Objective and Subjective Methods for Evaluating the Usability of Schematic Maps: The Case Against Informal Expert Assessments

Figures & data

Figure 1. Since 2000 the official RATP Paris Metro map has been a conventional design comprising horizontal, vertical, or 45° diagonal straight lines joined by tightly-radiused curves. The equal intervals of angle rotation make this a regular octolinear design. Image and design ©RATP, all rights reserved, reproduced with permission.

Figure 2. SNCF operates frequent services on Paris RER lines, and the organization has issued its own version of the Paris Metro map which is best described as multilinear, using straight lines but at a variety of different angles. For historians of the Paris Metro map, this version is similar to maps issued by RATP prior to the design shown in Figure 1 (see Ovenden, 2009). Image and design © SNCF, all rights reserved, reproduced with permission.

Figure 3. The Paris Metro map issued by Île-de-France Mobilités (formerly STIF) took a different approach to the RATP design. Octolinear lines dominate, but some deviate from this, and widespread use of gentle curves means that the design rules are best described as hybrid. Image and design © Île-de-France Mobilités, all rights reserved, reproduced with permission.

Figure 4. Octolinear schematized central regions of the London Underground. The upper layout is topographically accurate, the lower layout distorts, for example, the relative positions of Paddington and Notting Hill Gate (in a similar way to the official configuration). Guo (2011) observed that around 30% of journeys between Paddington and Bond Street are via Notting Hill Gate, with the implication that the configuration of the official map implies that this route is a reasonable option. The route via Baker Street is considerably shorter in reality. Image and designs © Maxwell J. Roberts, all rights reserved, reproduced with permission.

Figure 5. The curvilinear Paris Metro map, designed by the author, consistently out-performs the RATP official octolinear design (see Figure 1) in terms of time necessary to plan complex (i.e. two-transfer) journeys between pairs of stations. Image and design © Maxwell J. Roberts, all rights reserved, reproduced with permission.

Figure 6. Octolinear London Underground map created by computer algorithms overseen by Martin Nöllenburg and Soeren Terziadis at TU Wien (Roberts, 2019b). Image and design © Martin Nöllenburg and Soeren Terziadis, all rights reserved, reproduced with permission.

Table 1. The criteria for schematic map design as outlined by Nöllenburg (2014) for the purpose of computer-automated schematic map layout.

(R1) Do not change the network topology. This important rule prohibits structural distortions such as modifying the circular edge orders around vertices or introducing additional edge crossings. In graph theoretical terms the combinatorial embedding of the layout is preserved. This rule is respected by almost all methods and thus already included in the definition of Problem 1. (R2) Restrict edge orientations. The vast majority of metro map layout methods uses the octilinear set of orientations, that is, horizontal, vertical, and ±45°-diagonal orientations. Other orientations such as hexalinear (based on 60° angles) are possible, too. Some methods do not use straight-line edges at all and resort to curvilinear edges, for example, based on Bézier splines.

(R3) Draw each individual metro line as straight/monotone as possible and avoid sharp turns. For traditional polyline drawings, this implies to use as few bends as possible with preferably obtuse angles. For curvilinear drawings, preferably uniform curvature and few inflection points should be used. (R4) Metro lines pass straight through interchanges. Interchange stations are higher degree vertices, where it is particularly important that metro lines are visually easy to follow without ambiguities. This is supported if no metro line changes its orientation in an interchange. (R5) Use large angular resolution. This rule aims to distribute incident edges evenly around vertices.

(R6) Minimize geometric distortion and displacement. Many approaches try to stay as close to the input geometry as possible in order to maintain the user’s mental map of the city and the resemblance to geography. Some methods apply this rule only locally, that is, the relative positions of pairs of adjacent vertices should be maintained. (R7) Use uniform edge lengths. Since distances in a metro map are not linked to geographic distances, any edge in the layout ideally has the same length. This often implies that dense parts of the network in the city center are enlarged and peripheral stations move closer together.

(R8) Keep unrelated features apart. This rule ensures that there is some minimal clearance between non-incident vertices, edges, and around station labels. (R9) Avoid large empty spaces in the map. This rule asks for a balanced local feature density in the whole map. (R10) Use unobtrusive and clearly legible placement of station labels. The precise interpretation of this rule differs between different layout algorithms. Overlapping labels and occlusions are usually prohibited. Horizontally aligned text is mostly preferred, but horizontal metro lines can also be labeled with diagonal labels. Often all labels of stations between two neighboring interchanges are placed coherently on the same side of the path between them.

Table 2. Good Practice in Diagram Design: Prescriptions for optimizing schematic map design taken from Ovenden (2009).

Remove unnecessary topography though leaving important waterways or green spaces – often stylized – is sometimes considered useful for a vague orientation. No streets are shown though occasionally key highways like arterial roads may be acceptable. Straighten-out lines – avoid changing direction of a line unless totally necessary. Use only horizontal and vertical lines and one diagonal angle. 45 degrees is the most common angle for all diagonals though 20, 30, 35, 60 and 65 have also been utilized to good effect. Create at least one simple axis – north-south or east-west if relevant, though stylized ‘circle’ lines are also utilized as perfect spheres. Compress the suburbs or areas where stations are spaced far apart geographically. Enlarge central area or areas where stations are in close proximity geographically. Do not change line direction under a station marker especially at correspondences. Do not change line direction more than once between stations – it hampers clarity. Leave adequate space for the text of the station name – it’s as crucial as plotting the lines. Station names should NEVER crash over lines. Station names must be close to station marker and clearly relate to only one station.

