Unit IV

Software Quality Improvement Techniques Software systems have recently proliferated greatly and become a pervasive presence both in the life of individuals and in society at large. Accompanying the expansion of software use, it is essential to ensure the high quality of software. 1. Sufficient software testing, 2. Verification, and . !ault elimination are the most important techniques for improving software quality. "here are many methods for quality improvement. "hese cover product improvement, process improvement and people based improvement. #n the following list are methods of quality management and techniques that incorporate and drive quality improvement$ 1. #S% &''($2'') * guidelines for performance improvement. 2. #S% 1++'(,($ 2''+ * information technology * process assessment * -art ($ .uidance on use for process improvement and process capability determination. . /!0 * quality function deployment, also 1nown as the house of quality approach. (. 2ai3en * 4apanese for change for the better5 the common 6nglish term is continuous improvement. +. 7ero 0efect -rogram * created by 869 9orporation of 4apan, based upon statistical process control and one of the inputs for the inventors of Six Sigma. :. Six Sigma * :;, Six Sigma combines established methods such as statistical process control, design of experiments and !<6A in an overall framewor1. =. -09A * plan, do, chec1, act cycle for quality control purposes. >Six Sigma?s 0<A#9 method >define, measure, analy3e, improve, control@ may be viewed as a particular implementation of this.@ ). /uality circle * a group >people oriented@ approach to improvement. &. "aguchi methods * statistical oriented methods including quality robustness, quality loss function, and target specifications. 1'. "he "oyota -roduction System * rewor1ed in the west into lean manufacturing. 11. 2ansei 6ngineering * an approach that focuses on capturing customer emotional feedbac1 about products to drive improvement. 12. "/< * total quality management is a management strategy aimed at embedding awareness of quality in all organi3ational processes. !irst promoted in 4apan with the 0eming pri3e which was adopted and adapted in ASA as the <alcolm Baldrige 8ational /uality Award and in 6urope as the 6uropean !oundation for /uality <anagement award >each with their own variations@. 1 . "C#7 * meaning Dtheory of inventive problem solvingD 1(. B-C * business process reengineering, a management approach aiming at ?clean slate? improvements >"hat is, ignoring existing practices@. 1+. %/< * %bEect,oriented /uality <anagement, a model for quality management.
F G

-roponents of each approach have sought to improve them as well as apply them for small, medium and large gains. Simple one is -rocess Approach, which forms the basis of #S% &''1$2'') /uality <anagement System standards, duly driven from the ?6ight principles of /uality management, process approach being one of them. "he competencies which were hitherto rated as being smaller, are better recogni3ed and now acclaimed to be more potent and fruitfulD. "he more complex /uality improvement tools are tailored for enterprise types not originally targeted. !or example, Six Sigma was designed for manufacturing but has spread to service enterprises. 6ach of these approaches and methods has met with success but also with failures. Some of the common differentiators between success and failure include commitment, 1nowledge and expertise to guide improvement, scope of changeHimprovement desired >Big Bang type changes tend to fail more often compared to smaller changes@ and adaption to enterprise cultures. !or example, quality circles do not wor1 well in every enterprise >and are even discouraged by some managers@, and relatively few "/<, participating enterprises have won the national quality awards. "here have been well publici3ed failures of B-C, as well as Six Sigma. 6nterprises therefore need to consider carefully which quality improvement methods to adopt, and certainly should not adopt all those listed here. #t is important not to underestimate the people factors, such as culture, in selecting a quality improvement approach. Any improvement >change@ ta1es time to implement, gain acceptance and stabili3e as accepted practice. #mprovement must allow pauses between implementing new changes so that the change is stabili3ed and assessed as a real improvement, before the next improvement is made >hence continual improvement, not continuous improvement@. #mprovements that change the culture ta1e longer as they have to overcome greater resistance to change. #t is easier and often more effective to wor1 within the existing cultural boundaries and ma1e small improvements >that is Kaizen@ than to ma1e maEor transformational changes. Ase of 2ai3en in 4apan was a maEor reason for the creation of 4apanese industrial and economic strength. %n the other hand, transformational change wor1s best when an enterprise faces a crisis and needs to ma1e maEor changes in order to survive. #n 4apan, the land of 2ai3en, 9arlos .hosn led a transformational change at 8issan <otor 9ompany which was in a financial and operational crisis. Iell organi3ed quality improvement programs ta1e all these factors into account when selecting the quality improvement methods.

