An MIS Data Quality Methodology Based on Optimal Error Detection.

Abstract As organizational information systems increase in use, cost, complexity, and importance, the quality of the data upon which decisions are based becomes critical. Presently, quantitative measures of an organization's data quality do not exist, with organizations relying on periodic audits of selected applications to evaluate the accuracy of their systems. In this paper, a quantitative measure is developed by formulating the error rate of stored MIS records, whereby MIS records are classified as being either "correct" or "erroneous." This approach has several desirable characteristics including: (1) specification of a bench- mark for comparing the actual error rate with the theoretically smallest attainable error rate, and (2) specification of a method for allocating records to the categories in an optimal manner. Following development of the formulation, a methodology is provided for assessing an organization's data quality. Since quality control methodologies are beneficial only when appropriately applied in the specific organizational setting at hand, guidelines are given for deriving values for the parameters required in the formulation. These values can be obtained from a variety of sources, both mechanized and manual. The paper concludes with a discussion of the relationship between the parameters in the formulation and the verification mechanisms designed to insure data quality.