Replication and Pooling
Michael B. Bracken · Yale University Press eBooks · 2013
This chapter is concerned primarily with using replication to rule out chance as an explanation for study results. New hypotheses are derived from many sources, such as from observations in the laboratory or in the clinic, by analogy from other data, as well as by failure to replicate an early study result. If the scientific process is to move forward in an orderly and systematic fashion, and if clinical guidelines, legal opinions, and public health policies are to be based on sound evidence, validating a study result through replication must be a priority. It must be noted that replication does not mean exactly repeating the methods used in previous experiments because mistakes committed there may carry over to the new experiment. Instead, replication should be a series of studies of improving methodology as one learns from the previous mistakes. New studies are not merely replicas of earlier research; they should be based on all of the evidence to date.