How does the design of the research inform the use of a particular statistical test?