Quality and quantity of data.

Abstract

When collecting data, deciding what type and how much to obtain will depend in large measure on how the data will be used and the group(s) that will be given the data, as well as the risk-taking or risk-averse position adopted. Among the most important issues involving the quality of data are knowing what plans and activities will lead to obtaining high-quality data and recognizing whether the data obtained achieve one's goals. Factors to consider in deciding on the quality and quantity of data to collect include the following: robustness of data required for extrapolatability; potential use of the data; degree of bias that is acceptable; the quantity of data that must be collected in a study or trial which must be considered separately from the quantity to be submitted in a regulatory dossier; practicality of the study; cost of the study; and time available and necessary to complete the study. In preclinical experiments involving discovery research, the protocols prepared are generally loose and can be readily modified during the study in order to take advantage of serendipitous observations or sudden ideas of the experimenter; this is not generally the case with well-controlled clinical trials, although in some circumstances feedback loops can be established to monitor data collection during a trial. A feedback loop of adjusting the amount of data to collect during a study has both advantages and disadvantages (described in the text) when compared with creating a relatively inflexible plan in advance of starting a study, as required for most clinical trials.