Lies, damned lies and statistics
Kirsten Juhl, Per Markestad · Norwegian Archaeological Review · 1991
Differences in archaeological find material are most often implicitly taken to represent differences in the underlying prehistoric material in a one‐to‐one relationship. Contrary to this assumption the theory is adopted that a certain type of archaeological material must be treated as random samples from a large prehistoric artefact population and therefore will show random variations. These variations must be analysed statistically. A statistical test is developed to help decide if the differences are real or just a result of the random sampling. The test is applied to a Late Palaeolithic example from the archaeological literature.