Several data models exist at the conceptual level, the most popular being the Extended Entity Relationship Model (EERM). However, the EERM may be difficult to use by end-users, when specifying their data requirements, and is typically used by systems analysts as a precursor to mapping the data to the relational model, in normalized form. A possible reason why the EERM may be difficult to use by end-users or novice data modelers is that the constructs in the EERM may be abstract with regards to the actual usage of the data. In this research in progress, we present an exploratory empirical study that investigates the effects of abstraction of concepts on novice analysts’ data modeling performance.