Working Papers 2001 – Abstracts
On the Nature and Role of Hypothesis Tests
Alan McLean
Hypothesis testing is widely regarded as an essential part of statistics,
but its use in research has led to considerable controversy in a number
of disciplines, especially psychology, with a number of commentators suggesting
it should not be used at all. A root cause of this controversy was the
overenthusiastic adoption of hypothesis testing, based on a greatly exaggerated
view of its role in research. A second cause was confusion between the
two forms of hypothesis testing developed by Fisher on the one hand and
Neyman and Pearson on the other. This paper discusses these two causes,
and also proposes that there is a more general misunderstanding of the
role of hypothesis testing. This misunderstanding is reflected in vocabulary
such as "the true value of the parameter".
Keywords: Hypothesis test, Significance, p value, Probability
Model, Statistical model, Prediction.
Next Abstract
