Data Standardization Means Quality Information from Distributed Systems Copyright 1992 CAUSE From _CAUSE/EFFECT_ Volume 15, Number 4, Winter 1992. Permission to copy or disseminate all or part of this material is granted provided that the copies are not made or distributed for commercial advantage, the CAUSE copyright and its date appear,and notice is given that copying is by permission of CAUSE, the association for managing and using information resources in higher education. To disseminate otherwise, or to republish, requires written permission.For further information, contact CAUSE, 4840 Pearl East Circle, Suite 302E, Boulder, CO 80301, 303-449-4430, e-mail info@CAUSE.colorado.edu DATA STANDARDIZATION MEANS QUALITY INFORMATION FROM DISTRIBUTED SYSTEMS by Lore Balkan, Gerry McLaughlin, and Bruce Harper ************************************************************************ Lore Balkan is Data Standards Project Coordinator in the Institutional Research and Planning Analysis office at Virginia Tech, a project addressing data management quality issues. An active participant in CAUSE activities, she is also Region 18 president of the Data Processing Management Association (DPMA). Gerry McLaughlin is Director of Virginia Tech's Institutional Research and Planning Analysis office which provides information and data management support for the University administration. He was a co- recipient of the 1989 CAUSE/EFFECT Contributor of the Year Award. Bruce Harper is a member of the Standardization of Data Elements and Codes Project team at Virginia Tech as well as the records management department, where he provides computer support. He is also responsible for the maintenance and update of University policies and procedures. ************************************************************************ ABSTRACT: Increasing campus-wide interest in accessing and using data from distributed databases prompted a close look at data management and administration at Virginia Tech. To increase the value of administrative information for University-wide decision-making, the office of Institutional Research and Planning Analysis has undertaken a project to standardize data elements and codes across all administrative systems and to institutionalize an ongoing standardization process to continually improve the quality of data. At Virginia Tech, administrative systems have always emphasized online user access, primarily to support sourcepoint data capture for administrative operational functions. System support groups exist for each major administrative area. Management information is supplied from a series of extracts produced by these system support groups from each operational system. The office of Institutional Research and Planning Analysis is a primary user of these extracts, integrating data to create merged extracts for institutional reporting. However, there is increasing interest across the University in accessing and using this extracted, merged, and downloaded data for decision support activities. The inherent challenge of integrating diverse and non-standardized data from a variety of systems is magnified considerably as demand continually broadens both the customer base and the scope of information desired. This phenomenon has led to the Institutional Research office undertaking a special project to improve the quality of data through the standardization of data elements and codes across all administrative systems. This article begins with a brief discussion of the data administration and data management principles that formed the basis for our data standardization project, then describes the components of the project and our progress to date, including lessons we have learned. From data administration to data management Standards are agreements on language and procedure to enable clear, complete, and efficient communication. Many IS organizations created a central data administration department in the early seventies to help discern and enforce minimal standards and to ensure that systems could work cooperatively. Since most of these early systems were developed internally by a systems development group and usually for one database management system, the focus was on securing, cataloging, and standardizing database definitions. A data dictionary was often closely coupled and integrated with the particular database management system in use. Today we are more likely to hear about information resource management than data administration. The function now goes well beyond the initial database support function of most early data administration departments. The following concise and comprehensive data administration mission statement proposed by the Data Administration Standards and Procedures Working Group of the Data Administration Management Association (DAMA) clearly has relevance for the information management issues we face today: * To combine activities, standard methods, human resources, and technology for the central planning, documentation, and management of data from the perspective of the meaning and value to the organization as a whole. * To increase system effectiveness by controlling data through uniformity and standardization of data elements, database construction, accessibility procedures, system communication, maintenance, and control. * To provide guidance for planning, managing, and sharing of data and information effectively and efficiently in automated information systems. This mission is significantly more expansive than earlier mission statements. It describes an essential management function to optimize the information systems resources, even as technology rapidly changes, users become more diverse and increase in numbers, and the information support environment continues to become more distributed and complex. It recognizes that while data administration must be acutely aware of evolving technology and must be involved in the resulting change management, the data administration function itself is not driven by technology. In fact, the reverse is true: effective data administration focuses on building a stable and quality information resource that makes it possible for an organization to quickly respond to and take advantage of continuous technological innovation. Moreover, with people throughout the institution creating, managing, and disseminating electronic data, data quality must receive attention throughout the organization. This attention must go well beyond control activity within the information systems organization and be understood and embraced by all of an institution's management. The term _data management_ recognizes the fact that managers and technicians alike throughout our organizations must manage data and attend to the quality issues. In discussing structures for information resource governance, Miselis makes the important point that "since information is an institutional resource that is developed and used campus-wide, the management structure charged with ensuring that there is effective and efficient use of computerized information should also be campus-wide."