Modeling Consumer Opinion Using RIDIT and Grey Relational Analysis

Modeling Consumer Opinion Using RIDIT and Grey Relational Analysis

Rohit Vishal Kumar (Xavier Institute of Social Service, India) and Subhajit Bhattacharyya (Xavier Institute of Social Service, India)
DOI: 10.4018/978-1-5225-0997-4.ch010
OnDemand PDF Download:
No Current Special Offers


In order to understand consumers, researchers are forced to gather primary data on Likert scale. Such data is usually considered as ordinal or at best interval scaled data. One key requirement in research is to identify components which have high individual contribution to understanding the research problem. Hence the concept of ranking of the components comes under consideration. Most of the ranking techniques are based on simplistic mean ranks or overtly complicated methods. In this chapter the authors highlight two techniques - Grey Relational Analysis (GRA) and RIDIT - for the purpose. In this chapter the authors explain the techniques of the two methods and then try to show the simplicity and efficiency of GRA and RIDIT algorithms in analyzing a commonly available dataset. The outcome of the GRA and RIDIT analysis is also compared with the commonly used techniques and the authors would examine if GRA and RIDIT does a better job at ranking data than the commonly used techniques.
Chapter Preview

Review Of Literature

In 1958, Irwin D. J. Bross had propounded the use of RIDIT (“Relative to an Identified Distribution”) analysis for dealing with ordinal variables. Ordinal variables can be subjective scales such as – severe, moderate, or minor – or may take a numerical form. The measurement system relies heavily on the skills of the researcher involved in the measurement of the variable. According to Bross, these variables may not be adequately analyzed by Chi Square and t-test as they may not be appropriate due to assumption of a parametric distribution used by the test. (Bross, 1958). RIDIT calculates the probability that a given observation is better than a visible observation from a reference distribution. The outcomes from the RIDIT analysis can be used to arrange Likert scale items either in an ascending or in a descending order based on importance. This can also examine the relations among scale items in terms of degrees of significance or agreement. RIDIT approach is “distribution free” in nature, because there is no need of assumption about the distribution of the population under the study (Bross, 1958; Fleiss, Levin, & Paik, 2003).

Complete Chapter List

Search this Book: