ABSTRACT
The last several chapters introduced ways of statistically testing hypotheses couched in terms of a mean or difference between means. Sometimes communication researchers are interested not in a mean but instead on how a set of research units are distributed across a set of categories. For example, communication theories and research hypotheses often make predictions about how people’s responses to questions should be distributed across a set of response options, how frequently certain categories of objects (e.g., types of television shows) are found in a population (all prime time network television broadcasts) or the choices that people make in response to a question or their preference for an object out of a set of objects presented to them. Or, as discussed in the last chapter, sometimes communication researchers are interested in comparing two groups on an outcome measure that is dichotomous, such as responses to a yes/no question or whether or not a person chooses to purchase a product, agrees with a message, or talks to someone about a problem they are having, for example. Such questions can be tested by assessing whether two nominal variables (e.g., group membership and the yes/no decision) are statistically independent. The tests described in this chapter all share the feature that they focus on the analysis of frequencies. We start by determining how to test whether the distribution of a categorical variable conforms to an expectation or deviates from that expectation.
