Usability Evaluation of Centered Time Cartograms

Open Geosci. 2016; 8:337–359

Research Article Open Access

Rehmat Ullah*, Eskedar Zelalem Mengistu, C.P.J.M. van Elzakker, and Menno-Jan Kraak Usability evaluation of centered time cartograms

DOI 10.1515/geo-2016-0035 Keywords: centered time cartograms; usability; Dutch rail- Received Nov 16, 2015; accepted Mar 21, 2016 way network Abstract: A time cartogram visualizes travelling-times between locations. It replaces the geographic distance by time distance and distorts the underlying map accordingly. 1 Introduction By distorting the map, time cartograms may give a more in- tuitive and clear picture of travelling-times. The distortion A time cartogram visualizes travelling-times between loca- of the map, however, can make time cartograms harder tions. It replaces the geographic distance by time distance to recognize and use. Although cartograms are becom- and distorts the underlying map accordingly. Two types of ing widespread in use, very little is known about their time cartograms exist: centered and non-centered ones. A usability. This study focuses on the usability of centered centered time cartogram visualizes travelling-times from time cartograms: time cartograms that visualize travelling- a fixed starting location to other destinations inare- times from a fixed starting location to other destina- gion [1–4], while a non-centered time cartogram visualizes tions in a region. We created several centered time car- travelling-times between all pairs of locations [5–7]. tograms to answer spatio-temporal questions related to The use of cartograms is growing due to their capti- the Dutch railway network. Two experiments were per- vating design [8] and also due to the availability of meth- formed: a laboratory test and an online survey. In the lab- ods to create them automatically [e.g. 7, 9–12]. The simple oratory test, we used eye-tracking, thinking aloud, and reason is because they show what researchers need to see video-recording to compare four different designs of cen- and researchers outside of cartography do not find these tered time cartograms to find out which one (or com- hard to understand. The United Kingdom’s Chief Medical bination) of these performs better in answering spatio- Officer used cartograms throughout her annual report to temporal questions and thus, to establish a favorable de- show disease patterns across the UK [13] – whereas a nor- sign strategy for these cartograms. In the online survey, mal map would be much less useful for such projections. centered time cartograms were evaluated against a geo- Dent [14] described cartograms as “innovative,” “interest- graphic and schematic map for accuracy, response time, ing,” and “stylish”. According to Dorling [15, p. 53], “What- and preference. The first experiment suggested that among ever you choose to use cartograms for, from studying par- various designs, the centered time cartogram with empha- ticipation in elections, to the spread of a disease, or the sized railroads is the most preferred design and the cen- social structure of a country, the very different perspec- tered time cartogram without railroads is the least pre- tives they show are likely to alter the way you imagine ferred. The second experiment indicated that overall, cen- the processes behind these patterns to be operating.” Koc- tered time cartograms perform better than the two other moud and House [16, p. 236] stated that “The cartogram solutions in performing spatio-temporal tasks, particu- is a useful tool for visualizing the geographical distribu- larly when the task has a dominant time-related compo- tion of ‘routine’ data in a variety of disciplines, includ- nent. ing politics, social demographics, epidemiology, and busi- ness.” Shimizu and Inoue [17] reported that cartograms are a highly effective method for visually representing statis- *Corresponding Author: Rehmat Ullah: Department of Geo- tical data. Similarly, Wu and Hung [18] found cartograms Information Processing, Faculty of Geo-Information Science and very useful for creating strong visual impacts. Earth Observation (ITC), University of Twente, P.O. Box 217, 7500 Despite the many positive claims, some researchers AE, Enschede, the Netherlands; Department of Computer Systems Engineering, University of Engineering and Technology, Peshawar, are doubtful of the effectiveness of cartograms because Pakistan; Email: [email protected]; Tel: +31 53 4874 373 of their distortion in shape [e.g. 19–22]. In the literature, Eskedar Zelalem Mengistu, C.P.J.M. van Elzakker, Menno-Jan we find only a few studies on the usability of areacar- Kraak: Department of Geo-Information Processing, Faculty of Geo- tograms [e.g. 23–25], and only limited research has been Information Science and Earth Observation (ITC), University of reported on the usability of time cartograms [26, 27]. In Twente, P.O. Box 217, 7500 AE, Enschede, the Netherlands

Figure 1: Four alternative designs of the CTC from Enschede. (a) with railroads (WR). (b) without railroads (NR). (c) with emphasized railroads (ER). (d) with emphasized time-circles (EC). The concentric circles indicate travelling-times in increments of 10 minutes from the city of Enschede to all other stations in the province of Overijssel. recent years, various methods have been developed for recording to compare four alternative designs of CTCs (i.e. constructing time cartograms [e.g. 1–3, 5, 17, 18, 28–36], with railroads, without railroads, with emphasized rail- however, the practical value of such cartograms has not roads, and with emphasized time-circles – see Figure 1) yet been established. This motivated us to conduct a de- to test whether railroads add to the readability of CTCs (or tailed and systematic usability evaluation of centered time provide a false impression of time) and to examine the im- cartograms (hereafter abbreviated as CTC) to investigate pact of visual hierarchy on the readability of CTCs. These whether they are indeed useful. designs are abbreviated WR, NR, ER, and EC, respectively. Several CTCs were created to answer spatio-temporal Four groups of participants were asked to perform spatio- questions related to the Dutch railway network. To cre- temporal tasks using these designs and report which one ate these cartograms, we used the method proposed by of these provided an effective and efficient visual repre- Ullah and Kraak [2]. Two user studies were conducted: sentation and best satisfied their requirements. Forty re- a laboratory test and an online survey. In the laboratory sponses were collected and analyzed. In the online survey, test, we used eye-tracking, thinking aloud, and video- CTCs were evaluated against geographic and schematic Usability evaluation of centered time cartograms Ë 339 maps by asking spatio-temporal questions to determine same for all TPs. To avoid potential bias, the tasks and cor- whether they can answer some questions better than the responding maps were presented in random order. Each other two solutions. Eighty-eight persons participated in test was then followed by a questionnaire where the TPs the survey. ranked each design on a Likert scale from ‘strongly agree’ The remainder of this paper is organized as follows: to ‘strongly disagree’, considering three usability aspects: Section 2 discusses the design, implementation, results, (a) pleasant to see, (b) easy to understand, and (c) use and conclusions of the laboratory test. Section 3 reports without confusion. The questionnaire aimed to capture the on the design, implementation, results, and conclusions feelings of the TPs about each design which could then be of the online survey. Section 4 draws final conclusions and interpreted in combination with their suggestions during looks at possibilities for future work. the process of thinking aloud. Potential TPs were invited to participate through an email. They were asked to complete an online survey about 2 The laboratory test their profile and whether they were willing to participate in the actual user test. Profile information was used to al- locate participants to particular test groups (see Figure 3 2.1 Objective, design, and implementation below). Afterwards, a separate invitation was sent to all willing participants to participate in the actual user test. In this research, a laboratory test was conducted applying Test completion time for each TP was on average 40 min- eye-tracking, thinking aloud, and video-recording to com- utes (10 minutes for introduction + 5 minutes for warm-up pare four different CTC designs (see Figure 1). The test ob- tasks + 25 minutes for the actual test). jectives comprised: In total, 48 persons indicated their willingness to par- • To find out which alternative CTC design (or com- ticipate in the user test by completing the profile survey, bination) performs better when answering spatio- however, among those 48 respondents 8 failed to arrive temporal tasks for testing. Among those final 40 test participants, 11 listed • To discover which usability issues are involved when Bachelor, 23 Master and 6 PhD as their highest educational test participants are working with alternative CTC degree. The participants, from 25 different nationalities, designs were postgraduate students and staff members of the Fac- • To investigate whether railroads add to the readabil- ulty of Geo-Information Science and Earth Observation at ity of CTCs (or provide a false impression of time) the University of Twente in the Netherlands and were thus • To determine whether the distortion in CTCs indeed knowledgeable about maps. Participants had varied aca- affects their readability demic backgrounds, with the majority from GIS and Re- mote Sensing fields and most were occasional users of For the laboratory evaluation, CTCs were constructed Dutch trains. Approximately half of the participants knew for the railway network in the Dutch province of Overijs- about cartograms in general, but had limited or no knowl- sel. This network has 33 train stations. To prevent test re- edge about CTCs. sults being biased by a particular CTC shape and to in- Participants play a key role in usability studies [37, 38]. vestigate whether a distortion in CTC shape affects read- Accordingly, one of the pre-test preparations for the labo- ability, six stations (i.e. Steenwijk, Zwolle, Gramsbergen, ratory test was the collection of the participants’ profiles. Enschede, Raalte, and Deventer) were selected as starting We used thirteen criteria to assign participants to particu- stations (see Figure 2). The four different CTC designs were lar test groups with the intention to make the groups as presented to test participants (TPs) in four groups. The similar as possible in their composition. The criteria in- first group worked with CTCs with railroads, the second cluded – given in order of decreasing priority – experi- with CTCs without railroads, the third with CTCs with em- ence of travelling using Dutch trains, knowledge about the phasized railroads, and the fourth with CTCs with empha- Dutch railway network and the geography of the Nether- sized time-circles. Each group performed eight different lands, duration of stay in the Netherlands, map-use and spatio-temporal tasks (listed in Table 1). The tasks were de- map-making experience, knowledge about cartograms, signed to be real-world to allow testing of the CTCs from a usual method of train trip planning, department or course, real-world approach and provide an avenue to understand educational level and background, country of origin, age, the practical value of these cartograms. The first six tasks and gender. Highest priority was given to the user charac- were different (but with similar complexity) for each start- teristics that can influence the test results the most. Using ing station, while the seventh and eighth tasks were the these criteria TPs were divided into four groups (as illus- 340 Ë R. Ullah et al.

