Le présent article s'inscrit dans la suite logique d'un texte précédemment publié. Les auteurs y dépouillent une variété de définitions opérationnelles de la satisfaction au travail à partir desquelles ils opèrent un choix en accord avec leur position théorique. Ils procèdent ensuite à une revue sommaire des instruments de mesure disponibles et des considérations méthodologiques pertinentes afin de suggérer un devis propre à faciliter l'élaboration d'un instrument de mesure adéquat.
The present study was undertaken in order to point out some important guide-lines to follow when measuring job satisfaction. To do so, the authors go through the following steps.First, a critical analysis is carried out in order to determine the extent to which the five operational measures of job satisfaction, that have been reviewed, are, in fact, matching the authors' conceptual option of job satisfaction. The job satisfaction definition developed by the team of the Work Adjustment Project at the University of Minnesota is retained. Satisfaction then, is defined in need fulfillment terms as a « ... correspondent between the reinforcer system of the work environment and the individual's needs ». The Minnesota Satisfaction Questionnaire, based on this definition, requires the subject to respond in terms of the difference between what he would like to receive and what he does receive.Second, a descriptive review of some of the job satisfaction questionnaires, currently used, is presented in order to know for what purpose these job satisfaction instrument measures were used and to disclose the most important job facets related to satisfaction such questionnaires should include. Nine factors of job satisfaction are retained: 1) general factor; 2) intrinsic factor; 3) extrinsic factor; 4) working conditions ; 5) recognition ; 6) coworkers ; 7) supervision ; 8) autonomy and 9) self-actualization.Third, some of the most important psychometric considerations searchers should take into account, in building that sort of questionnaire, are discussed. The most appropriate type of questions seem to be the Likert one. Reliability should be tested through homogeneity of the items, internal consistency of the items (split-halves or odd-even) and stability of the instrument. Validity, on the other hand, should be tested by more than one means, such as content validity, concurrent validity, and construct validity.