[1] At Virginia Tech, guidelines for administrative information resource management have been applied as University policy, providing the basis and describing campus-wide roles for managing data in a distributed environment: The University is the _data owner_ of all University administrative data. University officials, such as the Controller, the Associate Vice President for Human Resources, and the Registrar, are _data custodians_ for data in their functional area. _Data stewards_ are usually delegated the responsibility for daily data maintenance and dissemination activities by data custodians. Data users are individuals who need University data in order to perform their assigned duties and are therefore authorized access by data custodians. The function of applying formal guidelines and tools to manage the University's information resource is termed _data administration_ and is performed cooperatively and collectively by custodians, stewards, and users.[2] The success or failure of data management depends on management's understanding, support, and active participation in these distributed data administration roles. Failure to obtain participation will cause the organization to raise barriers, such as the following: Those who in the past have written administrative applications tend to be set in their ways of operation. Hardware and software vendors try to sell what they have. Operational personnel have numerous individual pieces of automated support that they want to keep. Managers are concerned more about making the next deadline than about improving "their" data for someone else's needs. Everyone is already overwhelmed with just trying to meet existing commitments with little or no time to scale another learning curve.[3] Clearly operational, managerial, and executive personnel must buy into the belief that improving data quality is worth the investment of time and money. This requires providing a product that has value and can be marketed. This product is an integrated toolset that has four sequential supporting parts: _people_ who perform activities; _activities_ that utilize data; _data_ that are manipulated using tools; and _tools_ that assist with creation, reference, update, and deletion of data.[4] The entire toolset needs to be engaged to ensure that the institution effectively manages its data and thus has quality information. The overall efficiency of this integrated toolset in terms of quality information systems can quickly justify the costs involved. In building a supportive distributive environment for data management at Virginia Tech, we have focused on data element definitions and code standardization. Standards can provide the firm foundation on which to build the integrated toolset for quality data management. The essence of standardization is the adoption of a common language that enables shared understanding and provides capability to integrate multiple data sources. As such, it is a never-ending process that continually improves the quality of the information resource. Standardization for quality The process of continually improving the quality of information begins with a set of values. Durell's "Ten Commandments of Data Administration Standards" provide an excellent perspective on these values (see last page).[5] The Standardization of Data Elements and Codes Project at Virginia Tech represents our second major stride for data management. It is a direct follow-up to the development of the guidelines for administrative information resource management mentioned earlier. Like the development of the guidelines, development of standards is a consensus-building process and, as such, requires a great deal of coordination, cross- functional interaction, and cooperation. The purpose of the standardization project is to increase the value of University administrative information by implementing standards for description, definition, and validation of administrative data. The goal is to create both the policy and the technical tools to obtain, store, and manipulate the administrative operational data for University-wide decision-making. While we focus on the tools, we do so in the context of the entire toolset. We want our standards to serve as a foundation for generalized access and use of information from the variety of administrative data sources by a broad base of users. Moreover, we want to institutionalize an ongoing standardization process to continually improve the quality of the data. We contend that standardized data are quality data with the following attributes: (1) assignment of data custodial responsibility; (2) audits for accuracy and measures for accountability; (3) systematic edit and validation; (4) clear and meaningful usage; (5) consistency over time; (6) cross-reference to all occurrences; and (7) accessibility. Each of these requires that the institution provide the following support: People and Activities 1. Data custodial accountability. 2. Data stewardship to ensure proper edit and ongoing completeness and accuracy. Data 3. A single official source of critical entities such as facility or department with the list of standard values for key attributes. 4. Standardized data descriptions, definitions, and documentation. Tools 5. Procedures to retain and successfully use historical data. 6. Query capability for users to identify appropriate data sources and procedures to address specific information needs. 7. Ready access to timely and properly secured data by trained users. The standardization project adopted a goal statement based on the Shewart Cycle[6] to address the above data quality issues. The premise of this cycle is that never-ending improvement requires movement through the steps of _Plan, Do, Check,_ and Act (PDCA). As a result of validating or checking the doing of prototype work, improvements become a new standard or act and one goes back to _plan_, looking for new opportunities to continually improve.