|
Table 1. Results from the field study |
|
Performance of passengers (n = 105) searching for departure times en the trams indicator of Amsterdam Central Station |
|
Performance | Proportion of the sample |
| Correct even when taking another train and changing trains would decrease travel time | 61% |
| | |
| Sub optimally corrects correct, but there was a train that left before and arrived before
| 18% |
| Incorrect: mentioned a train that did not go to the named destination, even with a change of trains | 5% |
| | |
| Do not know | 15% |
| | |
| The mean delay for the total sample resulting from those passengers who mentioned a sub optimally correct train, computed over all passengers however | 6 min |
| | |
| Mean search time | 20 s |
|
3. The cause of the problem |
3.1 The method
We presented an Amsterdam trains indicator to passengers who had just left Amsterdam by train (see Figure 2 at the right). We asked them to search for the departure of five trains. After that we presented the same indicator but without the information concerning the structure, in other words, we blacked out the information concerning the structure, in other words, we blacked out the destinations, the header tend destination, and the header “via“ (see Figure 3, immediately below, at the right). The next step was to ask questions about the structure of the indicator. |  |
Figure 2 Experimental trains indicator
Trains as the basis for travel information arrangement.
Time is the basis of this structure, Passengers using this system generally will not know exactly which train to take nor the departure time. They just start searching somewhere and hope that the information they are looking for can be found at the beginning of their route. When they find an appropriate train it is advisable to continue searching. Often, somewhere else, there is a train that will arrive sooner at their destination. We used this trains indicator in the comparative research.<
|
| |
There are some differences between passengers searching for their own train in a station hall and passengers in this study searching for trains mentioned by the interviewer. In reality, passengers will only be looking for one train whereas during our investigation we asked the passengers to search for the departure times of five trains. We asked about every station an equal number of times; in reality main line junction stations cause a lot more problems than line stations and the number of passengers to main line junction stations is much greater. Another difference was that the only trains and destinations considered were: |
that were inland only;
that were presented on the indicator;
that would leave within one off-peak hour;
that would depart according to schedule.
These differences between the experiment and real situations will cause an underestimate of problems; in reality there will be more problems than we report. |
Figure 3. Trains by times indicator without information on the structure. The columns 'end-destination' and 'via' are empty. Do you know what information was in this column?' (pointing to the empty end-destination column and the empty via column.)
|
|
3.2 Results
We hypothesized that the passengers did not know or use the structure displayed on the trains indicator. The results confirm our hypothesis (see Table 2); passengers correctly answered only 15-47% of the questions on structure. Only 6% gave a correct answer to every question on structure.
These results suggest that the passengers did not know or were not conscious of the way the information was arranged. Nevertheless, we cannot conclude that the passengers could not deal with the structure chosen by the designers of the indicator. |
When interpreting these results, one should bear in mind that we obtained them in a memory testy and when a passenger could not remember the structure this does not imply that they did not use the structure. The figures are possibly an overestimate.
However, there is another indication. Earlier research, (Verhoef, 1984) showed that 46% of passengers did know their own destination but did not know the end destination of the train. Included in this are 9% who thought they knew the end destination of the train but mentioned the wrong destination.
It can be concluded that a structure based on the end destination of the transport vehicle is not a good structure for these kinds of systems.
|
| |
|
Table 2 Knowledge of the structure of the trains indicator (n = 353)
|
Proportion | |
of the | |
sample | What is remembered |
| |
15% | Difference between left and right panels |
11% | Contents of the column end destination |
47% | Contents of the column via |
|
4 A solution to the problem | The indicators we investigated did have conceptual structures. However, there were too many structures and the structures used were inadequate.
|
|
4.1 Too many structures | It was found in the initial fieldwork that four different types of train indicator existed, each with it’s information structured in a different way. Indicators were found that had: |
a left and right panel, with the contents of the left panel summarized in a header with nine destinations, and on the right panel the summary consisting of 10 destinations;
two columns, one with “end“ destinations and one with “via“ destinations;
each line on the panel referring to one train;
lines which were ordered chronologically.
