A parameter of a model becomes a decision support indicator if it is used as an argument of a decision function. This implies consideration of a wider model than the initial one and to acknowledge its complexity. It is also necessary to include the estimation of parameters in indication theory and to consider a data driven approach. Data analysis leads us to provide sets of indicators. With reference to the statistical concept of sufficiency (Fisher 1925), such a set may also be considered as a set for indication, equivalent to the original data for any subsequent purpose of indication. From the same data set, several models may be built providing different sets for indication. None of them can be considered as the best; this depends both on the context (e.g. fishing practices) and on the addressed questions. No model must be chosen to the exclusion of alternatives: an illustration is given with models in relation to a catch-effort data set, with a case study on Senegalese small scale fisheries.
