Display Characteristics and Errors at a Complex Data Entry Task

Abstract

Some of the most expensive data points ever collected, outside of satellite transmissions and particle accelerators, have come from clinical trials. In large clinical studies it is common to have complex treatment protocols, large samples, and multiple distributed data collection centers. For data to be useful the study's administrators must ensure that criteria, treatments, measurements, and reporting procedures remain uniform across a large number of geographically distributed data collection sites each with its own unique administrative structure and allocation of responsibilities for the data collection task. PoP is a software system developed over the past seven years at the Epidemiology Data Center (EDC) in the Graduate School of Public Health at the University of Pittsburgh. PoP was designed to assist in the collection of data and to provide quality assurance for clinical trial data. Data is entered locally into microcomputers running PoP at each of the participating sites. Data entry screens are closely matched to the paper forms on which the data is collected in order to facilitate entry and reduce errors. To further reduce transcription errors, PoP requires an independent second “verification” entry of each item of the form before it can pass on to a sequence of logical checking procedures. After data has passed all local verifications it is electronically transferred to a computer at the EDC.