Correspondence Analysis: Theory, Practice and New Strategies

7Ordered Non-Symmetrical Correspondence Analysis

7.1 Introduction

Since the early 1980s, the statistical analysis of categorical data has been constantly evolving. In particular, there has been increasing emphasis placed on differentiating data analysis for ordered and unordered categories. A variable is called nominal when its categories differ in quality, not quantity, and when they are unordered. Examples of such variables are professional status, gender, eye colour, political party, brand, product and religion. Conversely, a variable with an ordered categorical scale is called ordinal. Examples of ordinal categorical variables and their ordered scales are used plentifully in scientific investigations and social sciences, with applications ranging from sensory evaluation experiments to medical and market research evaluations for measuring diagnostic ratings, attitudes and opinions. People's opinions, educational attainment, customer and job satisfaction, quality of life and diagnostic rating, among others, are commonly measured on an ordinal scale. In these cases, responses are often measured using a Likert scale (Frey et al., 2000), which is commonly used for assessing attitudes. In fact, an item on a Likert scale consists of two parts: a statement of an attitude and a scale on which people express their agreement with that statement. For example, the level of agreement, or disagreement, of an issue may be measured using ordered responses that can be balanced. These may ...