The extent to which you can come to conclusions about one thing (often a population) based on information about another (often a sample).