Table 1: List of spatio-temporal tasks for the laboratory evaluation.

Task Nature Description T1 TEMPORAL Mark is a new graduate student at ITC in Enschede. This Monday he wants to go to the IND in Zwolle to collect his residence permit. The documents collection timings are from 11:30 am to 12:30 pm. At what time should he leave from Enschede to reach Zwolle at 11:00 am in the shortest time possible? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T2 TEMPORAL Robert is in Enschede to attend a 3-day workshop at ITC. The workshop runs from 09.00 am to 05.00 pm daily. As he has never been in the Netherlands before, on one of the workshop days he wants to visit another place that is close to Enschede. Could you please list all cities to Robert that are reachable by train within 20 minutes from Enschede? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T3 TEMPORAL Which station is closer to Enschede in time: Oldenzaal or Almelo? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T4 SPATIO- Which path takes the longest distance to reach Dalfsen from Enschede: via Mariënberg or via TEMPORAL Raalte? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T5 SPATIAL What is the number of possible paths (backtracking – you cannot visit a station twice – is not allowed) to reach Wierden from Enschede? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T6 SPATIO- List the intermediate stations while travelling (via the shortest path in time) from Enschede to TEMPORAL Goor. How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T7 SPATIO- Smith needs to travel frequently between Enschede and Zwolle for his job assignments. Where TEMPORAL should he rent an apartment so that he is time-wise evenly close to both places? How well did the provided cartogram assist you in performing this task? ○ very good ○ good ○ average ○ poor ○ very poor

T8 SPATIO- Given the four alternative visualizations of the CTC with different designs, perform the below TEMPORAL task using all four designs and which alternative would you prefer to use to perform this task and why? ○ 1 ○ 2 ○ 3 ○ 4 List the intermediate stations while travelling (via the shortest path in time) from Raalte to Holten. Usability evaluation of centered time cartograms Ë 341

Figure 2: CTCs constructed for six different stations of the Overijssel railway network. (a) Steenwijk. (b) Zwolle. (c) Gramsbergen. (d) En- schede. (e) Raalte. (f) Deventer. The concentric circles show travelling-times in increments of 10 minutes from the starting station. trated in Figure 3) using a technique called “matrice or- four alternative design solutions of the CTC from six dif- donnable” [39]. ferent starting stations in random order. Tobii Studio is A Tobii X60 eye-tracker (with peripherals) and Tobii a platform that is able to achieve integrative recording of Studio software 2.2 was used to prepare and present the eye-tracking, video, sound, and screen logging and these 342 Ë R. Ullah et al.

Figure 3: Visual grouping of test participants into four groups. outcomes may be analyzed in various (statistical and vi- ment was programmed for uninterrupted sound (for the sual) ways. The user tests were conducted in the ITC us- thinking aloud process) and video-recording of the whole ability laboratory which is a well-equipped laboratory de- testing procedure. signed for user research (Figure 4). The testing environ- Usability evaluation of centered time cartograms Ë 343

Figure 4: Test environment for the laboratory evaluation.

Three pilot tests were conducted before the actual user tests to assess the test scenarios for any potential prob- Figure 5: Correctness, satisfaction, and response time measurements for the temporal task T1 for the four CTC designs – i.e. with lems. Three experts in usability testing attended the pilot railroads (WR), without railroads (NR), with emphasized railroads tests to evaluate the presentation of the CTCs and the over- (ER), and with emphasized time-circles (EC). Each box plot shows all test set up. Their comments were used to detect and the median (line within the box), the 25th percentile (lower end of remedy any potential test set up problems before the final the box), the 75th percentile (upper end of the box), and 1st to 99th test execution. percentile (solid lines). The dots show the response time of each participant.

