Some Common Types of Investigation

Abstract

When the history of science in the first half of the twentieth century comes to be written there is no doubt that Statistics will occupy a very important chapter. This new discipline has been developed largely in the last fifty years and already has come to play a leading role in most branches of scientific research whether they are concerned with well-controlled laboratory experiments or with large, less precise and less well-defined observational surveys in the field. Indeed, if we confine our attention to research in the behavioural and medical sciences, with which this book is concerned, it is hard to imagine investigations being satisfactorily carried on today without the aid of statistical procedures and methodology. The main reasons are threefold: in the first place the measurements which it is possible to make of variables such as reaction times, muscle, tension, learning speeds, intelligence, not to mention anxiety, depression, thought disorder, and so forth, are so variable-not only from person to person but also for a single person at different times—that it would be difficult to describe our findings without the aid of summary statistical measures. In the second place it is generally impracticable to record data for all members of the class of individuals in which we are interested in any particular investigation and we have to be content with a sample from it. Here again statistical theory is essential in telling us how best to draw our sample and in assisting us to make inferences about the complete class on the basis of data derived from a sample only. Finally, it is the case that much of the information derived from experiments and investigations comes from making comparisons between deliberately chosen, or perhaps just recognizable, sub-groups of people or things, so that once more we must turn to statistical theory to discover how the comparisons may reliably be made.