CDF Statistics Web Page

Who We Are?

The CDF Statistics Committee consists of

• Louis Lyons, Oxford (chair)
• Craig Blocker, Brandeis
• John Conway, Rutgers
• Luc Demortier, Rockefeller
• Giovanni Punzi, Pisa
• Pekka Sinervo, Toronto

Our mandate from the CDF Spokespersons is as follows:

"The charge of the Statistics Committee is to identify typical problems in statistical analysis of CDF data and to propose solutions which conform to sound statistical procedures. Examples of such problems are setting limits, fitting to theoretical predictions and merging of results from different analyses/experiments. The committee is requested to document their recommendations in the form of CDF notes and public WEB pages, and present the proposals to the CDF Scientific Council for approval.

The committee is requested to identify the first set of problems to be treated by June 30 and to provide well documented recommendations for that set by January 2001. The particular problem of setting limits should be treated on a shorter time scale to take advantage of recent information from the CERN and Fermilab Limits Workshops."

There are a number of software tools that have been developed to perform certain types of statistical calculations. We list the ones that we have "qualified" below:

• POISSON is a program found in the P_U area of the Run I Offline code (use Version_7_12). It calculates upper limits assuming poisson statistics and folding in systematic uncertainties. It can accommodate background estimates, uncertainties on backgrounds (assuming Gaussian distributions), and uncertainties on overall normalization factors (such as efficiency and acceptance).
• CALCCL is a program found in the P_U aqrea of the Run I Offline code (user Version_7_12). It calculates 90 and 95% upper limit confidence level limits on likelihood distributions, folding in any additional Gaussian uncertainties on normalization factors.

There are a number of techniques we recommend to address specific sorts of statistical problems. We have gathered these below:

We list below a set of useful statistical references:

• The Particle Data Book,