The process of knowledge acquisition which has the strongest truth cla
im is the research process of natural science, based on testing hypoth
eses to destruction. But the application of this process to phenomena
beyond those for which it was developed, namely, the natural regularit
ies of the physical universe, is problematical. For research into soci
al phenomena there is increasing interest in ''action research'' in va
rious forms. In this process the researcher enters a real-world situat
ion and aims both to improve it and to acquire knowledge. This paper r
eviews the nature and validity of action research, arguing that its cl
aim to validity requires a recoverable research process based upon a p
rior declaration of the epistemology in terms of which findings which
count as knowledge will be expressed.