Patterns for visualization evaluation

Abstract

We propose a pattern-based approach to evaluating data visualization: a set of general and reusable solutions to commonly occurring problems in evaluating visualization tools, techniques, and systems. Patterns have had significant impact in a wide array of disciplines, particularly software engineering, and we believe that they provide a powerful lens for characterizing visualization evaluation practices by offering practical, tried-and-tested tips, and tricks that can be adopted immediately. The 20 patterns presented here have also been added to a freely editable Wiki repository. The motivation for creating this evaluation pattern language is to (a) capture and formalize “dark” practices for visualization evaluation not currently recorded in the literature, (b) disseminate these hard-won experiences to researchers and practitioners alike, (c) provide a standardized vocabulary for designing visualization evaluation, and (d) invite the community to add new evaluation patterns to a growing repository of patterns.