During engineering design, different representations are used to convey information about a systems' components, functionality, spatial layout, and interdependencies. These varying representations may have an impact on the interpretation of a system and consequently the decision-making process. This paper presents a research study that tries to capture these different interpretations by investigating how designers divide a system into subsystem clusters. These subsystem clusters can be considered partitions of a set-in combinatorial mathematics. Given designers' subsystem clusters for three products across three representation modalities, three different analysis methods for finding the most likely partition from observed data are presented. Analysis shows that the Variation of Information analysis method gives the most coherent and consistent results for the search of a most likely cluster. In addition, differences in clustering behaviour are observed based on representation modality. These results show that the way an engineer or designer chooses to represent a system impacts how that system is interpreted, which has implications for the decision-making process during engineering design.