0.1134

Tests performed with unequal sample sizes

First Edition

343

© All Rights Reserved TreQna 2005

Treqna Base Manual

np Chart
• Charts defectives when sub group size is constant and based on Binomial distribution

NP Chart of Rejects
30
1

25 20 15 10 5 0 1 2 3 4 5 6 Sample 7 8 9 10 LCL=1.21 __ NP=10.8 UCL=20.39

First Edition

Sample Count

344

© All Rights Reserved TreQna 2005

Treqna Base Manual

p Chart
• Charts defectives when sub group size is not constant and based on Binomial distribution

P Chart of Rejects
0.14 0.12 0.10 Proportion 0.08 0.06 0.04 0.02 0.00 1 2 3 4 5 6 7 8 9 Sample 10 11 12 13 14 15 LCL=0 _ P=0.0478
1

UCL=0.1118

Tests performed with unequal sample sizes

First Edition

345

© All Rights Reserved TreQna 2005

Treqna Base Manual

Process Control System
• A process control system is a strategy to sustain the improvement in the process. It includes the action plan for any deviation or variation from the control state. • A process control system includes: • Risk Management • Mistake-proofing • Statistical process control (SPC)

Avoidance Risk Management Mistake Proofing Statistical Process Control Control

Process control system ensures ongoing process control, which also means that the process keeps delivering the improved capability. Factors which make a process control system effective: •Clarity of requirements before making the control plan •Clear communication to involved parties •Good training •Buy-in from stakeholders – level of involvement and ownership from stakeholders can help

First Edition

346

© All Rights Reserved TreQna 2005

Treqna Base Manual

Risk Management
• Identify risk which can deteriorate the performance of the process • Evaluate each risk for probability of happening and impact on the process • Make action plan to limit or reduce the risk

First Edition

347

© All Rights Reserved TreQna 2005

Treqna Base Manual Risk Management • Identify risks in the following categories: • Technology related • Financial • Decision making • Business • Use FMEA as a risk assessment tool.1) First Edition 348 © All Rights Reserved TreQna 2005 . (Refer to M.

Treqna Base Manual Mistake Proofing • A technique to eliminate errors • Also known as Poka Yoke (Japanese) • Proactive approach to stop errors from happening • Example: • Fuel reserve indicator in vehicles • Shape of SIM card allows only one way to place it in cell First Edition 349 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Mistake Proofing • Helps to limit X’s within the tolerance so that the process performance does not go out of control • Identifies the loopholes and helps implementing proactive steps for warning or avoidance of situation which can lead to X’s stepping out of tolerance First Edition 350 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Mistake Proofing • How to mistake proof: • Understand origin of defects • Identify sources of errors resulting in defects • Identify process steps where mistake proofing can be done • Redesign process steps such that they are resistant to errors. First Edition 351 © All Rights Reserved TreQna 2005 . The options to make errors are reduced.

Treqna Base Manual Errors • Sources of errors: • Process Variation • Procedural incorrectness • Measurement accuracy • Human errors • One can use fishbone analysis to brain storm and identify the sources of errors First Edition 352 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Control Plan • Control plan for a project should include following things: • CTQ • Input and out variables • Specifications and tolerances • Control measures and tools • Monitoring and sampling plan First Edition 353 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Control Plan Sheet • Following is a sample control plane sheet Control Plan Date Created: Date Revised: Data Control Reaction Sample Collection Tool Plan Size Frequency Measure Process Step Input/Output Specifications Target Project Y. X’s to be measured Process step where measured Whether input or output measure Target for the measure Specifications and tolerances How often the data is to be collected Actions to be taken when the measure goes out of control Sample size to be collected Control charts. Mistake proofing other tools First Edition 354 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Who owns it? Responsibility plan: • Who owns the activity? • Who is the back up? • Define procedures and operational definitions • Escalation matrix: • Escalation matrix as a part of reaction plan First Edition 355 © All Rights Reserved TreQna 2005 .

Treqna Base Manual Control .Summary • Statistical process control • Eight tests for control charts • Variable and Attribute control charts • Control limits vs. Specification limits • Risk management and FMEA • Mistake proofing • Control plan • Elements of control plan • Additional details First Edition 356 © All Rights Reserved TreQna 2005 .

Sign up to vote on this title
UsefulNot useful