The Effect of Data Quality Tag Values and Usable Data Quality Tags on Decision-Making

摘要

Prior research has shown that supplying decision-makers with data quality (DQ) tags, metadata about the quality of data used in decision-making, can impact decision outcomes in certain circumstances. However, there is conflicting evidence as to how or when decision outcomes are affected. In order to improve experimental soundness, the current research addresses possible sources of ambiguity in previous DQ tagging experiments with respect to DQ tag design. In particular, usability, semantics, and the rationale for assigning DQ values are explicitly considered in the DQ tag design of experiments. DQ tag values are designed to test the question of whether DQ tag impact is influenced by the relative importance of the "poor quality" attribute(s) to the specific decision task, thus potentially explaining discrepancies noted in observed results of prior research. The results showed no significant impact on decision choice, confidence, or efficiency; even with moderately rather than critically important attributes identified as being of poor quality. There was, however, some evidence of decreased consensus with DQ tags. These findings thus contraindicate the inclusion of data quality metadata in decision-making applications as a general business practice.