ABSTRACT
In the first four chapters of this book we discussed simple descriptive statistics. In this chapter and subsequent chapters we shall deal with inferential statistics. Descriptive statistics provide us with a quantitative "picture" of some aspect of the world. Such statistics may be drawn from populations or samples and, strictly speaking, apply only to the population or sample from which the statistic was derived. For example, the decennial census of the United States gives us many descriptive measures of the American population. These census data apply only to the United States and cannot be used to describe any other population. Inferential statistics also give us a picture of some aspect of the world. But inferential statistics differ from descriptive statistics in that the former enable us to generalize beyond the data at hand. For example, public opinion polls such as those of Gallup or Harris measure the opinions of approximately 1,500 people and from the descriptive measures derived from the sample infer a description of the opinions of the entire American population. What makes such inferences possible? How can we trust that the opinions of 1,500 people will accurately reflect the opinions of millions? The answers to such questions lie with the concepts of probability and sampling, the subjects of this and the next chapter.
