The first step in designing an experiment was blithely described in chapter 2 thus: ‘We begin with a prediction that we are interested in testing’. It would be more accurate to say that we end the first stage of scientific inquiry with a prediction that we are interested in testing. A prediction is a specific statement about the relationship between two variables which is the outcome of earlier thinking of a more general and disorganized kind. We know surprisingly little about the processes which underly this kind of thinking. But certainly the formulation of predictions is a more creative and intuitive endeavour than we are often led to believe in journal articles and scientific reports. Very few psychological hypotheses are actually formulated by direct deductions from some more general theory. They must surely arise from some kind of interaction between the experimenter’s intuitions, theoretical ideas and factual knowledge. In contrast to the more formal aspects of experimental design, the generation of a research problem defies logical analysis, and there are certainly no guidelines for the formulation of fruitful or successful approaches.