***

Statistical Software Quality Assurance Techniques Ihat are the steps implied by statistical quality assuranceJ 1. #nformation about sHw defects, 2. An attempt is made to trace each defect to its underlying cause, . <ove to correct the problems that have caused the defects. Three ain Aspects of Statistical Quality Assurance Statistical quality control and assurance >S/9A@ consists of three main parts. 1. Statistical -rocess 9ontrol >S-9@ 2. Acceptance Sampling >AS@ . 0esign of 6xperiments >0%6@ to improve quality and productivity. Some terminolo!y" #n both S-9 and AS, samples are drawn. Although the terminology is not uniform or universsal, we refer to the individuals sampled as units and to different products in the line of goods made by a company as items. So we might say, DIe too1 a sample of : units of #tem A and a sample of ) units of #tem B.D Samplin! from a population vs" samplin! from a process" !rom the samples, decisions are made about the population or process generating the items ma1ing up the samples.

!or AS, there is a population of already produced items, called a lot. !or S-9, there is no existing population5 rather, we are examining the process generating the samples.

(1) Statistical Process Control

<ethods of statistical process control, especially the control chart, are used to try to determine whether a process is in statistical control. A process is in statistical control if it is on target and the variation is what would be expected in random observations of a stable process. Ihen this is the case, the variation among items is due to chance and there is no reason to see1 special, assignable causes of variation. %n the other hand, when the process is out of control, it may be possible to locate some specific cause for the variation, remove the cause, and improve the future performance of the process. #f a process is in control and yet its performance is unsatisfactory, the process itself must be changed. !or example, better technology must be found to produce a process with less variability.

(2) Acceptance Sampling

"he obEect of acceptance sampling is to evaluate each lot of material. "he lot is a population, and the situation is the same as hypothesis testing based on a sample from a population. Kere you should thin1 in terms of estimating the mean of each separate lotL, whereas in process control you thin1 in terms of estimating the mean of the processL. #n acceptance sampling you inspect a sample from the lot and use definite statistical

standards to infer the quality of the lot from the quality of the sample. "he standards are set according to what is required of the product. #n process control, depending upon the application, there may be a standard, or you may Eust be monitoring the process to see that there is no large deviation from what is usual and customary.

#ro$ucer%s ris& an$ consumer%s ris&

9onsider the situation where each unit is either defective >not in conformance@ or o1ay. Sampling will be used to estimate the proportion defective. Statistical methods are used to set a sample size and an acceptance num'er" "he sample si3e is the number of units in the sample. "he acceptance number is the largest number of defectives in the sample for which we still will accept the lot. #f the number of defectives in the sample is greater than the acceptance number, the lot from which the sample came is reEected. "his entails two ris1s. "here is the ris1 that a good lot will yield a bad sample and hence lead to reEection of the lot even though it is good. And there is the ris1 that a bad lot will yield a good sample and hence lead to acceptance of the lot, even though it is bad. "he first ris1 is called the producer's risk >or seller?s ris1@. "he second is the consumer's risk >or buyer?s ris1@. #n statistics these two types of error are called Type I and Type II errors.

(3) Design of Experiments

Approximately the last third of the course is focused on 0%6, the design of experiments to achieve /uality Assurance through #mprovement by -lanned 6xperimentation. #n particular, we will study the Analysis of Variance. <ethods 1. 9ause 6ffect 0iagram 2. !lowchart . -areto 9hart (. Cun 9hart +. Kistogram :. Scatter 0iagram =. 9ontrol 9hart MMM

Quality Appraisal #ro!rams "he Appraisal -rogram oversees the quality and consistency of the Software ?s process appraisal technology and encourages its effective use. #ts four main functions include communications to the appraisal community5 appraisal quality control5 training, authori3ing, certifying, and providing resources for Nead Appraisers and "eam Neaders5 and monitoring and reporting appraisal results. "hrough the S6# Appraisal -rogram, the highest quality candidates are selected and trained as Nead Appraisers. %ngoing training and resources are provided for Nead Appraisers. Appraisal (lasses !or benchmar1ing against other organi3ations, appraisals must result in consistent ratings. "he S6# has developed a document to assist in identifying or developing appraisal methods that are compatible with the 9<<# -roduct Suite. "his document is the Appraisal Cequirements for 9<<# >AC9@. S)I Appraisal (lasses "he AC9 describes a full benchmar1ing class of appraisal as 9lass A. %ther 9<<#, based appraisal methods might be more appropriate for a given set of sponsor needs, including self,assessments, initial appraisals, quic1,loo1 or mini,appraisals, incremental appraisals, and external appraisals. "hus, a particular appraisal method is declared an AC9 9lass A, B, or 9 appraisal method. "his designation implies the sets of AC9 requirements that the method developer has addressed when designing the method. "he S9A<-# family of appraisals includes 9lass A, B, and 9 appraisal methods. S(A #I A is the most rigorous method and the only method that can result in a rating. S(A #I * provides options in model scope, but the characteri3ation of practices is fixed to one scale and is performed on implemented practices. S(A #I ( provides a wide range of options, including characteri3ation of planned approaches to process implementation according to a scale defined by the user. Asing S9A<-# B, every practice in the appraisal scope is characteri3ed on a three point scale indicating the ris1 of 9<<# goal satisfaction if the observed practices were deployed across the organi3ational unit. <odel scope is not limited to the -rocess Areas but could include sets of related practices.

