Before we can undertake any effective data analysis we must be clear about the precise question to be answered and confident that it is properly represented by its statistical image. Ambiguous questions or misguided mappings are likely to lead to mistaken answers, incorrect conclusions, and flawed decisions. But all too often we use a statistical tool or method because it’s one with which we are familiar or which is widely used, without considering carefully whether that tool really answers the question we mean to ask. This talk illustrates this challenge with a variety of examples of situations where the “obvious” method is not the right method, ranging from simple to sophisticated statistical challenges.