The present study examines the difficulties novice data modelers face when asked to provide a data model addressing a given problem. In order to map these difficulties and their causes, two short data modeling problems were given to 82 students who had completed an introductory course in database modeling. Both problems involve three entity sets with relationships between them, either ternary or binary. The students' solutions were classified according to the types of errors they committed. More than half of the students provided faulty solutions. After an analysis of these results, open interviews were conducted with a selected group of students in order to figure out the reasons underlying the students' erroneous decisions regarding the data model. Among the reasons for their erroneous solutions were insufficient experience, lack of reflection on their solution, and lack of immediate feedback. In addition, the authors suggest instructional modifications derived from the research results.