[7] This PDCA paradigm is reflected in the four steps to meeting the project goal: To discover, define, document, and apply tools and techniques for standardizing University information by: Step 1: Identifying critical and key University entities and related data elements and codes (P) Step 2: Defining and documenting entities and related data elements and codes (D) Step 3: Measuring and verifying data and code quality and integrity (C) Step 4: Establishing an ongoing process of managing standardized entities in terms of data element edit, validation, update, alteration, audit, correction, and distribution (A) It is important to emphasize that the PDCA cycle is iterative. Moreover, for large problems or projects, multiple PDCA cycles must take place within a larger cycle. The four steps we have identified for standards are, in essence, a prototype methodology commonly used in research and development endeavors. Figure 1 shows the data standardization cycle feeding a larger quality information cycle. [FIGURE 1 NOT AVAILABLE IN ASCII TEXT VERSION] Establishing the standardization project In keeping with the PDCA cycle, the first six weeks of the standardization project were devoted to developing the project plan. This plan described the current environment, situation, and problems, and identified the four standardization steps. It also addressed project staffing with job descriptions, established milestones, and discussed the major project activities. Two individuals were assigned full-time to the project. However, the project plan clearly noted the need for additional technical support from computing center personnel and, even more critically, for project support from personnel in the various administrative offices. Cooperation of supervisors and staff in the operational areas is critical for making necessary modifications to local operational systems and adopting procedures to check and adapt to changing information requirements. It is basically their efforts that ensure enduring quality of the standardized data and set priorities for continuous improvement. It was especially important to develop project milestones to establish the project's credibility. The first three-month milestone involved completing a draft University standard to carry out the first two steps. For the second three-month milestone, a particular entity in one system area was selected to apply and refine the first two standards. The result was recommended procedures to continue standardization for other critical and key University entities and codes in other system areas. This milestone also involved work on the third and fourth steps in the cycle. University administrators, data custodians, data stewards, and information systems groups were called upon to help coordinate existing standards with the emerging University standard and to collaborate with prototype applications. The third milestone identified in the project plan was to refine the draft University standard for the third and fourth steps. Again, this refinement involved establishing recommended procedures to continue standardization beyond the prototype. Other entities in other system areas would begin to be identified and prioritized for standardization. The final milestone projected forward an additional six months. This milestone anticipated the promulgation of standardization results by providing user interfaces and access to standard definitions and code lists using available tools such as data dictionaries, relational databases, and an online query to extracts. It is only at this point that the results of the standardization project take the form of an obvious product. This milestone also projected the establishment of a recognized data management function to continue standardization and coordinate data quality initiatives with the various data custodians and a user group for each system area. Standardization activities The project plan received endorsement when it was distributed and presented to the University's executive management. Figure 2 shows the model, proposed in the plan, where data from operating systems would be merged together into the Administrative University Data Base that would then support diverse users. A key component of the planning model was the repository or the dictionary of information about information. One critical University entity, facility, stood out as an obvious early candidate for standardization, in part because it was relatively limited in scope. Additionally, facilities data had already been recognized as an area in need of improvement and therefore had high visibility. There was agreement that it should be the first entity targeted for standardization. We proceeded to identify major project activity areas as smaller, manageable "chunks" and prepared a plan for each. These activity areas each involve engaging or fortifying some part of the integrated toolset--people, activities, data, and tools. [FIGURE 2 NOT AVAILABLE IN ASCII TEXT VERSION] Work with formal and informal University groups (people) This activity involves engaging formal and informal groups to provide insight and feedback to the standardization process and to serve as a vehicle for promoting never-ending data quality improvement. The following efforts have been met with encouragement, enthusiasm, and responsiveness: * periodic progress reports to an informal management group; * meetings with a data steward group to monitor and review project developments; * assembly of a facilities focus group to assist with analysis and implementation of standards for the facility entity; * coordination with census file teams to standardize the census point-in-time snapshot process and data; * participation with the human resource system requirements team on a reengineering project; * interaction with the information systems departments regarding information dissemination strategies; and * roundtable discussions with the Administrative Systems Users Group (ASUG) to review project strategy. Additionally, we meet periodically with data custodians and always include them when we have major meetings with their data stewards. Develop the process and procedures for standardization (activities) The standardization process is embodied in the four steps of the project goal. As the project progresses, procedures to address each step are documented. A "living" document