This is a sound textbook for Information Technology and MIS undergraduate students, and MBA graduate students and all professionals looking to grasp a fundamental understanding of information quality. The authors performed an extensive literature search to determine the Fundamental Topics of Data Quality in Information Systems. They reviewed these topics via a survey of data quality experts at the International Conference on Information Quality held at MIT. The concept of data quality is assuming increased importance. Poor data quality affects operational, tactical and strategic decision-making, and yet error rates of up to 70%, with 30% typical are found in practice (Redman). Data that is deficient leads to misinformed people, who in turn make bad decisions. Poor quality data impedes activities such as re-engineering business processes and implementing business strategies. Poor data quality has contributed to major disasters in the federal government, NASA, Information Systems, Federal Bureau of Investigation, and most busineses. The diverse uses of data and the increased sharing of data that has arisen as a result of the widespread introduction of data warehouses have exacerbated deficiencies with the quality of data (Ballou). In addition, up to half the cost of creating a data warehouse is attributable to poor data quality. The management of data quality so as to ensure the quality of information products is examined in Wang. The purpose of this book is to alert our IT-MIS-Business professionals to the pervasiveness and criticality of data quality problems. The secondary agenda is to begin to arm the students with approaches and the commitment to overcome these problems. The current authors have a combined list of over 200 published papers on data and information quality.

Introduction to INFORMATION QUALITY

By Craig Fisher Eitel Lauría Shobha Chengalur-Smith Richard Wang

AuthorHouse

Copyright © 2011 Craig Fisher, Ph.D., MARIST College, Eitel Lauría, Ph.D., MARIST College, Shobha Chengalur-Smith, Ph.D., SUNY Albany, and Richard Wang, Ph.D., MIT
All right reserved.
ISBN: 978-1-4685-3028-5

Contents

Preface............................................................................xiChapter 1..........................................................................1Information Systems and Impacts of Poor-Quality Data...............................1Chapter 2..........................................................................27Total Quality Management...........................................................27Chapter 3..........................................................................49The Multiple Dimensions of Information Quality.....................................49Chapter 4..........................................................................79Information Products and Total Data Quality Management.............................79Chapter 5..........................................................................108Statistics.........................................................................108Chapter 6..........................................................................153Controlling Information Product Quality............................................153Chapter 7..........................................................................178Measuring and Tools for Assessing Data and Information Quality.....................178Chapter 8..........................................................................215Decision Support and Business Intelligence.........................................215

Chapter One

Information Systems and Impacts of Poor-Quality Data

Information systems (IS) help organizations get, process, store, manage and report information so that everyone can do their jobs as efficiently and effectively as possible. IS organizations exist in every large corporation, government entity, and educational institution in the United States. Although these systems have had many names—electronic data processing, data processing, information systems, information services, information processing, data centers, information technology—the key words are data and information. The purpose of IS has always been to provide data and information to those who need it.

However, even after 40 to 50 years of IS experience, organizations are often less—rather than more—satisfied with the information they are getting. Experience in developing IS includes many years of improving techniques and processes to assure quality results. Some techniques are: structured programming, data validity checks, check digits, test and conversion strategies, inspections, and computer-assisted software engineering tools. Processes include application development methodologies often modeled after engineering development processes, joint application development, various levels of integration tests and user acceptance tests, rapid application development (RAD) methodology, use of various models such as data flow diagrams, and entity relationship diagrams. Object-oriented approaches, including a unified modeling language, are now being used. These can be found in modern textbooks on systems analysis and design or management information systems.

Even with all of the improvements in processes and techniques, complaints of poor-quality data and bad information are frequent. Many of the processes and procedures mentioned in the previous paragraph take time and are full of errors. As a result, people bypass what seems to them a bureaucratic, time-consuming process. For example, the rapid development processes called RAD allow the developer to skip many of the bureaucratic steps. Another factor that contributes to poor-quality data is end user computing (EUC). Although one of the primary goals of modern Windows-type facilities is to make it easy for end users to quickly build systems, this does not guarantee quality. Are organizations relying too much on EUC? Compare the training of an end user with an IS student. The IS student takes at least one semester of database development in which normalization and referential integrity are covered in detail. Many end users are not even aware of these data quality (DQ) techniques.

End-user-created databases are generally not easily integrated into the corporate IS. Data definitions, the abbreviations and names of the data, the system architecture, the software, and the interfaces involved vary so much from user to user and department to department, that any attempt to create an integrated database or data warehouse seldom ends satisfactorily. The existence of good development procedures is not effective if people can simply bypass them, whether through ignorance or design. An overall approach which includes treating information as products, modeling information management systems, building information product maps, and applying total quality management (TQM) approaches, will be discussed in later chapters.