2.2 Results and discussion better than the other cartograms. The results were consis- tent with our hypothesis: NR showed higher correctness. From the huge test records of the Tobii software that in- In contrast, the correctness of ER was lower while WR and cluded visualizations such as gaze plots, heat maps, and EC performed equally well. However, the TPs who worked statistical graphics for area of interests (AOIs), the test with NR took more time to complete the task. The thinking results of the research were organized and prepared for aloud data showed that the reason for a longer response analysis. Result gathering and preparation were comple- time was that the TPs spent a greater deal of time locating mented by the verbal and action protocol data of the think- the route from source to destination. On the other hand the ing aloud process. TPs who worked with WR, ER, and EC took less time on av- The processing of test results was primarily directed erage to perform the task as the railroads helped TPs to towards the analysis of usability aspects such as correct- quickly locate the route and time-circles to calculate the ness, response time, and satisfaction. During task execu- travelling-time from source to destination. A majority of tion in the experiment, TPs were asked to either choose an the TPs liked all four alternative designs for this task. answer from the given choices or to write their response if The results for temporal task T2 are displayed in Fig- they felt their answer was not listed. These data were aug- ure 6. In this task the TPs had to list all stations that are mented with the help of the video-recording, reasoning, reachable within 20 minutes by train from a source sta- and eye-movements. The results for each task (see Table 1) tion. Again, the railroads were not needed for this task as are discussed in detail below. TPs were only required to list all stations located inside The results for temporal task T1 are presented in Fig- the 20-minute circle. For this task, NR was again hypothe- ure 5. In this task the TPs had to state the departure time sized to perform better than the rest. Accordingly, the cor- from a source station to reach a destination at 11:00 am rectness of NR was higher. The three other cartograms per- in the shortest time possible. To perform this task, TPs formed equally well as two TPs only answered the ques- did not require the railroads and only needed the time- tion partially. In those instances, the TPs either did not circles to calculate travelling-time from source to destina- list all stations reachable within 20 minutes or listed addi- tion. Accordingly, for this task NR was expected to perform tional incorrect stations (i.e. requiring more than 20 min- 344 Ë R. Ullah et al.

Figure 6: Correctness, satisfaction, and response time measure- Figure 7: Correctness, satisfaction, and response time measurements for the temporal task T2 for the four CTC designs – i.e. with ments for the temporal task T3 for the four CTC designs – i.e. with railroads (WR), without railroads (NR), with emphasized railroads railroads (WR), without railroads (NR), with emphasized railroads (ER), and with emphasized time-circles (EC). Each box plot shows (ER), and with emphasized time-circles (EC). Each box plot shows the median (line within the box), the 25th percentile (lower end of the median (line within the box), the 25th percentile (lower end of the box), the 75th percentile (upper end of the box), and 1st to 99th the box), the 75th percentile (upper end of the box), and 1st to 99th percentile (solid lines). The dots show the response time of each percentile (solid lines). The dots show the response time of each participant. participant. utes of travelling-time). The average response time of NR the task easily, particularly using WR, NR, and ER. On av- was longer, followed by ER and then WR and EC in or- erage, the response times of all four groups were similar. der. However, the average response times for all four car- The majority of TPs found all four designs satisfactory for tograms were quite comparable. Most TPs ranked all four this task. designs as ‘good’ and ‘very good’. The results for spatio-temporal task T4 are summa- Figure 7 depicts the results for temporal task T3. In rized in Figure 8. In this task the TPs had to determine this task the TPs were required to find the station that was the path which takes the longest distance to reach the closer in time to the source station among two given sta- destination from the source. To perform this task, TPs re- tions. To perform this task, TPs again only needed the time- quired both the railroads (to find the routes from source to circles to calculate travelling-times to the two stations. For destination) and the time-circles (to calculate the longest this task, we expected NR to perform better than the other route). For this task, ER and EC were expected to perform cartograms. In CTCs, travelling-times are indicated by con- better than WR and NR. CTCs visualize travelling-times centric time-circles while the railroads only show connec- from the starting station to all other stations by replac- tivity between stations. This task was specifically devel- ing the geographic distance with time distance. Distance oped to determine whether the railroads provided a false in CTCs always means time (time distance). Here, we pur- impression of travelling-times because tracks do not al- posely asked TPs to determine the path which takes the ways follow a direct time from circle to circle. For this pur- longest distance to investigate their understanding of the pose a pair of stations was carefully selected for each CTC CTC and to assess whether they found the term ‘distance’ where the railroads could be misleading. For example, we confusing. Since most TPs were frequent users of maps chose Oldenzaal and Almelo for the CTC from Enschede. in general but were not knowledgeable about CTCs, they In this instance if the railroad is followed, Almelo seems found the task confusing. In each group, many TPs either closer to Enschede than Oldenzaal. But this is not correct. stated that the task cannot be completed with the provided Oldenzaal is closer to the 20-minute circle than Almelo. CTC or performed the task wrongly by misinterpreting the As evident from the correctness graph however, the rail- distance for physical distance. Only TPs who were already roads did not cause confusion. All participants performed knowledgeable about CTCs understood the task well and Usability evaluation of centered time cartograms Ë 345

Figure 8: Correctness, satisfaction, and response time measure- Figure 9: Correctness, satisfaction, and response time measurements for the spatio-temporal task T4 for the four CTC designs – ments for the spatial task T5 for the four CTC designs – i.e. with i.e. with railroads (WR), without railroads (NR), with emphasized railroads (WR), without railroads (NR), with emphasized railroads railroads (ER), and with emphasized time-circles (EC). Each box plot (ER), and with emphasized time-circles (EC). Each box plot shows shows the median (line within the box), the 25th percentile (lower the median (line within the box), the 25th percentile (lower end of end of the box), the 75th percentile (upper end of the box), and 1st the box), the 75th percentile (upper end of the box), and 1st to 99th to 99th percentile (solid lines). The dots show the response time of percentile (solid lines). The dots show the response time of each each participant. participant. provided clear answers. Due to the confusion all four car- ever, one TP in WR spent a considerable length of time to tograms performed poorly for this task, however the cor- complete the task successfully. As apparent from the sat- rectness of ER and EC was slightly better compared to the isfaction graph, the TPs were quite satisfied with WR and two other CTCs. The TPs who worked with NR took more less satisfied with NR. time to perform the task, as they spent a considerable time The results for spatio-temporal task T6 are illustrated in identifying the paths from source to destination. The in Figure 10. In this task the TPs were required to list all in- TPs in the other three groups took less time to complete termediate stations while travelling (via the shortest path the task. The satisfaction results for this task were aligned in time) from a source station to a destination. In this task, with the results of correctness and accordingly most TPs TPs needed both the railroads and time-circles to find the ranked the cartograms between ‘very poor’ and ‘average’. shortest path from source to destination and list all sta- Figure 9 shows the results for spatial task T5. In this tions along it. Unlike in T4, we specifically asked TPs to list task the TPs had to identify the number of all possible the intermediate stations along the shortest path in time to paths (without back-tracking) from a source to a destina- see how they reacted to the term ‘time’. In contrast to T4, tion. To perform this task, TPs only needed the railroads TPs did not face any difficulty and understood the task very to see the connectivity between stations in order to locate well. For this task, we expected ER and EC to perform bet- the paths from source to destination. For this task, WR, ER, ter and the results were in line with our hypothesis as the and EC were hypothesized to perform better. The majority correctness of ER and EC was higher and the correctness of of TPs understood the task and provided answers without the other two cartograms (WR and NR) lower. Again, this confusion and the results were aligned with our hypothe- task could not be performed by TPs using NR. However, sis: WR, ER, and EC performed better. NR performed the some TPs again performed the task correctly due to their worst because it was not possible to perform this spatial familiarity with the Overijssel railway network, while oth- task without railroads. However, due to their familiarity ers used their instinct. On average, the response times of with the Overijssel railway network, two persons were still all four groups were comparable with the exception of two able to perform the task correctly using NR. The average TPs (one in ER and one in EC) who took an atypical length response times of WR, ER, and EC were comparable. How- of time to complete the task successfully. In line with the 346 Ë R. Ullah et al.

