Managing Organizational Data Resources: Quality Dimensions

The importance of data resources as a major component of Information Resources Management has long been recognized (Guimaraes, 1988). Many organizations are discovering that imperfect data in information systems negatively affects their business operations and can be extremely costly. Results from a survey indicate 50 percent of IS managers reported losing valuable data in the last two years and at least 20 percent with losses costing $1 million or more (Panettieri, 1995). Another survey reports 70 percent of the IS managers having their business processes interrupted at least once due to imperfect data (Wilson, 1992). Still another study showed that the nature of the problems associated with defective data ranges widely, from damaged files and lost data accounting for 23 percent of the responses, cost overruns (17 percent), conflicting reports (16 percent), improper regulatory reporting (13 percent), improper billing (9 percent), poor decisions (7 percent), delivery delays or errors (6 percent), and others (9 percent) (Knight, 1992). We believe imperfect data can result from practice-oriented and structure-oriented causes. Practice-oriented causes result in systems capturing or manipulating imperfect data (i.e., not designing proper edit checking into data capturing methods, or allowing imprecise/incorrect data to be collected when requirements call for more precise or more accurate data). Operational in nature, practice-oriented causes are diagnosed bottom up and typically can be addressed by the imposition of more rigorous data-handling methods. Structure oriented causes of imperfect data occur when there exists a mismatch between user requirements and the physical data implementation designed to meet the requirements. The imperfections are inadvertently designed into the implementation. Correcting structural causes more often requires fundamental changes to the data structures and is typically implemented top down. Structural problems result when a user cannot obtain desired results due to lack of access and/or lack of understanding of data structure, as opposed to getting an incorrect value or representation. Adopting an organization-wide perspective to data quality engineering integrates development activities using data architecture. Failure to develop systems as coordinated architecture components results in fragmented data resources whose definitions apply at best within system boundaries. One additional consequence is that data interchange among company systems and those of partner organizations is more difficult. Structurally defective data results in unfavorable outcomes such as: 1) providing the correct response but the wrong data to a user query because the user did not comprehend the system data structure; 2) organizational maintenance Little guidance has …