Due to high cost and complexity of Intelligent Tutoring Systems (ITS), current systems typically implement a single teaching strategy, and comparative evaluations of alternative strategies are rare. We explore two competing strategies for teaching database normalization. Each data normalization problem consists of a number of tasks, some of which are optional. The first strategy enforces the procedural nature of the data normalization by providing an interface that requires the student to complete the current task (i.e. a part of the problem) before attempting the next one. The alternative strategy provides more freedom to the student, allowing him/her to select the task to work on. We performed an evaluation study which showed that the former, more restrictive strategy results in better problem-solving skills.