Journal of International Information Management


Firms sometimes use two-operator data entry as a method to achieve or maintain database quality. When in-house staff are used, the firm typically selects data entry operators from a pool of junior staff and then assigns them into operator pairs, often on a random basis. Keying discrepan cies between operator pairs are compared to determine incorrect entries, in the same row and column. Because the likelihood of making an error on a given key varies among operators, the objective of this study was to optimize database quality by systematically matching operators. The model was developed by having 32 operators key data into two databases and monitoring the location of each operator error. The database quality of all operator combinations were compared to determine optimal operator pairings. This resulted in 319% fewer errors in a second database over the expected number of errors which would have occurred from random operator pairings, and produced a database that was nearly 99.95% error-free. Regression analysis used operator error rates and total number of errors from each operator pair as independent variables. The dependent variable was the number of errors committed by each operator pair in the second database. The model explained 69% of the variability, and was used in a subsequent study using 28 different operators who entered a second database which was different from the first study. This resulted in a 0.85 correlation between the predicted versus actual in the second group.