This paper presents a new and better method for teaching the derivation of the appropriate form of a multi-attribute utility function under various assumptions of utility independence. This new method has three advantages over the usual mathematical method for teaching this subject. First, because our method is a natural elaboration of the method often used to teach single-attribute utility assessment, students learn multi-attribute utility assessment as an extension of single-attribute utility assessment. Second, because our method uses the same visual representation used in teaching single-attribute utility assessment (the tree), students who learn by using visual representations may learn the multi-attribute concepts better. Finally, our method clarifies the assessments needed to determine the weights for each term in the multi-attribute utility function.