Using four different strictures will dilute the ability of users to obtain the information they need quickly and accurately.
|
4.2 The wrong structure |
A conceptual structure should, of course, be logically but several logics are possible. For instance, take the way in which departure times of trains are arranged. On British railway stations this information is arranged by destination. In Germany the same information is presented in chronological order. Netherlands Railways follow neither of their neighbours, they put together all information concerning trains that go in the same direction. In sum, there are many structures and many logics. Which ones are the best for travel information? What information does the available literature offers?
General cognitive psychology (e.g. Best, 1989) does not yet offer applicable knowledge. Neisser (1976) cannot help us more than stating that, what is perceived should fit the schemata of the user. More specific psychological work and ergonomics do not inform us either: for instance, Brenner (1982), Easterly and Zwaga (1984), Hancock (1987) Harrigan (1987). Janssen (1986) stated that there is little literature on the problem of structuring information concerning destinations.
It is the opinion of Adams et al (1984) that destination information should be arranged by directions. For instance, a two-floor building might have a sign posting system such as: |
Second floor administration
Selling department
Storehouse
Toilets
Workshop
First floor
First aid
Management
Reception |
There is a compatibility between the location of the destination on such displays and the direction in which to go. That is, left-side directions go on the left side of the display, upper directions on the top of the display (as in this example) and so on. Adams et al (1984) argue that this leads to a distinct layout, but unfortunately they do not define distinctness.
|
|
However, users scanning a signpost do not know the direction in which they have to go: so they do not know in which group to search for their destination on the signpost (on the left, on the right, upper, or lower groups?). Like users cannot use the direction when searching the visual field because they know their direction only after finding their destination in the field. It is our opinion that the proposal of Adams et al (1984) is wrong and we prefer that of Wright (1988). She stated that signposts at elevators should not arrange information by floor but by destination. For instance, in the example above, a better structure would be: |
Administration, 2nd floor
First aid, 1st floor
Management, 1st floor
Reception, 1st floor
Selling department, 2nd floor
Storehouse, 2nd floor
Toilets, 2nd floor
Workshop, 2nd floor |
Research carried out by Spijkers et al (1985) and by Corlett et al (1972) supports this theory. Spijkers et al investigated how to structure information on signposts. Their subjects searched in a list of 18 destinations for 0.8 to 2 s. Users performed better when the information was arranged according to information already available to them, e.g. known categories or the alphabet (see Table 3).
|
|
As a final argument we can make an analogy.
Suppose a telephone book is arranged in terms of the information for which you are searching, i.e. the telephone number. The book would not start with the subscribers whose name starts with the letter A, but with the subscriber whose number is the lowest. That, of course, does not work. |
| Correct | Basis for structure |
| response | |
| | Known |
| 91% | Alphabetical |
| 87% | Categorical and alphabetical |
| | Known and unknown |
| 76% | Categorical, directions alphabet |
| 79% | Categorical, random |
| 62% | Direction and alphabet |
| | Unknown |
| 49% | Random |
| Source: Spijkers et al (1985) |
|
Table 3 Percentage correct responses for different kinds of structure of information (n = 24)
|
4.3 Destination as a basis for a structure | The previous section provides a clear conclusion that the basis for the arrangement of information should be knowledge that is available to the user.
Returning to dynamic train indicators, let us examine the consequences. For train and flight indicators the basis for the structure should be information already available. It may be presumed that passengers know where they are going and thus the structure for information aimed at passengers should be their destination. Of course one could argue that passengers often have other travel information, such as a train or flight number. However, if they have this information it is because they adapted themselves to the system by learning. In this paper it is proposed that the system should be adapted to the user.
As there are many destinations, they have to be arranged somehow. We apply the same theory: the basis should be the information available. Not available are: gate number, airline company, platform number, train type, departure time, or flight number. The alphabet is available and known to the passenger. So an ergonomic structure of travel information on dynamic timetables is through destinations arranged alphabetically (see the Figure at the right). |
 |
Figure 4 Experimental destinations indicator
The basis of this structure is destinations that are arranged alphabetically. Passengers using this system know their destinations and they can also see that the destinations are structured alphabetically so they can start searching near their destination. We used this destinations indicator in the comparative research. |
5. Evaluation of the solution | It seems obvious to conclude that the structure of information should be based on available knowledge. A visit to the nearest airport, rail or underground station will teach that most authorities do not agree. Therefore we carried out comparative research.
|
|
5.1 Method | Of course, it was not possible to build and install a new indicator for research purposes only. Fortunately, for this kind of psychological research there is little difference between information presented by a dynamic panel and information presented on a sheet of paper, so we made a comparison with two A4 sized paper indicators.