S9A<-# 9 can be scoped at any level of granularity and the scale can be tailored to the appraisal obEectives, which might include the fidelity of observed practices to modelHgoal achievement or the return on investment to the organi3ation from implementing practices. Celiability, rigor, and cost might go down from A to B to 9, but ris1 might go up. (haracteristics of Appraisal (lasses (haracteristic Amount of obEective evidence Catings generated Cesource needs "eam si3e (lass A Kigh Oes Kigh Narge (lass * <edium 8o <edium <edium (lass ( Now 8o Now Small

aterials Appraisal +equirements for ( I ,A+(-. Version /"0 "he Appraisal Cequirements for 9<<# >AC9@ V1.2 defines the requirements considered essential to appraisal methods intended for use with 9apability <aturity <odel #ntegration >9<<#@ models. #n addition, a set of appraisal classes is defined, based on typical applications of appraisal methods. "hese classes are intended primarily for developers of appraisal methods to use with 9<<# capability models in the context of the 9<<# -roduct Suite. Appraisal methods, as used in this document, may be applied for different purposes, including assessments for internal process improvement and capability evaluations for supplier selection and process monitoring. "his document defines the requirements for such methods, but not necessarily the conditions or constraints under which they might be applied. "he approach employed to provide guidance to appraisal method developers is to define a class of typical applications of appraisal methods >which are based on years of experience in the process improvement community@ called appraisal method classes. Cequirements are then allocated to each class as appropriate based on the attributes associated with that class. "hus, a particular appraisal method may be declared to be an AC9 9lass A, B, or 9 appraisal method. "his designation implies the sets of AC9 requirements that the method developer has addressed when designing the method. S(A #I 1ea$ Appraiser *o$y of Knowle$!e ,S1A *2K"he Standard 9<<# Appraisal <ethod for -rocess #mprovement >S9A<-#@ is designed to provide benchmar1 quality ratings relative to 9apability <aturity <odel #ntegration >9<<#@ models. "he role of the S9A<-# Nead Appraiser, who conducts the S9A<-# process, has developed into a distinct profession that encompasses a wide and deep array of competencies. "he S9A<-# Nead Appraiser Body of 2nowledge >SNA B%2@ provides a multi, dimensional view of the competencies and associated s1ills that are needed to be a

successful S9A<-# Nead Appraiser. "he SNA B%2 is a driver for the S6# Appraisal -rogram?s approach to ensure quality of S9A<-# appraisals. #t establishes a comprehensive basis for curriculum, training, certification, observation, and quality assurance programs. S(A #I etho$ 3efinition 3ocument , 33"he Standard 9<<# Appraisal <ethod for -rocess #mprovement >S9A<-#@ is designed to provide benchmar1 quality ratings relative to 9apability <aturity <odel #ntegration >9<<#@ models. #t is applicable to a wide range of appraisal usage modes, including both internal process improvement and external capability determinations. S9A<-# satisfies all of the Appraisal Cequirements for 9<<# >AC9@ requirements for a 9lass A appraisal method. "he S9A<-# <ethod 0efinition 0ocument describes the requirements, activities, and practices associated with each of the processes that compose the S9A<-# method. #t is intended to be one of the elements of the infrastructure within which S9A<-# Nead Appraisers conduct a S9A<-# appraisal. -recise listings of required practices, parameters, and variation limits, as well as optional practices and guidance for enacting the method, are covered. An overview of the method?s context, concepts, and architecture is also provided. Implementin! a 4oal5Ali!ne$ )mployee Appraisal #ro!ram A web,based performance management system to help the company more strategically manage its goal,based employee performance appraisal process. 1. Align employee goals with overall organi3ational obEectives5 2. <onitor the appraisal program and provide regular updates to employees5 and . #ncrease on,time completion within a shorter annual cycle. ***