Predicting elections from biographical information about candidates: A commentary essay

Armstrong and Graefe apply the index method to predict presidential elections. They imply that the technique is also useful for business decision making. Their idea has merit and may be relevant when the decision context is dynamic, has few prior “observations,” and where domain knowledge exists. However, Armstrong and Graefe fail to adequately explain the variable selection process, clarify the conditions when the index method is appropriate, or identify the types of problems most amenable to the index method, and fail to discuss how the index method can be calibrated to help make single option decisions.