Data Management and Handling
Our programming experience includes the development of data management information systems on a wide variety of projects. For each project, our programmers work with study staff to develop a workable study management information system, which is used to account for the status of all sampled cases and to track progress through the various stages of data collection, that is, to record and monitor all critical study outcomes; produce reminder listings, showing ID numbers for cases for which an outcome is expected or overdue; print advance notice reports; notify study personnel of data collection events that need to occur for a study subject during a scheduled timeframe; generate study mailings, followup forms, call sheets, and other individualized letters or forms; and report and statistically summarize study progress in various ways. Study staff can generate reports on a demand basis to obtain management information, such as overall response rates, response rates by interviewer, results of tracing efforts, and productivity rates by interviewer
To manage data, we create tracking systems to manage study progress. For computer-assisted telephone interviewing (CATI), we use Blaise, a software package developed by Statistics Netherlands. Blaise is a flexible system that supports survey development, data collection, data editing, and the extraction and use of survey data. For our CATI studies, we take respondent contact information (e.g., interviewer, date, time, result) that is automatically generated within the CATI program and write it to a management database. The database is then updated daily by the CATI program, and additional data items are added as needed, such as status of incentive payments, receipt of consent forms, and nonresponse conversion results.
The management information system for each study uses internal logic checks to allow the entry of only valid outcome codes while simultaneously checking for consistency across events. When new events are entered into the tracking system, the validity checks are made automatically and feedback is provided if errors are found so that project staff can review them against source documents and make necessary corrections. The tracking system is also used to check study records against the research data sets created for the study.
Close collaboration between programmers and project staff has made possible easy-to-use, straightforward entry screens with built-in safeguards and other enhancements as requested by the users. It is our experience that much of the programming effort required to develop a data management system for one project is transportable to others.