- 1.2k Downloads
Students whose complete exposure to statistical methods comes from an introductory statistics class can easily get the impression that, with few exceptions, data analysis is all about examining continuous data, and more specifically, Gaussian (normally) distributed data. Anyone who actually analyzes data, however, quickly learns that there’s more to the world than just the normal distribution. In particular, many of the data problems faced in practice involve categorical data that is, qualitative data where the possible values that a variable can take on form a set of categories. For example, a market researcher might ask respondents whether they use a particular product (“yes” / “no” ), how often they use it (“never” / than once per week” / “between 4 and 10 times per month” / “more than 10 times per month”), how satisfied they are with the product (“completely dissatisfied” / “somewhat dissatisfied” / “neutral” / “somewhat satisfied” / “completely satisfied” ), and what competitors’ products they have also used (“Brand A” / “Brand B” /etc.). The first and last of these examples are nominal (unordered) categorical variables, while the middle two are ordinal (ordered). Such data often are summarized in the form of tables of counts (or contingency tables). Standard Gaussianbased methods are not appropriate for such data, and methods specifically designed for categorical data are necessary.
KeywordsSpace Shuttle Loglinear Model Introductory Statistic Categorical Data Analysis Complete Exposure
Unable to display preview. Download preview PDF.