A Comparison of Map vs. Text Directions for a Handheld Device in a Campus Setting: A Pilot Study Liz Atwater Jason Burke Andrea Kirk Department of Psychology George Mason University Institute for Systems Research University of Maryland Department of Computer Science University of Maryland December 2001 Map vs. Text Directions Which one is more effective? – Less time, less errors Does route complexity have an effect? Use by pedestrians instead of drivers – Lack of landmarks – No street names, etc. Rover Context-aware, location-aware – Location awareness via GPS, RF, IR, etc. Provides information depending on: – User profile – Device profile – Location – Context Useful in many domains – Tourism – Commerce Background Research Williams studies (1999) – pilots finding nearest airport using maps or text – Maps are faster and more accurate – ERF tasks had better results with track-up – WRF tasks had better results with north-up Aretz,1991 – ERF vs. WRF – Ego-centered frame track-up – World-centered frame north-up Butz, 2001 – landmarks at key decision points Experiment – Hypotheses Hypotheses: – Null: There is no statistical difference between completion time, consultation time and number of errors between text and map directions, regardless of route complexity. – H1: Users will complete the tasks faster using map directions. – H2:Users will make fewer errors using map directions. – H3: Users will need less consultation time using text directions. – H4: Completion time will rise with increasing route complexity. Experiment – Variables IVs & Treatments – Direction type: map vs. text – Route complexity: low, medium, high • Low: 3 decision points, 893 ft • Medium: 5 decision points, 897 ft • High: 7 decision points, 883 ft DVs – Completion time – Consulting time – Errors Experiment – Materials Subjects – 7 male, 5 female – Undergrad & grad UMCP students Other materials – Pre & post-task questionnaires – VZ-2 Experiment – Tasks Navigate 3 routes using directions Within-subjects for routes Between-subjects for direction type 2 stopwatches Route permutations: 123 213 312 132 231 321 Screen Shots Text Implementation Map Implementation Results – Completion Time Main effect for route: significant Main effect for direction: ns Interaction effect: ns Completion Time 400 350 Time (s) 300 250 Map 200 Text 150 100 50 0 High Medium Route com plexity Low Results – Consultation Time Main effect for route: significant Main effect for direction: ns Interaction effect: ns Device Consulting Time 120 100 Time (s) 80 60 Map 40 Text 20 0 -20 High Medium Route Com plexity Low Results – Errors Main effect for route: significant Main effect for direction: ns Interaction effect: ns Number of Errors 3.5 Number of errors 3 2.5 2 Map 1.5 Text 1 0.5 0 -0.5 High Medium Route Com plexity Low Observations Learning seemed to have a significant effect on the results Most errors occurred at non-dead ends People are different – Huge variance in user performance in both map and text implementations – Difficulty judging distances in text version – Rotate map for track-up bearings – Looking ahead caused problems Conclusions Need many more subjects Text directions are difficult to describe in college campus environment Feedback from “real” context-aware equipment could improve performance Track-up display for map could decrease orientation time Hybrid to accommodate variations in user cognitive strengths