is created and modified as standardization procedures are discovered so they can later be repeated for other critical and key University entities. The analysis process for standardization is broken down into decision points with checklists and criteria for evaluation. For identification of the critical and key University entities, the first of the four steps, we rely heavily on the already approved _Guidelines for University Administrative Information Resource Management_. The _Guidelines_ list the criteria for data inclusion in the logical Administrative University Data Base (AUDB): * It is relevant to planning, managing, operating, or auditing major administrative functions. * It is referenced or required for use by more than one organizational unit. Data elements used internally by a single department or office are not typically part of the AUDB. * It is included in an official University administrative report. * It is used to derive an element that meets the criteria above. As an entity is standardized, an official source or "entity master table" is either identified or created. At minimum, it must contain: (1) key data elements--those variables that provide validation and translate capability, and (2) AUDB data elements--those variables that are required to answer University-wide questions about an entity that should be generally available to management, possibly via extracts. There are four key data elements that should have an official set of valid values for any standardized entity. They are: (1) a standard coded representation across data sources; (2) a standard long name; (3) a standard short name; and (4) a standard abbreviation. One or more of these four key data elements should be included as an attribute of the standardized entity wherever it appears, thus paving the way to integrate information across distributed systems. Prior to creating a prototype of a standardized facilities entity master table, the following standard definition for facility was developed: A University _facility_ is a building, structure, site, or parking lot used by Virginia Tech. A _building_ is a roofed structure for permanent or temporary shelter. A building must be attached to a foundation, be roofed, be serviced by a utility in addition to lighting, and undergo regular maintenance. A facility that does not meet this criteria is considered a _structure_ and defined simply as something that is constructed. A _site_ is an identifiable location. An example of a site is the "drill field." Parking lots are special sites that warrant a unique category. A _parking lot_ is an identifiable and designated area for parking vehicles. The next step was concentrated work with staff in the Facilities Planning and Construction office to refine and implement standardized facilities information, distribute the established standard and companion data definitions for the facility entity, replace existing facilities information with standardized facilities information, and turn over data custodial responsibility for standardized facilities information to Facilities Planning and Construction. Assess project progress (activities) At each milestone or every three months, a management report on project progress is prepared and presented. Feedback is solicited and current activity and direction re-evaluated. If unforeseen obstacles surface, a plan to address the problem is formulated. It is worth noting that so far these "obstacles" generally call our attention to a necessary step or interface that we have overlooked and must address as part of the standardization process. They serve as opportunities to improve the quality of the process. Develop data definition and documentation standards (data) Data definitions, descriptions, and documentation are created for each data element in an entity master table. A second "living" document outlines a standard for data definition and documentation. In addition to identifying the particular descriptive information that must be supplied, this document also includes a standard naming convention to provide a common language for referencing and relating like data elements. The evolving standard has also been applied to the census files produced in Institutional Research, is being used by several users to define their local systems, and has potential for use as a requirements analysis tool for reengineering projects. Develop data quality assurance methods and measures (data) To check results of the standardization process, measurements of data accuracy must be developed. Moreover, these measures must be systematic and eventually automated so ongoing quality assurance can take place on all information sources containing standardized data. Techniques for measuring data quality across distributed systems must be identified and procedures for testing and making corrections established and documented. An important step already taken is creating a baseline, in the form of a facilities entity master table, to assess the quality of facilities information in a variety of existing files. Finally, as there are new occurrences of an entity, codes must be assigned, all standardized data sources updated, and users notified.The project is working with data custodians to establish users groups and internal procedures to address these change management issues. Develop retention procedures for historical data (tools) The valid lists of values for the attributes of a standardized entity must be available over time, and for given points in time. Therefore, another important tool for ongoing quality information is a set of procedures to capture and store historical snapshots of all standardized data. This capture must occur on a regular documented cycle and include a copy of the data definitions current at the time of the snapshot. Census snapshots must also capture the related code validation lists current at the time of the snapshot. In working with the data custodians and data stewards to develop these procedures it is usually necessary to clarify the following: The procedures to capture historical standardized data do not replace other requirements to maintain historical transaction data from source systems for audit purposes. They also do not replace standard backups that the operational offices must do for disaster recovery contingencies. Implement repository and data distribution (tools) A primary data management tool is the repository. It consists of data dictionaries for