Arc V iew and S-Plus
Chanter 3 Usability' Evaluation in G eovisualization
This chapter considers another tool, DecisionSite Map Interaction Services (Map IS) by Spotfire, Inc. (Boston, USA), which is m ore in the spirit o f the developm ents described in Table 3.1. DecisionSite is an interactive system for the visual and dynamic exploration o f data designed to support decision-making. It provides a num ber o f graphical displays which include scatterplots, parallel coordinate plots (PCP), histograms. Une graphs, data tables and details-on-demand, am ong others. This product was enhanced by Unkage to A rcExplorer— a stand-alone software appUcation built using ESRI's Map Objects embeddable m apping and GIS com ponents— which provides Umited GIS functionaUty (simple user interface, basic tools and data management). Thus, using a ‘loose’ approach, the functionaUty o f visuaUzation and GIS software was com bined as a m eans to develop an environm ent that supports spatial data users.
This chapter describes a study aimed at providing some evidence on the adequacy o f tools built as a coupUng o f two systems for supporting geovisuaUzation tasks. Although potential disadvantages o f this approach have been discussed, new technologies offered the possibiUty o f building better Unks between software products. T he foUowing sections discuss an evaluation where attention was given to participant behaviour while using the system. Furtherm ore, we discuss some usabiUty problem s with the system and its functionaUty which hindered users’ visual exploration o f spatial data. Therefore, concepts o f usabiUty and usabiUty evaluation are briefly revised next to better explain the purpose and scope o f the study.
3.2
U s a b i l i t y
It has been previously discussed (Chapter 2) that the field o f hum an-com puter interaction (HCI) is concerned with how to im prove users’ w ork processes. It also deals with understanding how people use com puter systems in order to develop or improve their design. T he aim is to m eet user requirem ents so that they can carry out their tasks “safely, effectively and enjoyably” (Preece et al. 1993; ISO 9241). UsabiUty is a sub-field o f HCI that deals with these aspects and the issue o f how to provide functionaUty that can be used effectively and with ease (Nielsen 1993).
UsabiUty describes a system’s fimess for use and its effectiveness for supporting the interaction between hum an operators and their com puter system. W hatever the aim o f the system, it should faciUtate users’ tasks or enhance their perform ance by providing appropriate analysis tools. Therefore, it can be specified in terms o f how weU intended users can m aster and perform tasks using the system (Butler 1996). From a user-centred
Usability' Evaluation in G eovisualization
perspective, the aim o f a software application is to satisfy users by providing relevant functionality which they find fit for purpose and can enhance their perform ance. “Testing usability means making sure that people can find and w ork with [a system ’s] functions to m eet their needs” (Dumas and Redish 1999: 4). Usability is therefore concerned with estabhshing w hether people can use a com puter-based tool to com plete their tasks effectively and with sufficient ease.
Usability engineering provides systematic m ethods to measure a system’s usability in terms of its learnabüity, efficiency, effectiveness, memorability, error rate and user satisfaction (Butler 1996; Nielsen 1993). This applies to all aspects o f a system’s user interface, which can be defined as the m edium through which a user interacts and communicates with the com puter (Preece et a l 1993, Nielsen 1993). T he ease o f learning a system is measured as the time it takes a potential user to “reach a specified level o f proficiency in using it” (Nielsen 1993: 28). Efficiency refers to the level o f productivity that the user m ust achieve once the system has been learned. Effectiveness on the other hand is “the accuracy and com pleteness with which users achieve specified goals” (Thomas 1996: 109). Memorability measures how easily a system is rem em bered either after a period o f no t using it or by casual users. An error in this context is defined as any action that does no t accomplish the desired goal (Nielsen 1993), which can be objectively observed and counted to obtain a measure o f a system ’s error rate. Satisfaction refers to how pleasant the system is to use. Preece et a i (1994) also consider throughput, flexibility and user attitude towards the system as aspects that m ust be addressed in assessing a system’s usability. Ease o f use or throughput is comparable to the concept o f efficiency and error rate as it is defined as “the tasks accomplished by experienced users, the speed o f task execution and the errors m ade” (Preece et a l 1994: 401). Flexibility refers to the extent to which the system can accom m odate tasks it was no t originally planned for. Attitude is com parable to user satisfaction or the “com fort and acceptability o f use” (Thom as 1996: 109). Ease o f use, satisfaction and learnabihty were the usability aspects investigated with the study discussed below.
UsabiUty evaluation involves recording both objective data on user interaction with the system during perform ance o f task scenarios and subjective data for diagnosing problems. Objective data may include the time spent on tasks, patterns o f operations during the test sessions, or errors. Empirical data can be gathered either in a usabiUty lab o r in user workplaces. D ata can be recorded using multiple cameras and program s for recording and playing back on-screen user interaction. Subjective data include verbal reports that can be recorded either during the evaluation or foUowing the session. These techniques aUow
Usabilirv Evaluation in G eovisualization
m onitoring o f user thoughts as well as the problem s they are aware o f having (Butler 1996). Similarly, qualitative interviews and questionnaires using rating scales can aid in eliciting user satisfaction with the software and perceptions on ease o f use and usefulness o f a system.
