ON THEOREMS OF DE FINETTI TYPE FOR CONTINUOUS-TIME STOCHASTIC-PROCESSES

Most statistical models used in practice admit a data reduction by mea ns of a sufficient statistic. This paper concerns the problem of chara cterizing explicitly those models that are compatible with a certain c hoice of data reduction. We suggest a general method for establishing such results when a stochastic process in continuous time is observed. We give examples both from diffusion and counting processes and discu ss some implications for statistical inference.