The issue being investigated is the role of the type of representation used for recording design ideas in idea generation techniques for conceptual design. An empirical study was conducted to test the hypothesis that graphical (pictorial) representation is better than textual (sentential) for recording of design ideas. This study used a standard set of outcome measures: fluency, quality, novelty, and variety of ideas generated by use of different representations. The expressiveness of each representation was also studied by examining the ability to convey key design information of an idea. The experiment was designed and data analyzed based on the principles of statistical DOE. The study confirms the hypothesis that graphical representation provide greater benefits for engineering design idea generation than textual representation.