A method to improve the accuracy of the data flow diagram (DFD) developed in the requirements analysis phase of the software development life cycle (SDLC) is proposed. The system analyst develops a DFD in consultation with the user or customer. This is a crucial phase: 50 percent of the errors in the later stages of the SDLC, such as system design and development, are due to inaccurate requirements analysis.

Earlier research had attempted to modify the process of creating DFDs, or proposed using alternative means other than DFDs. Freeman suggests that, instead of this strategy, the analyst should be given a refresher course in data flow diagramming before interacting with the client.

To test this theory, a small experiment was conducted with 16 subject-analysts. The acting analysts were management information systems (MIS) majors, with backgrounds in system analysis and design. Half of the group was refreshed on DFD before they interacted with users. Sixteen non-MIS majors acted as users. A refresher course did improve the accuracy of the DFD developed; there was a 17 percent increase in accuracy from a 20-minute refresher session.

The paper is well written. However, the results could have been presented in a more intuitive manner: the graph could have been plotted using the sorted accuracy score of the two categories of analysts, and the average score could have been displayed as a reference line in the plot. One of the analysts not trained with the refresher course shows better accuracy than the rest of the DFD-refreshed analysts. An explanation of this would have been interesting.