Figure 10: Correctness, satisfaction, and response time measure- Figure 11: Correctness, satisfaction, and response time measurements for the spatio-temporal task T6 for the four CTC designs – ments for the spatio-temporal task T7 for the four CTC designs – i.e. with railroads (WR), without railroads (NR), with emphasized i.e. with railroads (WR), without railroads (NR), with emphasized railroads (ER), and with emphasized time-circles (EC). Each box plot railroads (ER), and with emphasized time-circles (EC). Each box plot shows the median (line within the box), the 25th percentile (lower shows the median (line within the box), the 25th percentile (lower end of the box), the 75th percentile (upper end of the box), and 1st end of the box), the 75th percentile (upper end of the box), and 1st to 99th percentile (solid lines). The dots show the response time of to 99th percentile (solid lines). The dots show the response time of each participant. each participant. correctness results, the majority of TPs ranked ER and EC identify all possible routes between Enschede and Zwolle either ‘good’ or ‘very good’. and thus answered the task only partially. They also took The results for spatio-temporal task T7 are given in Fig- a marginally longer than the other cartograms because of ure 11. In this task the TPs were asked to identify a station the time spent in locating the routes. However, the TPs (or stations) which was evenly close time-wise to both En- seem to have found this visualization satisfying, which is schede and Zwolle. This was considered the most difficult not really strange because of a problem with the wording of task because it required two different cartograms (from En- this task. We did not state explicitly (but obviously should schede and Zwolle) to complete the task successfully. To have done) that TPs had to list all possible stations which perform this task, TPs required both railroads and time- are evenly close time-wise to both Enschede and Zwolle. circles: the railroads to locate the paths between Enschede This may have explained why many of the TPs in NR an- and Zwolle and the time-circles to find out the stations swered the task only partially correctly but ranked the de- which are in the middle time-wise. Four stations – Wier- sign the highest. den, Rijssen, Holten, and Daarlerveen – are evenly close In the spatio-temporal task T8, the TPs in each group time-wise to both Enschede and Zwolle. For this task, WR, were provided with all four alternative designs of the CTC. ER, and EC were expected to perform better than NR and They were asked to perform the given task using all four this hypothesis was confirmed by the testing. A majority designs and choose the one that best satisfied their re- of the TPs who worked with these cartograms performed quirements. The task required TPs to list all intermedi- the task correctly by listing all four stations. Here, the rail- ate stations while travelling (via the shortest path in time) roads helped to quickly identify the routes between En- from Raalte to Holten. To avoid potential bias, the presen- schede and Zwolle and the time-circles assisted to find the tation order of the cartograms was different in each group required stations. This is also the reason that those TPs (as illustrated in Figures 12(a)–12(d)). Note that here, the took less time to perform the task. On the other hand, NR heat maps show eye fixation in each group during the task performed reasonably well but while most TPs performed execution and not the preference for a particular design . the task correctly they but did not list all four correct sta- At the end of the task, TPs were asked to indicate their CTC tions. Due to the missing railroads, they were not able to preference (Figure 12(e)). The majority of TPs (27 out of 40) Usability evaluation of centered time cartograms Ë 347

Figure 12: Results for the spatio-temporal task T8. (a) – (d) illustrate the presentation order of the four designs and the densities of gazes in each group. (e) shows the number of participants selecting different cartograms – i.e. with railroads (WR), without railroads (NR), with emphasized railroads (ER), and with emphasized time-circles (EC). opted for ER, followed by EC (7 out of 40) and WR (6 out of more on the negative side for the three usability aspects, 40). Not a single TP selected NR. particularly ‘easy to understand’ and ‘use without confu- At the end of test, the TPs in each group were provided sion’. with a questionnaire where they rated the design accord- Overall, the results of the laboratory test suggested ing to three different categories: (a) pleasant to see, (b) that among various designs, ER is the most preferred de- easy to understand, and (c) use without confusion. The re- sign and NR is the least preferred. In general however, sults are summarized in Figure 13. Overall, the satisfaction the effectiveness, efficiency, and satisfaction of the CTCs of WR, ER, and EC was more on the positive side for the significantly depend on the nature and complexity ofthe three usability aspects. In the case of NR, satisfaction was task. Prior knowledge about CTCs also plays a role in this 348 Ë R. Ullah et al.

Figure 13: Satisfaction measurements based on the questionnaire for the four CTC designs – i.e. with railroads (WR), without railroads (NR), with emphasized railroads (ER), and with emphasized time-circles (EC). regard. Overall, no design was found to be equally suited always the same as what they actually do. We used video- to all tasks and there are apparent differences in the suit- recordings and thinking aloud data from the test to ana- ability of certain designs for answering specific tasks. We lyze the TPs’ behavior while performing the tasks. While found that while WR, ER, and EC are better suited for thinking aloud can help assess TPs’ responses and feelings spatial and spatio-temporal tasks, NR gives better perfor- when they execute tasks, eye-tracking also provides ad- mance for temporal tasks. We propose that optimal results ditional unobtrusive evidence of TPs’ behaviour. We con- could be achieved by presenting these cartograms in an in- sidered the TPs’ visual behavior as typical if the fixation teractive visualization environment where users can select points were placed on the area of interest and considered a particular design and turn different layers on/off depend- as atypical if otherwise. We defined the area of interest as ing upon the nature and complexity of the task. the corridor where we anticipated the TPs to look at while working with the CTC during task execution. Group 1 worked with CTCs with railroads (WR). TP7 2.3 Further analysis based on thinking and TP10 required more time for all tasks, particularly for aloud, eye-tracking, and profile data T1 (350 seconds and 211 seconds respectively versus the average time of 121 seconds). TP7 was familiar with car- In this section, thinking aloud and eye-tracking results are tograms and the geography of the Netherlands. In addi- discussed along with the responses data to identify the tion, she had map-making and map-use experience. She problem areas in the four alternative CTC designs. During completed T1 successfully but took very long. The gaze plot the test execution, what TPs say or believe they do, is not Usability evaluation of centered time cartograms Ë 349

