This communication presents the results of a second examination of decision taking by a section of the National comity of the Cnrs (Centre national de la recherche scientifique) in charge of research assessment and it relates them to those previously gathered (Gabriel, 2001). The modelling of the decision taking by the board of examiners concerning eligibility of the candidates to the junior researcher and senior researcher competitive entry examinations in 2001 of this section open to educational problematic has shown the weight of different contextual factors (ie. features of the candidates and of the examiners) independently of the effects linked to the scientific facts produced by the examiners and of the information accumulated during the collective treatment, auditions and further debates. This matter of facts has recommended the renewal of the examination with the aim to apprehend the respective parts of systematic and random variabilities in this kind of observation. It is what does this new examination which leans as previously, on a quantitative multifactorial and exploratory analysis of the decision taking, from the 2002 entry examinations session. The understandings and insights yielded by this enquiry into research assessment are discussed regarding the partnership with the Cnrs, but also considering that essential aspect of the educational activity that the assessment of scientific products represents.