It could be that modern day information systems are so complex and cumbersome that accidents and errors cannot be avoided. For example, Fleet Bank launched a $38-million customer relations management project to pull together customer information from 66 source systems. The project failed in less than three years because it was too difficult and time consuming to understand, reconcile, and integrate data from so many sources.

Home Depot's IS supports over a million products in approximately 2,300 stores with many databases. Home Depot is attempting to tie together thousands of its software applications, stores, and systems. Information systems consist of people, procedures, databases, systems hardware and software, and computer application programs. The complexity of the systems, and of its components, may lead to DQ problems. Clearly, a very comprehensive approach to quality is necessary, and this will be covered in the remaining chapters.

Information Systems

This chapter is focused on examples of problems and their effects on people, organizations, the Federal Government, and society in general. We will summarize the effects after we distinguish between data and information.

Data Versus Information

There is an old saying that "one person's data is another person's information." Several textbooks distinguish between data and information. Data is defined as isolated facts devoid of meaning. Information is defined as processed data that has meaning because of relationships established with other data. For example, if a retailer sold 10 dresses, that might be an interesting fact but doesn't have much meaning by itself. However, if that retailer sold 100 dresses yesterday and 10 today, then more meaning can be derived. Meaning is enhanced by establishing relationships with other data. Additional data might include comparisons of this retailer's sales to those of a competitor; or comparisons of sales over intervals, such as sales per month, per quarter, or per year. It might also be important to know how many dresses were sold during a sales promotion or during an employee incentive program to sell, for example, 100 dresses. Another view might be the rate of sales. If sales were very high—10 dresses in 30 minutes—that might make the manager take action that would not be taken if the rate of sales was very low. Depending on the users and their purpose, the previous examples might be indicative of either data or information. The floor manager may want to know how many dresses are on display to determine if more should be brought to the floor. The store manager might want to know average daily sales to compare with the sales goals and competitors' sales.

Here is a manufacturing example. The number of parts in a stock room is a critical piece of information and must be accurate. The shop manager needs to know that detail to determine if he can build a product at 10:00 A.M. The shop manager's information, however, is simply data to an executive, since the executive needs to combine it with many other indicators to determine how well the plant is running overall relative to his goals and his competitors' performance. Since there are many levels and interpretations of the differences between data and information, we will treat data and information interchangeably. The context will make it clear.

The effects of poor-quality data and information are more than just a nuisance; they inhibit people at all levels in all types of organizations from performing their jobs properly. Poor-quality data is prevalent in both the private and public sectors.

Data quality is one of the most critical problems facing organizations today. As executives become more dependent on IS to fulfill their missions, DQ becomes an even bigger issue. Poor-quality data is pervasive and costly. "There is strong evidence that data stored in organizational databases are neither entirely accurate nor complete" (p. 169).

Estimates of DQ problems range widely, but none are small. "Current data quality problems cost U. S. businesses more than 600 billion dollars a year" (p. 99). In industry, error rates as high as 75% are often reported, while error rates to 30% are typical. Of the data in mission-critical databases, 1% to 10% may be inaccurate. More than 60% of surveyed firms had problems with DQ. In one survey, 70% of the respondents reported their jobs had been interrupted at least once by poor-quality data, 32% experienced inaccurate data entry, 25% reported incomplete data entry, 69% described the overall quality of their data as unacceptable, and 44% had no system in place to check the quality of their data.

In the summer of 2002, a couple checked into a lodge and were given a room key. When they arrived at their room, they found it was already occupied. The clerk who assigned the room had simply grabbed a key off the row of hooks and assigned that room number. This type of error can also happen to anyone using a computer to enter or extract data. Any interruption, such as a telephone call or someone dropping by the office, could cause a momentary loss of concentration, leading to an error with a costly or sometimes embarrassing effect.

Because of the nature of the system, some incidents of DQ problems may be systemic or unavoidable. Quality problems caused by human error are certainly avoidable, but some data problems fall in between these two extremes. The user may not have made an obvious error, and the system may not have errors due to system design. It is these not-so-obvious problems that lead to many of the undetected problems that plague systems end users. Much of the remainder of this book is dedicated to exploring those errors and how to avoid them, but for now we continue with the problems and effects of poor-quality data.

Mandating Quality Information

Can We legislate Good-Quality Information?

As a prelude to providing discussion of the effects of poor-quality data, let us consider the gravity of the situation. The Federal Government has now made poor-quality data illegal. Did you ever believe that the government would have to pass a law requiringits agenciesto give out only good-quality data? Just a little reflection makes the situation seem absurd. One would expect that in the course of performing their tasks, those agencies would deliver good-quality information. However, industry and business leaders believed that agencies were disseminating information that forced their organizations to implement actions they weren't really required to take. Therefore, they pressed for "regulation of information."

