People often laugh at opinion polls and remember times when they were slightly incorrect. However, they often forget all the time when they were accurate. A poll should be conducted very carefully and should sample a cross-section of the population. The cross-section should be selected very carefully so that the right balance of ages, sexes and social classes can be achieved. At first it was very difficult for many people to accept that a sample of one or two thousand could reflect accurately the views of millions, but polls have now proved themselves. However, polls should not claim to do what they are not intended to do. The findings should always be examined very carefully before predictions are made with any degree of certainty.