ABSTRACT
This chapter describes the well-known discrete frequency distributions: the binomial, Poisson, negative binomial, geometric and logarithmic series distributions in the simplest case of an unstructured sample. Discrete data often follow various theoretical probability models. Graphic displays are used to visualize goodness of fit, to diagnose an appropriate model, and determine the impact of individual observations on estimated parameters. Discrete data analysis is concerned with the study of the tabulation of one or more types of events, often categorized into mutually exclusive and exhaustive categories. Binomial-type data arise as the discrete distribution of the number of “success” events in independent binary trials, each of which yields a success with a constant probability. Probability models for such data provide the opportunity to describe or explain the structure in such data, in that they entail some data-generating mechanism and provide the basis for testing scientific hypotheses and predicting future results.
