Virtual Reality vs. 2D Visualizations for Software Ecosystem Dependency Analysis – A Controlled Experiment

Abstract

Background/Context: Data is typically visualized using 2-D on- screen tools. With the advent of devices capable of creating 3D and Virtual Reality (VR) scenes, there is a growing interest in exploring these technologies for data visualization, particularly for complex data like software dependencies. Despite this interest, there is lim- ited evidence comparing VR with traditional 2D on-screen tools for such visualizations. Objective/Aim: This registered report aims to determine whether comprehension of software ecosystem dependencies, visualized through their metrics, is better when presented in VR scenes com- pared to 2D screens. Specifically, we seek to evaluate if answers obtained through VR visualizations are more accurate and if it takes less time to derive these answers compared to traditional 2D on-screen tools. Method: We will conduct an experiment with volunteers from various backgrounds, using two setups: a 2D on-screen tool and a VR scene created with A-Frame. The data will focus on web projects using the Node Package Manager (NPM) registry. Subjects will answer a series of questions in both setups, presented in random order. We will statistically analyze the correctness and the time taken for their answers to compare the two visualization methods