"How to Ensure the Replicability of an Ad Hoc Research Strategy: A Few Lessons Drawn from the Sociology of the Concept of Public Service"
This presentation will account for the replicability of a constructionist method that can be deemed unusual. Namely, set theory, used to compare the studied population with a control group, proved particularly appropriate for a “sample” that was not meant to be representative since it had not been randomly drawn from a population, but rather constituted by the weight that each actor exerted in the studied field. Also, in such a varied corpus comprising “conseillers d’État”, academics, politicians and trade unionists, it proved impossible to compare the sample of “conseillers d’État” with the thousands of members of this institution under the Third Republic while, at the same time, doing the same for all other groups, given the countless number of public servants in this 70 year period. Instead, we used other techniques (archival ethnography, geometric data analysis) to devise an unbiased “sample”.