Do's and Dont's in searches and fits of signals


I will discuss things to do and not to do in searches and fits of signals which show up as a bump in a mass spectrum on top of backgound. The following issues will be covered: Goodness Of Fit tests for background only hypothesis, choice of background parametrisation, background shape systematics and things to avoid like fitting after background subtraction. The talk content is heavily influenced by personal experience in the CMS Statistics Committee which I chaired during the last three years.




Olaf Behnke