Abstract
We formulate multiple correspondence analysis (MCA) as a nonlinear multivariate analysis method that integrates ideas from multidimensional scaling. MCA is introduced as a graphical technique that minimizes distances between connecting points in a graph plot. We use this geometrical approach to show how questions posed of categorical marketing research data may be answered with MCA in terms of closeness. We introduce two new displays, the star plot and line plot, which help illustrate the primary geometric features of MCA and enhance interpretation. Out approach, which extends Gifi (1981, 1990), emphasizes easy-to-interpret and managerially relevant MCA maps. Multiple correspondence analysis (MCA) is weil on its way to becoming a popular tool in marketing research (Hoffman & Franke, 1986). For example, Green, Krieger, and Carroll (1987) use MCA to analyze the relationship between consumers' choice profile predictions from a conjoint task and consumer demographic characteristics. In a similar vein, Kaciak and Louviere (1990) illustrate how MCA may be used to analyze data from discrete choice experiments. Carroll and Green (1988) apply individual differences MDS to normalized Burt matrices (a principal data matrix in MCA) to determine the relationship between consumer demographics and automobile characteristics with respect to number of cars in the household. More recently, Valette-Florence and Rapacchi (1991) perform an MCA on the attributes-consequences-values matrix derived from a laddering task to construct a product positioning map and Hoffman and Batra (1991) apply MCA to study the association between television program types and audience viewing behaviors.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call