Figure 14: Selected gaze plots of test participants for the mentioned tasks in group 1. in Figure 14(a) compares the visual behaviors of TP7 versus area of interest whereas those of TP5 are scattered. This TP3 (who was also familiar with cartograms and took only participant was seemingly confused while performing the 68 seconds to perform the task correctly) for T1. Unlike TP7, task. the fixation points of TP3 are only placed on the area of Group 2 worked with CTCs without railroads (NR). The interest. TP10 was not knowledgeable about cartograms, participants in this group consumed more time overall for did not have any map-making experience and had very all tasks and particularly for spatial and spatio-temporal little map-use experience. She performed T1 successfully tasks. TP7 was a frequent traveler of the Dutch trains but but consumed more time to understand the task and fa- did not know about cartograms. He completed T1 success- miliarize herself with the representation. This is evident fully but took very long (337 seconds versus the average from the gaze plot (see Figure 14(b)) and also supported by time of 185 seconds) to understand the task and familiar- the thinking aloud data. Figure 14(c) compares the visual ize himself with the CTC as depicted in Figure 15(a)). The behaviors of TP1 and TP5 for T2. Both participants were gaze plots in Figure 15(b) compare the visual behaviors for knowledgeable about cartograms and showed similar vi- T2 of TP3 (who answered the task incorrectly and took 87 sual behaviors to complete the task successfully in compa- seconds versus the average time of 109 seconds) and TP5 rable times (81 and 121 seconds respectively). TP8, being a (who completed the task successfully but required 208 sec- frequent user of the Dutch trains, performed T3 success- onds). TP3 was a frequent user of maps but was not fa- fully in only 43 seconds. In that instance, the gaze fixation miliar with cartograms whilst, TP5 was an occasional user points were placed on the desired stations (as shown in of maps but was knowledgeable about cartograms. Both Figure 14(d)). TP10 stated that T4 cannot be answered with participants showed different visual behavior. In the case the provided cartogram, which was incorrect. The scat- of TP3, the fixation points are not targeted on the area of tered fixation points of TP10 show that the participant did interest. This suggests that the participant was confused not understand the task well and faced difficulty in read- and faced difficulty in interpreting the cartogram. The fix- ing the cartogram (see Figure 14(e)). A comparison of the ation points of TP5 are also scattered, meaning that the gaze plots of TP1 and TP5 for T6 is presented in Figure 14(f). participant likely did not understand the task well and TP1 completed the task correctly, while TP5 answered the was confused. This is likely the reason why TP5 consumed task wrongly. The fixation points of TP1 are placed on the more time to complete the task. TP6 and TP9 (both famil- 350 Ë R. Ullah et al.

Figure 15: Selected gaze plots of test participants for the mentioned tasks in group 2. iar with cartograms) showed similar visual behaviors to T3 and completed the task successfully but consumed more and completed the task successfully (Figure 15(c)). Both time (115 seconds versus the average time of 84 seconds). participants took 52 seconds for the task. TP7 incorrectly The gaze plot indicates that this participant took a sub- stated that T4 cannot be answered with the provided car- stantial length of time to understand the cartogram. Fig- togram and took 308 seconds to decide. The gaze plot in ure 16(c) illustrates the gaze plots of TP3 and TP6 for T3. Figure 15(d) agrees with these results, showing that the Both participants had different profiles and showed differ- participant was not able to identify the different paths. Fig- ent visual behaviors to perform the task. TP3 answered the ure 15(e) illustrates the gaze plot of TP7 for T5. The par- task wrongly, requiring 41 seconds, whilst TP6 performed ticipant answered the task partially because he failed to the task successfully in 25 seconds. Unlike TP3, TP6’s fixa- define all possible paths. Additionally, he did not under- tion points are placed on the area of interest. A compar- stand the task well and also faced difficulty in reading ison of the gaze plots of TP1 and TP4 for T4 is given in the cartogram. The gaze plots of TP6 and TP9 for T6 are Figure 16(d). TP1 completed the task successfully and had shown in Figure 15(f). TP6 answered the task incorrectly gaze points concentrated on the area of interest. TP4 per- and TP9 completed the task successfully. The gaze plot of formed the task incorrectly and illustrated scattered gaze TP6 shows that the participant did not understand the task points, showing that the participant likely did not under- well. stand the task. Figure 16(e) and Figure 16(f) depict the gaze Group 3 worked with CTCs with emphasized railroads points of TP5 for T5 and T6 respectively and in both cases (ER). The gaze plot in Figure 16(a) shows the visual be- the fixation points were targeted on the area of interest and haviors of TP2 and TP10 (both familiar with cartograms the tasks were completed correctly. and with equal map-use experience) for T1. TP2 answered Group 4 worked with CTCs with emphasized time- the task wrongly and took 104 seconds while TP10 per- circles (EC). The gaze plots of TP1 for T1 and T2 are shown formed the task correctly and required 79 seconds. The fix- in Figure 17(a) and Figure 17(b) respectively. The partici- ation points of TP2 are not placed on the area of interest pant, being knowledgeable about cartograms, performed whereas the fixation points of TP10 are perfectly placed the tasks correctly and the gaze points are perfectly placed on the area of interest. Figure 16(b) shows the gaze plot of on the area of interest in both cases. Figure 17(c) depicts TP5 for T2. The participant did not know about cartograms the gaze plot of TP9 for T3. Here the participant answered Usability evaluation of centered time cartograms Ë 351

Figure 16: Selected gaze plots of test participants for the mentioned tasks in group 3.

Figure 17: Selected gaze plots of test participants for the mentioned tasks in group 4. the task wrongly and the scattered gaze plot verifies that T4 is provided in Figure 17(d). TP4, having no knowledge result. A comparison of the gaze plots of TP4 and TP7 for of cartograms, incorrectly stated that the task could not be 352 Ë R. Ullah et al. answered with the provided cartogram. Accordingly, the • The thinking aloud data illustrated a number of de- fixation points are not targeted on the area of interest and sign issues. Some participants stated that the leg- this illustrates the participant likely did not understand end was not clear enough and suggested to provide a the task well. In contrast, TP7,who also had no knowledge- clearer legend. Most participants commented on the able of cartograms, completed the task successfully and labeling, color, and point size of the stations, and illustrated gaze points which were placed on the area of color and line width of the railroads. All these prob- interest. TP8 answered T5 partially as he did not define all lems were corrected for the online survey based us- possible correct paths (as can be seen in Figure 17(e)). The ability evaluation. gaze plots in Figure 17(f) compare the visual behaviors of • Overall the results (particularly from task T3) indi- TP2 and TP10 for T6. Both participants performed the task cated that the inclusion of railroads adds to the read- correctly in 596 seconds and 63 seconds respectively. TP2 ability of the CTCs and does not provide a false im- seemingly took long to understand the task and familiarize pression of time. However, in some cases and par- herself with the cartogram whereas with TP10, the fixation ticularly for non-experienced users, the railroads points were perfectly placed on the area of interest. This is could be misleading. Users therefore need to be ed- the likely reason TP10 consumed less time. ucated about the role of railroads in CTCs. Various TPs worked with the CTCs from Enschede, • Overall, the responses, thinking aloud and eye- Zwolle, Deventer, and Steenwijk (less distorted) to per- tracking data inferred that the distortions of CTCs do form different tasks. Several other TPs performed similar not greatly affect the readability. tasks using the CTCs from Raalte and Gramsbergen (both highly distorted and far from reality). Because of the se- vere distortion in the geography, we expected poor perfor- 3 The online survey mance in the latter case. However, the responses, thinking aloud and eye-tracking data showed that the distortions of CTCs did not greatly affect the readability. For instance, 3.1 Objective, design, and implementation TP3 in group 1 performed T1 using the CTC from Enschede. TP10 in group 3 worked with the CTC from Raalte to per- This experiment was aimed to explore the usability of a form T1. Both participants were familiar with cartograms CTC in comparison with two other kinds of equivalent rep- and showed similar visual behaviors to complete the task resentations (i.e. a geographic and schematic map).The successfully in comparable times (68 seconds and 79 sec- laboratory test showed that the CTC with emphasized rail- onds respectively). See Figure 14(a) and Figure 16(a). TP1 roads (ER) was the most preferred design. However, the in group 4 performed T1 and T2 correctly using the CTCs participants suggested a number of improvements to the from Gramsbergen and Raalte respectively. The distorted design particularly relating to the legend which was not nature of the cartograms did not influence the visual be- clear enough. A modified legend was then provided toas- havior of TP1 as the gaze points were perfectly placed on sist participants in knowing what each of the elements the area interest in both cases (compare Figure 17(a) versus used in the cartograms represent. Many participants also Figure 17(b)). commented on the labeling of the stations. For better visibility and to allow a stronger reflections and evaluations on the actual tasks, and unlike in the laboratory eval- 2.4 Conclusions from the laboratory test uation, we therefore only labeled the stations of interest. Color and point size of the stations and color and The following can be concluded from the laboratory based line width of the railroads were also adjusted as per par- usability evaluation: ticipants’ suggestions. Compare Figure 1(c) versus Fig- ure 18(a). We then investigated whether the improved CTC • Overall, the results suggested that ER is the most was preferable over the other two map types in performing preferred design and NR the least preferred. How- spatio-temporal tasks. ever, the effectiveness, efficiency, and satisfaction of In this part, CTCs were constructed for the whole the CTCs were highly dependent on the nature and network of the Dutch railway. Seventy-three train sta- complexity of the task: WR, ER, and EC being better tions, well-spread over the country, were used to construct suited to spatial and spatio-temporal tasks and NR the cartograms. Four stations (i.e. Eindhoven, Maastricht, better suited to temporal tasks. Utrecht, and Schiphol) were selected as starting stations (see Figure 18). The alternative geographic and schematic Usability evaluation of centered time cartograms Ë 353

