The principle that the results of a statistical experiment or study should be independent of each other in order to be generalizable. It is a fundamental concept used to ensure that random samples provide accurate and objective results in research and analysis.