Inference problems in population genetics: DNA sequences, restriction endonucleases and ascertainment sampling.

A major trend of population genetics theory in the 1970s was the increased emphasis on inductive arguments, based on observed genetic data, rather than on deductive arguments based on theory and models. This occurred in part because the deductive theory had largely fulfilled its role of describing evolution as a genetic process, and in part because of the increasing amounts of data available on the genetic constitution of natural populations. Inference procedures raise difficulties not present in the deductive theory. Often conditional arguments are necessary since the data often must fulfil some condition to be observed. Different inference procedures, having different efficiencies, apply for data from different apparatuses. Care must be taken in deciding what it is that the inference concerns. These problems are illustrated by reference to restriction endonuclease techniques and ascertainment sampling.