Understanding data: important for all scientists, and where any nation might excel

Abstract

In some areas of science, the big rich countries are at an advantage. But in many cases, even well resourced research projects do not make optimal use of the data collected. It is argued here that it is sensible for individuals, for institutions, and even for nations to give some degree of priority to the process of understanding data. The key point is value for money. At relatively low cost, good understanding of data gives a relatively high pay-off, as much can be done with only a desktop computer and an educated human brain. University statistics departments are central to helping a scientist develop two types of skill: a repertoire of techniques for data analysis, and an ability to create a special model tailored to a specific dataset. The example of the first discussed here is the accelerated life hypothesis. There are useful techniques in textbooks that seem not to be known by many scientists working with survival times of organisms or equipment. As an example of the second, the correlation across different studies between response to treatment and response to placebo is discussed.