SummaryThe paper is in the form of a dialogue between Fred, a young mathematical statistician with a model building view of survey design and analysis and Harry, an experienced survey statistician who prefers the traditional sampling distribution basis. The two essentially different philosophical approaches are brought out in their conversation which is concerned with a proposed traffic noise survey. Harry has been commissioned to design a survey to evaluate the effect of various sources of traffic noise on people for future planning purposes, the underlying theme of the paper is the concept of the population for which inferences are required and the different approaches which the two statisticians take to achieving this.