Section 515 of the Treasury and General Government Appropriations Act of 2001 (PL106-544, H. R. 5658) directs the Office of Management and Budget (OMB) to issue government-wide guidelines that "provide policy and procedural guidance to Federal agencies for ensuring and maximizing the quality, objectivity, utility and integrity of information (including statistical information) disseminated by Federal agencies.".

During 2002, the year following approval of Section 515, generally known as the information quality act, federal agencies issued hundreds of pages of agency-specific guidelines that included "administrative mechanisms allowing affected persons to seek and obtain correction of information maintained and disseminated by the agency that does not comply with the OMB guidelines.". So we now live in an age where Federal laws are required to ensure that good-quality information is disseminated. In spite of this, good-quality information has not been achieved. There has been enormous public pressure for improved information because of situations like 9-11, terrorist tracking, the Iraq war, information warfare, downsizing, dependencies on electronic information, errors in everyday databases (such as billing files and credit ratings), corporate databases, NASA programs, and military problems. Bombing the wrong targets, the right targets at the wrong time, and problems such as those that occurred on the USS Vincennes are just a few examples of the latter.

Information Warfare

Threat exposure is increasing. Information systems and their databases are becoming more prolific every year, making them a prime target. Digital assets include everything from financial statements to patient records to proprietary algorithms. For example, 93% percent of corporate documents are created electronically. Every year, billions of email messages are sent worldwide. U.S. consumers spend billions of dollars over the Internet, and networked business-to-business (B2B) transactions are in the trillions of dollars.

A growing concern is that we do not know the extent and impact of poor-quality data. The FBI says that approximately 10,000 cyber-crime complaints were filed in the year 2000, in which losses totaled over $265 billion. But reported losses do not begin to tell the story. Most cyber crime goes unreported and undetected. Because of incomplete data, organizations do not know the actual extent of their losses and therefore cannot perform a proper evaluation to determine how much they should invest in security systems. Information is becoming increasingly important, which makes our exposure to poor-quality information even more critical.

Information is now a primary instrument of national power, and information warfare is an emerging area for conflict. Information is our most powerful weapon to secure our nation and combat terrorism. The office of the Assistant Secretary of Defense said that information warfare is "action taken to achieve information superiority by affecting adversary information, information-based processes, information systems and computer-based networks while defending one's own...." However, simply gathering huge amounts of information is not the answer. Although law enforcements agencies gather data, they do not always properly manage, analyze, and share information.

On November 25, 2002, President Bush signed the "Homeland Security Act of 2002" into law. This law represents a concerted effort by the Federal Government to prevent terrorist attacks in the United States and to reduce vulnerability and minimize damage from attacks should they occur. The newly formed Department of Homeland Security has made information a central element of its mission. One branch of the new department is called the "Information Analysis and Infrastructure Protection Directorate." It focuses on improving access—one dimension of DQ—to information that may help detect and disrupt terrorist activity. The directorate is also responsible for protecting that information. There is a strong need to improve information sharing, and the government has designated $4.5 billion to address interoperability and sharing issues.

Conventional approaches to defensive information warfare focus primarily on physical security, electronic measures, and encryption. These areas control only a small subset of the range of impacts that information warfare attacks can have on information DQ. For example, subtle changes to information timeliness, accuracy, and believability can have significant impact on military command and control, yet pass undetected through standard security safeguards.

Compared to physical asset systems, information systems have more layers of abstractions and many interdependencies between nodes. Some attacks on IS may be easily detected. However, many types of attacks are very difficult to detect until the bad effects have already rippled through the system.

Healthcare and Medical Profession

"In recent years, the importance of good data has grown exponentially as most of the industrialized world has adopted data-driven, or evidence-based, health care which uses comparative data methods for clinical treatment methods" (p. 85). Evidence-based decision making relies on consistent classification and reporting of medical information. A national survey of all accredited U.S. medical records managers examined variations in the classification of health care. In the healthcare industry, significant variation in the application of classification rules results in much misreporting of patient data. A recent study indicates that impacts occur in five areas.

• Over-reimbursement: errors caused over-reimbursement of billing claims in about 5% of records (p. 86).

• Under-reimbursement: Overall, 16.8% of respondents said that more than 5% of their records had significant under-reimbursement errors (p. 86).

(Continues...)

Excerpted from Introduction to INFORMATION QUALITYby Craig Fisher Eitel Lauría Shobha Chengalur-Smith Richard Wang Copyright © 2011 by Craig Fisher, Ph.D., MARIST College, Eitel Lauría, Ph.D., MARIST College, Shobha Chengalur-Smith, Ph.D., SUNY Albany, and Richard Wang, Ph.D., MIT. Excerpted by permission of AuthorHouse. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.