These were a trains indicator in which trains were arranged chronologically and a destinations indicator with destinations arranged alphabetically (see Figures 2 and 4 above). We presented both indicators to 304 passengers who had just left Amsterdam by train. We interviewed only those passengers who were traveling alone, who were Dutch-speaking, and who were not reading or working. Half the passenger started with the trains indicator: for the other half we reversed the sequence. The procedure was as follows. |
We asked the passengers their travel frequency, destinations the way they learned of their platform number, and whether they had seen and used the trains indicator in the station hall.
We presented the passengers with the train times or alphabetical destinations indicator and asked them to search for departure times of five trains. We scored the answers as correct when the train mentioned arrived at the destination requested, even when less travel time was possible by a change to another train. We regarded an answer as sub optimally correct when the train did arrive at the destination named but when another train would arrive sooner because it departed sooner, stopped at fewer stations or had a shorter route. For the sub optimally correct answers we computed the delays for those passengers if they had actually taken that train. These delays were totaled and divided by all passengers. This figure was defined as the mean delay across the sample, We scored an answer as “incorrect“ when the train did not lead to the destination even by changing trains somewhere. Not only the answers were scored but the reaction times as well.
After this we asked questions on the structure of the indicator they used.
We repeated this procedure with the other indicator.
Finally, we asked the passengers to give both indicators a rating. |
5.2 Results | The results indicate that there is a clear difference between a trains times indicator and an alphabetical destinations indicator, in terms of search time, delays, and the appreciation of the passengers (Tables 4, 5 and 6 below). |
When interpreting the results it should be noted that, as far as passenger experience is concerned, the chronological indicator was in fact at an advantage because this was the system that had been in use for several months, and 43% of the passengers that participated in this investigation had been looking at this indicator some minutes previously. An alphabetical indicator had probably never been seen before, but nonetheless proved the better on all measures.
|
| |
Table 4. Performance of passengers on two indicators, structured by train times or alphabetically by destination (sample size = 353)
| | | Destinations |
| | Trains | ordered by |
| Performance |
by times | alphabetically |
| Correct, even when taking | 81% | 94% |
another train and changing trains would decrease travel time.
| | |
Sub optimally correct
| 14% | 3% |
correct, but there was a train that left before and arrived before.
| | |
Incorrect mentioned a train that did not go to the named destination, even with a change of trains.
| 5% | 3% |
The mean delay caused by passengers who mentioned a sub optimally correct train (computed over all passengers)
| 3.6 min. | 0.9 min |
| | | |
| Mean search time is | 7s | 4s |
|
Table 5. Performance for junctions
Performance of passengers on the junctions Rotterdam and Utrecht (chronological indicator, n = 47, alphabetical indicator n = 29)
| | | Destinations |
| | Trains | ordered |
| Performance | by times | alphabetically |
| | | |
| Correct, even when taking another train and changing trains would decrease travel time | 36% | 86% | | | | |
| The mean delay caused by passengers who mentioned a sub optimally correct train (computed over all passengers) | 28 min | 4 min |
| Mean search time | 7s | 4s |
| |
Table 6. Subjective evaluation of the passengers (n = 148)
| | | Destinations |
| | Trains | ordered | | | by times | alphabetically |
| Rating score (min 0, max 10) | 6.1 | 8.2 | | Which is better | 15% | 85% |
|
| |
The number of errors
Incorrect choices were made by 3% of respondents for the alphabetical and by 5% for the chronological indicator. This might be expected since the task was a relatively easy search task. Such errors as occurred were visual confusions; for instance:
Haarlem instead of Heerlen
23 instead of 32.