standardized elements with data descriptions and definitions to provide both high-level and detail documentation, cross- reference multiple data sources, and support catalog type query, i.e., "what is available?" We have designed and developed this repository based on the Information Resource Dictionary Standard (IRDS), ANSI X3.138.1988. The repository rationale, standards, and functions are discussed in the data definition and documentation "living" document mentioned earlier. The system is set up so a user can access a data element dictionary for a particular data source and view a list of all the data elements (which also serves as a table of contents), look at a short description of individual elements, or look at the full data processing description of an element. Another important "tool" receiving attention from the project involves the method for disseminating standardized information along with standardized data definitions and descriptions. This effort is primarily one of communicating requirements to the University's information systems personnel. The objective is to be positioned to deliver standardized information in a consistent manner to a broad base of users operating in a variety of computing environments, using both client- server and mainframe technology. Meanwhile, the project has developed a prototype that gives users easy access to data element dictionaries, allows them to browse the facilities master table, and other master tables as they are developed, and provides the ability to read, copy, or print several standard reports from a master table. Continuous improvement equals progress The future will find us intensifying our involvement with others to extend and refine the standardization process. The project team will make itself available to campus groups and organizations that seek expertise and assistance with implementing standards in existing systems. Additionally, we will continue to work with our data custodians to develop and define ongoing data management functions and update job descriptions accordingly. Considerable progress has been made on standardizing the facility entity. We continue to refine the standardization process, collecting and disseminating standardized data definitions and standardized data for the facility entity and census snapshots from other systems, promoting compliance with standards in all data sources, and assisting reengineering task forces with data requirements definitions. Our primary focus is also shifting somewhat to the technical aspects of distributing standardized information and documentation to users so they can better manage their local functions and data and make decisions based on quality information. Lessons learned It is important to take time to consider what has been learned when evaluating progress. As a result of our work on the facility entity, we know considerably more about the process of standardization and are able to move more swiftly as we take on other entities. Additionally, we see improvement not only in the entities on which we have focused, but more importantly, in the expanding campus awareness of the merits of standardization. We are observing greater understanding of the importance and specifics of conscientious data management at all levels of the organization. We have learned the following: * If the distributed system support personnel do not find improving the quality of their data to be rewarding, then it will not happen. Benefits include the opportunity to expand and exercise their professional skills, more effective use of technology to make their lives easier, better service to their users, and recognition for their visible improvements in terms of both end-user product and end-user support. Modifying their job descriptions to reflect their data management responsibilities and data administration activities is also effective motivation. * From the typical user's viewpoint, occasional incorrect coding is not as serious a problem as the symptom, i.e., the inability to understand the meaning of data elements or the use of unstable or unknown criteria for including particular values in a database. * The warm, fuzzy feeling of "making progress" must be augmented and supported by setting and meeting visible milestones to maintain resource support and morale. * Success is very dependent on the project team's effective project management skills, as well as its ability to assemble and engage groups for both input and support. These skills should be reinforced and refined. * Team members must have access to custodians and have at least some champions in the upper administration. This is helped by treating standardization as a cross-functional project and stressing the importance of accountability and consensus building. Build on the concerns and successes of your champions. * Core team members should be physically located together and there should be a minimum of two people dedicated full-time to the project. This promotes the sharing of both technical and managerial skills, helps keep focus on the project goals, and provides a degree of shelter for coping with the inevitable frustrations of this type of project. * A focus on quality and never-ending improvement gets everyone's attention and shifts the focus from "why are we doing this?" to "what can we do, how, and when?" Endless debates and the telling of "war stories" are replaced with constructive discussions that purposefully build on recognition of those things that are already "better than they were." Looking back at "the way we have always done it" wastes everyone's time. * Credit for all improvements and successes should be spread as widely as possible. Individual ownership of project results fails to acknowledge the importance of a broad base of involved and empowered "team players" necessary for enterprise-wide data management. * There is no such thing as a tiny improvement. Every improvement is a "breakthrough," paving the way for the next ... and the next ... and the next. The PDCA cycle works well, providing prototypes that both demonstrate incremental improvement and serve as a baseline for continued improvement. * There is still much we don't know. We have benefited greatly from the "lessons learned" and shared by other institutions and we recognize many commonalities. We will continue to seek out opportunities to learn, and to learn how to enjoy learning by sharing, as we all strive for higher quality information. Looking