Conclusions : study is introduced, with more performance measures and firmsthan in the two previous case studies. Although different values are obtained with the Montecarlo simulation, the interpretation of the results remains the same. For comparing the results, we propose using the sum of absolute deviations (Z) and the maximum absolute deviation (D) between the multicriteria performance and the single-criterion performances, along with Spearman’s correlation as traditionally applied. In the three case studies, the simulation shows that the boundary in plane D–Z of the solutions obtained with goal programming encompasses the CRITIC and TOPSIS solutions. As the extended GP model can produce different solutions according to what value is assigned to the l parameter, it might be thought that it implies adding an element of subjectivity to the problem, given that decision makers have to choose between the different efficient alternatives. However, the selection of the performance measures themselves introduces a source of subjectivity that is inevitable. It is our considered opinion that the models provided by GP serve to enlarge the spectrum of methodologies reported in the literature. They also empower decision makers to choose from the various solutions the one that is best suited to their strategy, while always retaining the ability to derive the weights from the initial data. Our recommendation for future research is that it would be of great interest to analyze the sensitivity of the above methods with the goal of quantifying the impact the inclusion or exclusion of one or more single-criterion measures might have on the results. Such a study could provide another element for comparing the different methodological proposals. Additionally, as shown in Case Study III, an application of the methodology may consist in relating firm performance, size and firm value