3.3
U s a b i l i t y Study
Usability evaluation normally runs in parallel with a system’s developm ent cycle. In research, it can help to assess the suitability o f a concept that is being im plem ented using a computer- based tool, as well as the suitability o f the design for supporting some task. This section discusses results from the first study conducted regarding the use o f GIS-based tools for exploring spatial data visually and interactively. Its purpose was to provide a qualitative description o f the problem s that participants had while perform ing the tasks as well as identifying functionality that enhanced their work process. In other words, the discussion that follows points at strengths and weaknesses o f the system and concepts tested. These findings were used in a subsequent developm ent iteration to build an environm ent that overcom es some o f the limitations discussed next.
3.3.1
Goals
T he main aim o f this study was to evaluate w hether the functionality provided by the coupling between the lightweight GIS and DecisionSite served to create an environm ent that supported the exploration o f spatial data. Ultimately, the aim was to identify the potential o f linking systems to support geovisualization tasks for spatial data users. In addition, the study was designed so that it would be possible to review participants’ work processes, how they thought and behaved when using the system.
It has been argued that dynamic displays facilitate perception (Peterson 1994; Ware 2000). Dynamic graphics are usually interpreted as supporting interaction or direct manipulation o f graphical elements on a com puter screen that provides an almost instantaneous change o f elements (Becker and Cleveland 1987). T hose that prom ote the experim entation with data attribute com binations and transform ations, see dynamic displays as essential for exploratory analysis (DiBiase et al. 1994; Asche and H errm ann 1994; M acEachren 1994; O penshaw et at.
1994). Thus, it was hypothesised that the main shortcom ing o f DecisionSite Map IS (Figure 3.1) was the lack o f a real-time link between the two software com ponents. This refers to the value o f providing linked multiple graphical displays that support simultaneous or dynamic brushing as an im portant feature o f systems that allow the exploration o f data (Spence 2001; M acEachren 1995; M onm onier 1989).
. ' h a p r c r I ' . i l t i h n l'.\ . i l u . i r i ' i i l i n C i c ' A i H i . i l i / . U i ' i n
O S p u tf iie D e c i s io n s i t e 6 .2 - y 1 _ S H 0 R T ,x ls
Fte Ed* View ViMjafaabon lo o k ÿ/indow H *
O a ^ 4 ^
[^py #
010 IlKtia 1,529708.203 f 18.350.438 * ;9 ;* YKMUMMW U L»wti»-i 1 0 -1 3 . m % □ S 'I ^ :E ÏKL Æ J
0 5 1 0 15 20 25 X 35 40 45 5 0 K e O G 5 70 75 80 85 I*! HHNK>\”’ | B S c M te r Pin# IK 60 70 80 r* ] HHNOCA |~.ÿ. Scaltei Pk* CHILDN 0 0=1 COUNCL 0 96J
DETCHD 0 u m 96 B HHNOCA 0 ra— B j ü Column V a ^ ID ‘ 1556 EASTING 359336 NORTHING 173229 BIGHHD 0 BIGACC BUYING BOUGHT CHILDN COUNCL DETCHD HHNOCA HHTCAR QUALML OLDADL OLDPSN (94 ;0) J 844(X<of844fcscofd*vBibte(100 0 0 * l1 m a ik edFigure 3.1 DecisionSite Map IS (Spotfire, Boston). To the left of the image is ArcExplorer and DecisionSite appears to the right. The points on the map correspond to a set of records selected in DecisionSite which have been ‘transferred’ to the GIS as an image and laid on top
o f the map.
In DecisionSite Map IS, however, there is no simultaneous h i^ ig h tin g between the aspatial graphical displays (such as a scatterplot, PCP, histogram or line graph) and the map. In order to see how data in the former views relate to a map, the user m ust transfer the data from the particular view in DecisionSite to the map in ArcExplorer. In other words, if the user wants to associate attribute values represented on a scatterplot in DecisionSite to their actual location in space, the data m ust be selected on the scatterplot and then transferred from DecisionSite to the GIS. The selection is then represented on the map in the form o f an image file that plots a point on the corresponding geographical coordinates (see the red points on the map in Figure 3.1). N o t only is this form o f communication lacking a dynamic link between the two com ponents, it cannot be customised differently as ArcExplorer only supports data browsing and does not provide a development environment. It was therefore o f interest to find whether the available functionality sufficed for enabling the system as a spatial decision-making tool. In other words, to test the im portance o f having a dynamic link for attempting geovisualization tasks.
Usahilir\' Evaluation in G eovisualization
In addition, the identification o f potential problem s with the current im plem entation was a main goal since this study was a preliminary enquiry into the characteristics that similar software tools should provide. In this study, particular attention was given to measuring user perform ance with the system in terms o f their ability to generate hypotheses about the data during the tasks provided. It was n o t expected that participants would follow a strict series o f steps to solve the tasks but rather that they obtained plausible answers. T he test design and choice o f participants aimed at facilitating the evaluation o f these goals.