Investigating the Methodology of Warning Symbol Design

Abstract

The present study is a continuation of work done by Dr. Adam Piper in his 2010 dissertation on warning symbol design. Warning symbols are important because, if designed correctly, they communicate information about a hazard quickly and effectively so that an individual can react accordingly if exposed to it. Warning symbols themselves, as an alternate to warning messages, have the added benefit of a potential to traverse language barriers and add comprehensibility to a warning message. Several methods of symbol design have been proposed, but none that utilize the input from the general population from start to finish. Current methods include the designer method, the production method, and the focus group method. The present experiment attempts to expand upon these methods by utilizing the general population throughout the study. A group of general population participants was recruited to draw a symbol based on a given safety referent. These drawings were then semantically annotated by a naive set of individuals representing the general population. The resulting lists of attributes were entered into matrices and clustered via a word clustering software program developed by Feinberg (2014) called Wordle, which identified significant attributes through the size of each attribute in a word cloud (i.e., the larger the word appeared in the cloud, the more often it occurred in the matrices). This procedure resulted in the same general core attribute lists for both referents when compared to the list produced by a panel of safety experts in a previous study.

The most significant limitation of this study was the fact that it took a significant amount of time to check each individual matrix for accuracy, and subsequently consolidate the similar terms into a workable number of attributes for clustering. In the future, an updated method of data consolidation must be employed to reduce the amount of time required for pre-data analysis. Preliminary data analysis revealed that similar attribute lists were created via both the general population method and expert panel method. Therefore, it may be advantageous to utilize an amended version of the general population method presented in this study in order to reduce costs of symbol creation.