ABSTRACT
This chapter is largely critical as the title suggests. The negative criticisms expressed in it were the spur to all previous chapters. Without a feeling of dissatisfaction, there
is no covers a
11.1
The
lhd(θ ;y) = lhd(θ ;y1, . . . ,yn) = f (y1|θ) . . . f (yn|θ) is the second most important thing in (both Frequentist and Bayesian) Statistics, the first being common sense,
For this reason the first section is on likelihood. One of the main theses of this book is that it is possible to discard likelihood to a large extent. Its use must be restricted to very special situations where the model has been regularized and is consistent with the data. An optimal likelihood based procedure is then a border post delimiting what is the best possible under the most unfavourable circumstances. Common sense will always be needed. It is common sense which prevents likelihood from going astray.