ahead and stepping forward As expected, more progress has been made thus far with the people and activity parts of the toolset than with the data and tools parts. Clearly, the prototype workwe have done positions us to progress more quickly in successive system areas. As mentioned earlier, we also anticipate additional technical support to assist with appropriate tool selection and development to support standardization, data quality assurance, and, ultimately, readily accessible decision support systems with quality data. Future activities will include those already begun by the project. However, they will continually be expanded to include more of the University community addressing more of the total information resource. Also, greater attention will be paid to measuring improvement as the breadth of improvement expands. As our efforts bring forth end-user decision support products, there will likewise be new activity to support the users of those products with training, consulting, meeting with user groups, and continual product enhancement. This links directly with the more traditional institutional research function. Much of the groundwork has already been laid for each part of the integrated toolset necessary for distributed data management. The first sequential part of the toolset is _people_. This is the foundation we must continue to fortify. We do this largely by sharing our vision, stepping ahead, and then expanding our vision as we learn more. We are seeing our earliest vision become reality. We will continue to step ahead, inviting more people to become involved in formulating the expanding vision. Our strategy is to: * Influence data custodians to include the data stewardship activities in their system support job descriptions. * Assemble formal and informal University groups and focus their attention on data management obligations that should be part of their mission and purpose. * Support development of a high-level University data model to show interfaces between major systems and point out priorities for future standardization work. * Develop a prototype structure for the data management function that addresses coordination of data selection, capture and storage, manipulation, and delivery necessary to support management reporting requirements. We realized from the beginning that there was no "quick fix" to the apparent data quality problems. We also realized that our plan, in total, assumed a cultural change that would take time. Nevertheless, we set out with high expectations that we could make improvements and we have been successful. Keeping our goals before us every step of the way, we have tackled the job, one chunk at a time. We believe the future holds only more of the same, which includes sharing credit and celebration for each and every incremental improvement along the way. ************************************************************************ The Ten Commandments of Data Administration Standards 1. The first rule is that there are exceptions to every rule. No standard is applicable in every situation. However, the DA staff must not allow exceptions to become the norm. 2. Management must support and be willing to help enforce standards. If standards are violated, management must assist in assuring that the violations are corrected. 3. Standards must be practical, viable, and workable. Standards must be based upon common sense. The less complicated and cumbersome the standards, the more they will be adhered to. Keep standards simple. 4. Standards must not be absolute; there must be some room for flexibility. While some standards must be strictly adhered to, most standards should not be so rigid that they severely restrict the freedom of the data designer. 5. Standards should not be retroactive. Standards are to control and manage present and future actions--not to undo and redo past actions. In most cases, standards enacted today cannot apply to data design that began several months ago. 6. Standards must be easily enforceable. To achieve this, it must be easy to detect violations in standards. The more the process of auditing for the compliance of standards can be automated, the more effective will be the standards themselves. 7. Standards must be sold, not dictated. Even if upper management wholeheartedly supports DA standards, the standards must be sold to employees at all levels. DA must be willing to advertise the standards to all employees and to justify the need for such standards. DA standards demand that programmers and analysts change the way they design data. Any lasting and meaningful change must come from the employees themselves. 8. The details about the standards themselves are not important--the important thing is to have some standards. DA must be willing to compromise and negotiate the details of the standards to be enacted. 9. Standards should be enacted gradually. Do not attempt to put all DA standards in place at the same time. Once standards are enacted, begin to enforce them, but do it gradually and tactfully. Allow ample time for the non-DA staff to react and adjust to new standards. The implementation of standards must be an evolutionary, rather than a revolutionary, process. 10. The most important standard in data administration is the standard of consistency--consistency of data naming, data attributes, data design, and data use. ======================================================================== Footnotes: 1 K. L. Miselis, "Organizing for information resource management," in J. B. Presley, ed., Organizing Effective Institutional Research Offices, New Directions for Institutional Research, Volume 66 (San Francisco: Jossey-Bass, 1990), pp. 59-70. 2 L. Balkan and P. Sheldon, "Developing Guidelines for IRM: A Grassroots Process in a Decentralized Environment, CAUSE/EFFECT, Summer 1990, pp. 25-29. 3 G. W. McLaughlin and J. S. McLaughlin. "Barriers to information use: The organizational context," in P.T. Ewell, ed., Enhancing Information Use in Decision Making, New Directions for Institutional Research, Volume 64 (San Francisco: Jossey-Bass, 1989), pp. 21-34. 4 "A repository (IBM repository on its way)," Computer-world, 6 February 1989, pp. 87-90. 5 W. R. Durell, Data Administration: A Practical Guide to Successful Data Administration (New York: McGraw-Hill, 1985, pp. 31-32. ======================================================================== Data Standardization Means Quality Information from Distributed Systems