Visual Representation of Trichotomous Items: A Graphical Approach to Analyzing Preferences

Abstract

Data visualization is a fundamental tool in statistical analysis because it allows hidden structures and patterns to be revealed. Particularly when sets of three-mode categorical items such as “yes / no / don’t know” are used, such as in opinion questionnaires or in rating scales and perceptual surveys, tabular representation is not always straightforward to read, whereas a graph helps to capture differences and similarities with immediacy. This paper proposes a spreadsheet-executable approach to represent three-category items on a graph. Each item is placed in a Cartesian plane according to the proportions of positive (YES) and negative (NO) and neutral (Don’t Know) responses; the result is a visual map that allows one to identify the prevailing response of each item. The paper presents two types of triangular graph and proposes a graphical application of the chi-square test to delineate areas corresponding to heterogeneous distributions of responses within each item.