Figure 18: CTCs constructed for four different starting stations of the Dutch railway network. (a) Eindhoven. (b) Maastricht. (c) Utrecht. (d) Schiphol. The concentric circles depict travelling-times in increments of 30 minutes from the starting station. maps of the Dutch railway are shown in Figure 19. To make of the test, the participants performed the same four tasks sure that the tests were fair, the maps were designed so that using all three representations and ranked each on a Lik- they contained exactly the same amount of information. ert scale from ‘very good’ to ‘very poor’ based on how well In the geographic and schematic maps, time was encoded each type assisted them in performing the tasks. The ac- using labels along the network segments. tual test was preceded by a questionnaire in which partic- The test contained four spatio-temporal tasks (listed ipants were asked to answer questions about their profile. in Table 2) and consisted of two parts. In the first part, The test was implemented through an online sur- the participants were asked to perform the four tasks us- vey (created using the Lime Survey tool – www.limesur ing one of the three representations (selected at random). vey.org). Lime Survey was chosen because it provides time To select a representation at random, three option buttons statistics for responses. The survey was distributed to dif- (placed in a row and without a label to prevent any bias) ferent networks by sending email invitations. In total, 140 were provided. The first option button selected the geo- responses were collected; 88 responses were complete, graphic map, the second button selected the CTC, and the and the other 52 were incomplete. We considered the com- third button offered the schematic map. In the second part plete responses only. The responses were different for each 354 Ë R. Ullah et al.

Table 2: List of spatio-temporal tasks for the online survey.

Task Nature Description T1 TEMPORAL Mark is a new graduate student at TU Eindhoven. This Monday he wants to go to the IND in Zwolle to collect his residence permit. The documents collection timings are from 11:30 am to 12:30 pm. At what time should he leave from Eindhoven to reach Zwolle at 11:00 am in the shortest time possible?

T2 TEMPORAL Robert is in Utrecht to attend a 3-day workshop at Utrecht University. The workshop runs from 09.00 am to 05.00 pm daily. As he has never been in the Netherlands before, on one of the workshop days he wants to visit another place that is close to Utrecht. Could you please list all cities to Robert that are reachable by train within 30 minutes from Utrecht?

T3 SPATIO- Smith needs to travel frequently between Eindhoven and Maastricht for his job assignments. TEMPORAL Where should he rent an apartment so that he is time-wise evenly close to both places?

T4 SPATIO- List the intermediate stations while travelling (via the shortest path in time) from Schiphol to TEMPORAL Enschede

Figure 19: (a) A geographic map of the Dutch railway network. (b) A schematic representation of the Dutch railway network. representation; 55 participants worked with the CTC, 12 28 Bachelor, 43 Master and 15 PhD as their highest edu- with the geographic map, and 21 participants performed cational degree. The participants had different academic the tasks using the schematic map. The difference in the backgrounds, with the majority from Geo-Information Sci- number of respondents may be due to the order of the op- ence and Earth Observation. They came from 32 different tion buttons. With the CTC button being in the middle, nationalities, with 67 currently living in the Netherlands many participants performed the tasks using the CTC. and the other 21 in the rest of the world. A majority of Among the 88 participants who completed the survey, the respondents were frequent train travelers and also fre- 1 listed intermediate vocational education, 1 high school, quent users of maps. About half of the participants knew Usability evaluation of centered time cartograms Ë 355

Figure 20: Correctness and response time measurements for Figure 21: Correctness and response time measurements for the temporal task T1 for the three representations - i.e. CTC (C), the temporal task T2 for the three representations - i.e. CTC (C), schematic map (S), and geographic map (G). Each box plot shows schematic map (S), and geographic map (G). Each box plot shows the median (line within the box), the 25th percentile (lower end of the median (line within the box), the 25th percentile (lower end of the box), the 75th percentile (upper end of the box), and 1st to 99th the box), the 75th percentile (upper end of the box), and 1st to 99th percentile (solid lines). The dots show the response time of each percentile (solid lines). The dots show the response time of each participant. participant. about cartograms in general, but had limited or no knowl- The results for temporal task T1 are displayed in Fig- edge about CTCs. ure 20. In this task, participants had to report the departure time from Eindhoven to reach Zwolle at 11:00 am in the shortest time possible. For this task, the average cor- 3.2 Results and discussion rectness of the CTC was higher than that of the schematic map but only marginally lower than that of the geographic The participants were asked to perform four tasks related map. Similarly the average response time of the CTC was to spatio-temporal aspects (see Table 2). Firstly, they were shorter than both the geographic map and the schematic requested to perform the tasks using one of the three differ- map. In the CTC, the participants spent less time perform- ent representations. We collected participants’ correctness ing the task because they seemingly quickly calculated the and response times. Secondly, they performed the same travelling-time from Eindhoven to Zwolle by counting the four tasks using all three representations, and gave a rank number of circles. Alternately, participants who worked to each representation on a Likert scale from ‘very good’ to with the geographic and schematic map took longer be- ‘very poor’ considering how effective each representation cause they had to add the travelling-times along the net- was to perform the tasks. This part was aimed to measure work segments in order to ascertain the time distance from participants’ satisfaction. Below we discuss the results in Eindhoven to Zwolle. Most participants were satisfied with detail first for each task. the CTC. They ranked the CTC as ‘good’ and ‘very good’ and For each task, we calculated correctness, response the other two maps as ‘average’ and ‘good’ (Figure 24(a)). time, and satisfaction for all three representations. We ex- Figure 21 displays the results for temporal task T2. In pected the CTC to perform better than the other two solu- this task, participants had to list all stations that were tions for task T1 and task T2, and poorly for task T3 and task reachable within 30 minutes by train from Utrecht. This T4: as T1 and T2 are temporal tasks and T3 and T4 spatio- task was easy to perform using the CTC but difficult to temporal tasks. T3 was likely to be difficult for the CTC, as perform using the two other maps. Using the CTC, all one it uses two different cartograms (i.e. one from Eindhoven needed to do was to record all of the stations located inside and the other from Maastricht) to complete the task suc- the first (i.e. 30-minute) circle. Using the other two maps cessfully. however, participants had to add travelling-times along 356 Ë R. Ullah et al.

