ABSTRACT

Statistical significance" relates to the problem of inference, or how much confidence we have that a result that we've found in a sample of research participants relates to the larger population from which that sample was drawn. This chapter presents one tool, the funnel diagram, which can be used to detect the effects of such biases. In the research on gender differences, the idea that "men and women are the same" is conventionally taken as the default position or "null hypothesis". This functions like the presumption of innocence in the jury trial. Researchers are said to be like the prosecutors, gathering data and summarizing the evidence. The chapter discusses one clever method of detecting publication bias. Confirmation bias in the form of a tendency to prefer estimates with a particular sign results in a funnel graph that is asymmetric, with points on one side or the other of zero scarce or missing.