Good Data Management Posted on October 19, 2007 Charles F. Reynolds III (bio) advocates staying very close to your data.

A good data manager will do all kinds of quality control checks on the data to make sure that, for example, the rates of missing data are very low, to make sure that the values being entered into the field are credible, to make sure that there is good reliability of data entry and, also data retrieval. I would say it’s very important for investigators to look at their data as an ongoing process. It is a mistake not to look at the data close in time to the point when it was originated. If you are not close to your data, you don’t really know what’s going on in the process of your science and in your own laboratory.

My own practice has always been on a week to week basis to sit down with the team and to look at all of the data that were originated in the past week. People know that I care deeply about this and that I’m gonna hold them accountable for it. I pay a lot of attention to issues of interrater reliability, for example, because the work that I do is mostly longitudinal in nature. I do a lot of work with maintenance treatment strategies, you can’t afford to have missing data. Absolutely critical. So I work closely with data management staff, I pay a lot of attention to process issues within my team, to the clinical raters and the clinicians who are originating those data. Do they get them in a prompt way to the data management team? This is about teamwork and coordination between the raters and clinicians who originate the data and the staff who manage that data.

I also on at least a once a month basis ask the data managers to produce a report on the number of subjects that we have accrued, where people are in the pipeline in a particular protocol. We will have looked at their data but I want to know exactly what the performance of the protocol is, both respect to the clinicians who carry it out and the data managers who manage that information.

Looking at that report becomes an opportunity to strengthen the collaboration between the data mangers and the clinical part of the team. Again, it’s about teamwork. It’s about everyone feeling that what they do is fundamentally important and it’s about my communicating as the PI that I care about the job that you’re doing. I’m gonna hold you accountable for the quality of the job that you’re doing and I’m gonna look at your work very often.

« Back to Article