The aim of the study is to investigate and compare experimentally the error finding strategies of a notation-familiar group with degrees in computer science related fields and a domain-familiar group on a simulation conceptual modeling representation based on UML. The use of eye movement an verbal protocols together with performance data underline the differences such as error finding and reasoning between two groups. The experiment with 20 participants also reveals that the diagrammatic complexity and the degree of causal chaining are the properties of diagrams that affect understanding, reasoning and problem solving with conceptual modeling representations. In a follow-up study with 24 university students, it is seen that these properties are independent of gender. The study also emphasizes the combination of different data collection modalities, namely eye movements, verbal protocol and performance data to be effective in uncovering individual differences in human-computer interaction studies in the domain of software engineering.