This term emphasizes that we never have a set of data that does not have a background. It must have been collected in some fashion for some purpose. Something must be known about the variables being measured. The situation is probably not totally unique, someone will have collected data of a roughly similar nature before. There may even be papers or books on the type of problem being studied. All this information can contribute to our understanding of how to approach the details of the modelling exercise, although they should not prevent us from noticing the truly new and unique features of the data.