A methodology to deal with choice by a group of decision makers is here developed. Its first step
consists on obtaining individual evaluations of the available options. These evaluations are seen
as estimates of location parameters of random variables and each vector of individual evaluations
of the whole set of options is transformed into a vector of probabilities of being ranked as the
best choice by that individual decision maker. The next step is the probabilistic composition
of such individual vectors of probabilities into a unique vector of aggregate preferences. To do
that different composition procedures may be applied. The comparison of the results of distinct
composition strategies is employed to detect outliers in the individual evaluations and, fnally,
to filter the best options. After the initial evaluations are obtained, the whole process may be
automatically developed. This makes the methodology particularly useful when fast decisions
are needed. Its applicability is here illustrated by a case of daily revision of a stocks portfolio.