Figure 22: Correctness and response time measurements for the Figure 23: Correctness and response time measurements for the spatio-temporal task T3 for the three representations - i.e. CTC (C), spatio-temporal task T4 for the three representations – i.e. CTC (C), schematic map (S), and geographic map (G). Each box plot shows schematic map (S), and geographic map (G). Each box plot shows the median (line within the box), the 25th percentile (lower end of the median (line within the box), the 25th percentile (lower end of the box), the 75th percentile (upper end of the box), and 1st to 99th the box), the 75th percentile (upper end of the box), and 1st to 99th percentile (solid lines). The dots show the response time of each percentile (solid lines). The dots show the response time of each participant. participant. the network segments to ensure they were less than 30 The results for spatio-temporal task T4 are illustrated minutes. That extra arithmetic was obviously more time in Figure 23. In this task participants had to list the in- consuming. The results of T2 were in line with our hy- termediate stations while travelling from Schiphol to En- pothesis. The CTC was both effective and efficient for this schede via the shortest path in time. This task required task. Test participants also found the CTC more visually in- participants to first find the shortest path from Schiphol teresting and satisfying compared to the two other maps. to Enschede and then list all stations along it. For this The satisfaction score of the CTC was more on the positive task, the correctness of the CTC was higher than the ge- side compared to the geographic and schematic map type ographic and schematic map. The average response time where scores were more on the negative side (Figure 24(b)). of the CTC was also shorter than the other two maps. Us- The results for spatio-temporal task T3 are given in Fig- ing the geographic and schematic map, the participants ure 22. In this task participants had to identify a station (or had to add the travelling-times along all possible paths stations) which was time-wise evenly close to both Eind- from the source to destination to know the shortest path. hoven and Maastricht. Here, the participants needed two That was obviously time consuming and is evident from different cartograms to perform the task successfully; from the results. For this task, the participants ranked all three Eindhoven and from Maastricht. Both cartograms were maps between ‘average’ and ‘good’, with the ranking of provided to the participants. This task was considered to CTC slightly more on the positive side (Figure 24(d)). be difficult for the participants who worked with the CTC. Moreover, we calculated two statistical measures – We also expected the participants to require more time us- mean and mode – to rank the three representations. In ing the CTC in comparison with the other two maps. How- order to compute the mean ranking, numbers from 1 to 5 ever the CTC performed equally well versus the other map were used to indicate the range from ‘very poor’ to ‘very types. The correctness of the CTC was higher and the aver- good’. The mean ranks are listed in Table 3. The ranking age response time comparable compared to the two other order based on the mean rank value is: CTC (3.89); ge- map types. For this task, a majority of the participants ographic map (3.46); schematic map (3.37). Mode is the ranked the CTC between ‘good’ and ‘very good’ and the value that occurs most often in a set of data. We counted other two representations between ‘average’ and ‘good’ the first and second frequently occurring rank forthe (Figure 24(c)). three maps. The results are shown in Table 4. The rank- Usability evaluation of centered time cartograms Ë 357

Table 3: Mean scores for the three representations - i.e. CTC (C), schematic map (S), and geographic map (G).

Map C G S Average score 3.89 3.46 3.37

Table 4: First and second mode scores for the three representations - i.e. CTC (C), schematic map (S), and geographic map (G).

very poor poor average good very good First frequent score SSGCC Second frequent score G G S G, S G

Figure 24: Satisfaction measurements for all tasks for the three representations – i.e. CTC (C), schematic map (S), and geographic map (G). ing order based on the mode scores is CTC>geographic tasks, particularly when the task had a dominant map>schematic map, which is in agreement with the rank- time-related component. ing order based on the mean rank value. • CTCs are appealing, as most participants ranked them as ‘very good’ and ‘good’.

