One might argue that unemployment in an area affects the number of crimes, due to the higher number of people who feel alienated from society. Therefore the unemployment rate is the independent variable. Alternatively, one might argue that crime rates depend on factors other than unemployment, and that unemployment is a consequence of crime as businesses move to safer areas. according to this argument we would use crime rate as the independent variable. The choice of dependent and independent variable is, in other words, affected by the theoretical arguments from which research hypotheses are derived. It is not an issue that the statistical tests themselves can prove. Statistical tests cannot explore the complete nature or channels of any causality. All that statistics can do is compare differences given the way that the problem has been posed.
All these consideration involved in organizing data are summarized in Table 13.1.