... analysis.1
Based on the series of lectures Statistical methods for frontier physics'' given at the VI LNF Spring School Bruno Touscheck'', Frascati, Italy, May 2001.

Email: dagostini@roma1.infn.it. URL: http://www-zeus.roma1.infn.it/agostini.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... rules2
Note that the translation of these rules into confidence'' is not straightforward, unless the log-likelihood (or the ) has a nice parabolic shape and all values where the likelihood finally concentrates are initially considered about equally likely.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 37\%.3
This value should be compared with 39.3% obtainable in the case of a perfect 2-D Gaussian, thus indicating that the final distribution comes out naturally approximatively of that kind.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.