Equalize station spacing wherever possible – it looks neater. Line width should be in aesthetic balance with the overall space and point size of station names. Make all lines the same width unless they are part of another service i.e. Tram/mainline rail, etc. Station markers should be universal (except at interchanges). Station markers should be universal (except at interchanges) Make a clear distinction between ordinary stations and interchanges – this greatly helps passengers planning journeys. Interchange station markers should be universal though at extremely complex or lengthy interchanges a white-line connector or variation of this symbol may be necessary as long as it fits into the overall style of others. Station markers should be ticks or ‘blobs’ dots, open circles and ‘blobs’ are the most common. Ticks come second. Avoid ‘line breaks’ as these can be extremely confusing. Stations names should all be horizontal – if at an angle, all should follow same angle; mixing horizontal names with angled ones looks bad. Background colour is generally white – pastel shades can be used if they do not clash with the line colours. Very dark backgrounds (black, dark blues, etc.) can work with the same proviso but primary colour backgrounds should be avoided. Simplify complex interchanges/service patterns – badly drawn correspondences will let down the entire diagram.

Keep all station names the same size though in systems where terminal stations play a key role, there may be a good reason to enhance their appearance in bold/caps/reversed out, etc. Keep all stations on the same side of the line – this is rarely possible all over but looks neat on a long simple line. Use upper and lower case text – it is more humane. Stay within corporate identity – the in-house operator style should permeate every feature of the diagram—anything that does not fit will look hideously out of place when the diagram is on display in situ. All text must be in the operator’s house font – never condense/alter a font to squeeze text in to a badly designed space. Station names to read EXACTLY as on the platform signage – this avoids passenger confusion. Abbreviations on the map to squeeze-in text are unacceptable – however heritage signage may still show abbreviations, in this case use the full name on the diagram. Diagrams do not need to be totally abstract – with modern computer graphics and good design most diagrams can give reasonable reflection of at least the key features of city geography without losing all points above. REMEMBER: Someone else may re-design your diagram at a later stage!

Figure 7. Three configurations of the same section of the London Underground network from the stylized designs in the internet rating study, illustrating octolinear (top), multilinear (centre), and curvilinear (bottom) approaches. All three have the same priorities, the simplest possible line trajectories inside the Circle Line. Image and designs © Maxwell J. Roberts, all rights reserved, reproduced with permission.

Figure 8. The matrix of nine London Underground maps designed by the author and used in the internet rating study of Roberts et al. (2017). Image and designs © Maxwell J. Roberts, all rights reserved, reproduced with permission.

Table 3. Ratings of schematic map usability comparing Expert and Non-expert groups, reanalysing data from Roberts et al. (2017). Responses have been scaled such that 100% would indicate every single person rating a map as easy to use and 0% would indicate every single person rating a map as difficult to use.

	Expert Group (n = 274)	Non-expert Group (n = 375)	Overall
Ratings of maps with identical design rules
Octolinear maps	63%	65%	64%
Multilinear maps	35%	41%	39%
Curvilinear maps	32%	28%	30%
Ratings of maps with identical design priorities
Stylized maps (simple line trajectories, topographical inaccuracies)	62%	60%	61%
Geographical maps (complex line trajectories, topographically accurate)	37%	41%	39%
Compact maps (complex line trajectories, topographical inaccuracies)	31%	34%	33%

Table 4. Criteria for rating individuals as holding a coherent simplicity theory of schematic map usability (top), and a coherent octolinearity theory (middle) along with the ratings given by an individual that would yield a categorization of simultaneous simplicity and octolinear theories.

	Octolinear	Multilinear	Curvilinear
Criteria for categorising an individual as holding a coherent simplicity theory
Stylized	Octolinear stylized given highest rating in this column	Multilinear stylized given highest rating in this column	Curvilinear stylized given highest rating in this column
Geographical	Rating exceeded by Octolinear stylized map	Rating exceeded by Multilinear stylized map	Rating exceeded by Curvilinear stylized map
Compact	Rating exceeded by Octolinear stylized map	Rating exceeded by Multilinear stylized map	Rating exceeded by Curvilinear stylized map
Criteria for categorising an individual as holding a coherent octolinearity theory
Stylized	Octolinear stylized given highest rating in this row	Rating exceeded by Octolinear stylized map	Rating exceeded by Octolinear stylized map
Geographical	Octolinear geographical given highest rating in this row	Rating exceeded by Octolinear geographical map	Rating exceeded by Octolinear geographical map
Compact	Octolinear compact given highest rating in this row	Rating exceeded by Octolinear compact map	Rating exceeded by Octolinear compact map
Matrix of ratings for an individual to be categorized as simultaneously holding coherent simplicity and octolinearity theories
Stylized	Easy to use	Neutral	Neutral
Geographical	Neutral	Hard To Use	Hard To Use
Compact	Neutral	Hard To Use	Hard To Use

Ovenden, M. (2009) Paris Underground: The Maps, Stations, and Design of the Métro New York: Penguin Books.

 Google Scholar

Guo, Z. (2011) “Mind the Map! The Impact of Transit Maps on Travel Decisions in Public Transit” Transportation Research Part A 45 pp. 625–639.

 Google Scholar

Roberts, M.J. (2019b) Tube Map Travels London: Capital Transport Publishing.

 Google Scholar

Nöllenburg, M. (2014) A Survey on Automated Metro Map Layout Methods. Schematic Mapping Workshop 2014, University of Essex, 2nd to 3rd April. https://www.researchgate.net/publication/281587114_A_survey_on_automated_metro_map_layout_methods (Accessed 12th June 2023).

 Google Scholar

Roberts, M.J., Gray, H. and Lesnik, J. (2017) “Preference Versus Performance: Investigating the Dissociation Between Objective Measures and Subjective Ratings of Usability for Schematic Metro Maps and Intuitive Theories of Design” International Journal of Human Computer Studies 98 pp.109–128.

 Web of Science ®Google Scholar