3.3 Conclusions from the online survey 4 Final conclusions and future work The online survey based usability evaluation illustrated the following: This study suggested centered time cartograms (CTCs) as • Overall, CTCs performed better than the schematic an intriguing, but as yet untested, possibility for represent- and geographic map in answering spatio-temporal ing space-time phenomena and presented a detailed and 358 Ë R. Ullah et al. systematic user evaluation of these cartograms. Two user be embedded in an interactive visualization envi- tests were conducted: a laboratory test and an online sur- ronment where users can select a particular repre- vey. In the laboratory test, we used eye-tracking, thinking sentation and also switch on/off different layers de- aloud, and video-recording to compare four different de- pending upon the nature and complexity of the task. signs of centered time cartograms to find out which de- Additionally, for transport networks in which seg- sign (or combination) of these performs better in answer- ments and interconnections are dense and numer- ing spatio-temporal questions. In the online survey, we ous, there may be readability issues. To address this evaluated centered time cartograms against a geographic problem, the interactive environment should have and schematic map to determine whether they better fa- zoom and pan functionalities to allow visualization cilitated the answering of certain questions compared to of the network at different resolutions. Moreover, geographic and schematic maps. Based on our results, the the environment could permit users to choose line following conclusions can be drawn. width, line color, point size, and point color for in- creased visibility. • Among the four alternative designs, the CTC with emphasized railroads was the most preferred design and the CTC without railroads the least preferred. However, the effectiveness, efficiency, and satisfac- References tion of the CTCs were greatly dependent on the nature and complexity of the task. The CTCs with rail- [1] Bies S, van Kreveld M. Time-space maps from triangulations. In: roads, with emphasized railroads, and with empha- Didimo W, Patrignani M, editors. Graph Drawing. Volume 7704, sized time-circles were better suited for spatial and Lecture Notes in Computer Science: Springer Berlin Heidelberg; 2013. p 511–516. spatio-temporal tasks while the CTC without rail- [2] Ullah R, Kraak M-J. An alternative method to constructing time roads was better suited for temporal tasks. cartograms for the visual representation of scheduled move- • The inclusion of railroads to these CTCs generally ment data. Journal of Maps 2014; 11(4):674–687. increases their readability and does not provide a [3] van Campenhout M. Travel time maps [MSc thesis]: Technical false impression of time. However in some cases, University Eindhoven; 2010. [4] Angel S, Hyman GM. Transformations and Geographic Theory. and particularly for non-experienced users, the rail- Geographical Analysis 1972; 4(4):350–367. roads were misleading. Users therefore need to be [5] Buchin K, van Goethem A, Hoffmann M, van Kreveld M, Speck- educated about the role of railroads in CTCs. mann B. Travel-time maps: linear cartograms with fixed vertex • Distortions in CTCs do not greatly affect their read- locations. In: Duckham M, Pebesma E, Stewart K, Frank A, ed- ability and visual behavior. itors. Geographic Information Science. Volume 8728, Lecture • CTCs were effective and also more efficient than ge- Notes in Computer Science: Springer International Publishing; 2014. p 18–33. ographic and schematic maps in answering spatio- [6] Dorling D. Cartogram. In: Monmonier M, Collier P,Cook K, Kimer- temporal tasks, particularly when the task had a ling J, Morrison J, editors. History of Cartography: Cartography dominant time-related component. in the Twentieth Century. Volume 6. Chicago: Chicago University • CTCs provide an alternative method to visualize Press; 2015. p 1960. travelling-times and are appealing due to their capti- [7] Inoue R. A new construction method for circle cartograms. Car- tography and Geographic Information Science 2011; 38(2):146– vating design. We suggest this method as an intrigu- 152. ing possibility for representing space-time phenom- [8] Bhatt LM. Investigating the appropriateness of Gastner- ena and believe that this visualization method de- Newman’s cartogram versus conventional maps in visual serves greater attention. We hope that our work will representation and modeling of health data [MSc thesis]: encourage the design and use of CTCs by cartogra- Southern Illinois University at Carbondale; 2006. phers and information visualization researchers. [9] Gastner MT, Newman MEJ. Diffusion-based method for pro- ducing density-equalizing maps. Proceedings of the National • In general, the laboratory test and the online sur- Academy of Sciences of the United States of America 2004; vey illustrated that no single visualization was suit- 101(20):7499–7504. able for all tasks and that certain representations [10] Keim DA, North SC, Panse C. CartoDraw: a fast algorithm for are better suited to answering specific tasks: one generating contiguous cartograms. Visualization and Computer representation may be better to answer one task Graphics, IEEE Transactions on 2004; 10(1):95–110. [11] Sagar BSD. Cartograms via mathematical morphology. Informa- and another representation may be better to an- tion Visualization 2014; 13(1):42–58. swer another task. We suggest that optimal results [12] van Kreveld M, Speckmann B. On rectangular cartograms. Com- would be obtained if these representations could putational Geometry 2007; 37(3):175–187. Usability evaluation of centered time cartograms Ë 359

[13] Davies S. Annual Report of the Chief Medical Oflcer, Volume [27] Ullah R, Kraak M-J, van Elzakker C. Using cartograms to explore One, 2011, On the State of the Public’s Health. London: Depart- temporal data: do they work? GeoViz 2013. Hamburg2013. ment of Health 2012:11. [28] Goedvolk A. De nieuwe relatieve afstand voor het openbaar ver- [14] Dent BD. Communication aspects of value-by-area cartograms. voer. Nieuwe Geografenkrant 1988; 10:6–7. The American Cartographer 1975; 2(2):154–168. [29] Spiekermann K, Wegener M. The shrinking continent: new time- [15] Dorling D. Area cartograms: their use and creation (concepts space maps of Europe. Environment and Planning B: Planning and techniques in modern geography). Norfolk, UK: Environ- and Design 1994; 21(6):653–673. mental Publications 1996. [30] Cauvin C. A systemic approach to transport accessibility. A [16] Kocmoud C, House D. A constraint-based approach to construct- methodology developed in Strasbourg: 1982–2002. Cybergeo: ing continuous cartograms. In: Poiker T, Chrisman N, editors. European Journal of Geography 2005. Proceedings of the 8th International Symposium on Spatial Data [31] Ahmed N, Miller HJ. Time-space transformations of geographic Handling. Vancouver, BC: Taylor and Francis; 1998. p 236–246. space for exploring, analyzing and visualizing transportation [17] Shimizu E, Inoue R. A new algorithm for distance cartogram con- systems. Journal of Transport Geography 2007; 15(1):2–17. struction. International Journal of Geographical Information Sci- [32] Axhausen KW, Dolci C, Fröhlich P, Scherer M, Carosio A. Con- ence 2009; 23(11):1453–1470. structing time-scaled maps: Switzerland from 1950 to 2000. [18] Wu Y-H, Hung M-C. Non-connective linear cartograms for map- Transport Reviews 2008; 28(3):391–413. ping traflc conditions. Cartographic Perspectives 2012(65):33– [33] Kaiser C, Walsh F, Farmer CQ, Pozdnoukhov A. User-centric time- 50. distance representation of road networks. In: Fabrikant S, Re- [19] Fotheringham AS, Brunsdon C, Charlton M. Quantitative geogra- ichenbacher T, Kreveld M, Schlieder C, editors. Geographic In- phy: perspectives on spatial data analysis. SAGE Publications; formation Science. Volume 6292, Lecture Notes in Computer 2000. Science: Springer Berlin Heidelberg; 2010. p 85–99. [20] Tobler W. Thirty five years of computer cartograms. Annals of [34] Chen X. Seeing differently: cartography for subjective maps the Association of American Geographers 2004; 94(1):58–73. based on dynamic urban data [MSc thesis]: Massachusetts In- [21] Yau N. Alternative to cartograms using transparency. Flowing- stitute of Technology; 2011. Data2008. [35] Ramaer L. De vervaardiging van temporele kartogrammen. 100 [22] Roth RE, Woodruff AW, Johnson ZF. Value-by-alpha maps: an al- Jaar veranderingen in de reistijd per trein in beeld. Geo-Info ternative technique to the cartogram. The Cartographic Journal 2011;2011-10/11:11-13. 2010; 47(2):130. [36] Denain J-C, Langlois P. Cartographie en anamorphose. Mappe- [23] Sun H, Li Z. Effectiveness of cartogram for the representation of monde 1998; 49(1):16–19. spatial data. The Cartographic Journal 2010; 47(1):12–21. [37] Nielsen J. The use and misuse of focus groups. Software, IEEE [24] Kaspar S, Fabrikant S, Freckmann P. Empirical study of car- 1997; 14(1):94–95. tograms. 2011; Paris, France. [38] Nielsen J. Estimating the number of subjects needed for a think- [25] Reyes Nuñez JJ, Juhász B. Hungarian survey on the use of car- ing aloud test. International Journal of Human-Computer Stud- tograms in school cartography. International Journal of Cartog- ies 1994; 41(3):385–397. raphy 2015; 1(1):5–17. [39] Bertin J. La graphique. Communications 1970; 15:169–185. [26] Hong S, Kim Y-S, Yoon J-C, Aragon CR. Traflgram: distortion for clarification via isochronal cartography. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems. Toronto, Ontario, Canada: ACM; 2014. p 907–916.