GIScience 2016 Short Paper Proceedings A 3D Virtual Environment for Spatio-Temporal Analysis: Theoretical Approach, Proof of Concept, and User Study R.N. Stewart 1 , C. Wilkerson 1 , E. Ragan 1 , M. Agreda 1 , D. White 1 , S. Duchscherer 1 , J. Piburn 1 Oak Ridge National Laboratory, P.O. Box 2008, 1 Bethel Valley Rd, Oak Ridge, TN 37831 Email: stewartrn@ornl.gov, cwilker7@vols.utk.edu, eragan@tamu.edu, mdagreda@yahoo.com, whiteda1@ornl.gov, piburnjo@ornl.gov Abstract We present a 3D virtual environment leveraging 3D game dynamics for statistical analysis on spatio-temporal (ST) data. We present a theoretical construct, a proof-of-concept implementation (STWorld), and preliminary results from a human-computer interaction (HCI) study. Results include a novel integration of the Unity game engine with open source tools PostGreSQL, R, and the D3.js graphics library. Preliminary results suggest that STWorld may be judged more enjoyable and incur better procedural memory retention than traditional dashboard interfaces. We conclude with findings on gender bias and the gap between gamers and non-gamers. 1. Introduction Applying statistical tools to spatio-temporal (ST) data using traditional statistical packages requires access and understanding of statistical routines, data manipulation, navigation of user interfaces, and basic programming skills. These are obstacles for casual analysts impeding education and application. Web-based tools such as the World SpatioTemporal Analytics and Mapping Project Tool (Stewart et al. 2015) have helped in mitigating these. Alternatively, educational gaming research has demonstrated a strong and positive connection between gameplay, enjoyment, learning, and memory retention of new content and procedures. Jabbar and Felicia (2015) provide a review of the extensive body of work in this domain, and Armes et al. (1999) explored engaging statistical visualizations in virtual 3D space. In this paper, we examine the potential for leveraging first-person gameplay dynamics for analysing spatiotemporal data and raise the following questions: Can we leverage benefits of 3D gaming for the study of ST data? How does efficiency in learning and exploration compare to a traditional interface? How does memory retention compare to a traditional web-tool? Are effects influenced by gender or gaming experience? We present a novel theoretical construct for analytics in 3D space, implement a novel proof- of-concept called STWorld, and report on a small user study aimed at these questions. 2. Theoretical Approach We propose a 3D virtual environment populated with familiar objects whose real-world purpose and behaviour are replaced by abstract data and analytical operations. The user enters a room with a first-person perspective, where cardboard boxes represent PostGreSQL databases, a gun used to query data sets, a magnifying glass that performs statistical analysis, and walls that become interactive graph and map spaces (Figure 1). The intent is for these repurposed objects to serve as comfortable, memorable metaphors for otherwise abstract data operations. By distributing abstract analytics as objects in a room, we aim to improve memory retention of how to locate, access, and analyse data (Ragan et al., 2012). While not a game in the strictest sense, we repurposed foundational game knowledge for ST analytics by adopting