When we ask questions such as "What fraction of all smokers prefer cigarette brand A?", "What fraction of all rural homes have running water?", of "What fraction of all brand A coffee cans really contain one pound?" we are dealing with statistical problems. To solve these problems a three step procedure is followed. The first step is to construct the experiment or determine the sampling procedure. This consists in choosing questions which will give you the exact information you seek, excluding as much as possible any possibility of ambiguity. The second step is the collection and analysis of the data from the questionnaires, or one of the other data-gathering devices.