05 instead of 50 |
Search time
On mean search time the chronological indicator took 7 seconds and the alphabetical indicator 4 seconds, per train found.
|
|
The delay
The mean delay that the passengers would have experienced if actually taking the train they mentioned was, when using the chronological indicator, four times greater than when using the alphabetical indicator. 3.6 min to 0.9 min. |
Performance with difficult destinations
Junction stations give more problems than other stations, and with these the chronological indicator scores much more poorly and the alphabetical indicator scores only a little worse than for line stations (compare Tables 4 and 5 above). The cause of these problems is that the trains schedule of junction stations is complicated. These kinds of station can often be reached by stop train and express train; they can be reached directly or by a detour; and several trains for these kinds of station may be leaving at almost the same time. The results allow for two conclusions. First, the alphabetically ordered destinations indicator is better than the time-ordered one, and particularly for more complicated situations. Second, since the proportion of passengers actually using junction stations is much larger, the results may be an underestimate of the advantages to be gained from using the new structure. |
The evaluation of the passengers
After having searched for five departure times on an chronological indicator and five on an alphabetical indicator, the passengers scored each indicator (see Table 6). The latter was rated as better by 85% of the passengers.
|
6. Final word | We would advocate that information for passengers will greatly improve when the basis for the structure of the information is destinations ordered alphabetically, and not trains or flight numbers. However, as can readily be seen in practice most railway and airport authorities have another opinion. |
The database that controls the indicator is based on train numbers and to construct a destinations indicator would mean converting these train numbers to destinations. For other technicians such as train drivers, bus divevdrs and pilots, trip numbers are more essential components of their mental model than are destinations. | A further argument that may be leveled against an alphabetical indicator is that it is not consistent with other information systems such as railway timetable books and timetable posters. However, from data reported in this paper it can be concluded that passengers hardly notice and remember how information is structured, and we have reason to believe that in this case consistency lacks positive effects. Finally, one should not reject a good system because it is inconsistent with existing bad systems. | A common argument against ergonomic design that the costs are higher, but this does not apply to a destinations indicator. They are in fact about 20% to 30% cheaper. The reason for
|
7. Acknowledgements |
The author is indebted to his colleagues in the Ergonomic Advisory Group of Netherlands Railways
for their help through discussions, field observations and the construction of research instruments.
|
References |
Adams, W., Venemans, P.J. & Mannaerts, A.A.J. , (1984). Bewegwijzering binnen gebouwen: een aanzet voor en checklist. Tijdschrift voor Ergonomie. Vol. 9, no 4 pag. 387-391
Best, J.B. , (1989). Cognitive Psychology St. Paul etc.: West Publishing Cy.
Brebner, J. , (1982). Environmental Psychology in Building Design London: Applied Science Publishers Ltd.
Corlett, E.N., Manenica, I, & Bishop, R.P. , (1972). The design of direction finding systems in buildings Applied Ergonomics. no june pag. 66-69.
Easterby, R., & Zwaga, H. , (1984). Information Design Chichester etc.: Wiley and Sons Ltd.
Hancock, P.A. (Ed.) , (1987). Human Factors Psychology Amsterdam: Elsevier Science Publishers.
Harrigan, J.E. , (1987). Human Factors Research: Methods and Applications for Architects and Interior Designers. Amsterdam: Elsevier.
Janssen, W.H. , (1986). Een evaluatie van enkele systemen voor lokale bewegwijzering Soesterberg: Inst. voor Zintuigfysiologie.
Neisser, U. , (1976). Cognition and Reality Principles and implications of cognitive psychology. San Francisco: Freeman & Company.
Spijkers, W.A.C., Bechtold, M.F.V.M., & Venemans, P.J. , (1985). Indelen van bestemmingen Tijdschrift voor Ergonomie. Vol. 10, pag. 2-9.
Verhoef, L.W.M. , (1984). Welke informatie moet op een treinaanwijzer staan? Utrecht: NV Nederlandse Spoorwegen, PZ 2.4. /CAE rapport no 349c.
Wright, P. , (1988). Functional literacy: Reading and writing at work. Ergonomics. Vol. 31, pag. 265-290.
|
More applied cognitive psychology for design, other than public |
|
| 
