Map are neither objective nor exhaustive. They are only an interpretation of reality. Map-making is the result of a series of choices, subjective readings, a way of seeing things, sometimes involving a manipulation of reality, often an approximation of it. The resulting graphic images should enable the user to perceive the information represented instantaneously and memorize it easily. Effective images rely on the rules of graphic semiology. The representational conventions of traditional geography – north up, Europe-centered, and the Pacific Ocean cut in two, with conventional limits – no longer suffice to map the world.

Any mapping project raises the question of the availability, quality and coherence of sources and statistical data, these reflecting the actors that produce them. Despite the difficulty of assembling the relevant data on certain subjects and the challenges involved in representing the density of complex exchange networks, the present work attempts to achieve a balance between a vision of juxtaposed states and that of an increasingly transnational and global world, for each of the six major themes it deals with.

Identification of sources and data collection

Researching, collecting and processing data and displaying the results graphically are the main stages in the process of designing the document. Decisions must be made according to rigorous criteria, because they condition the interpretation of the graphic documents produced (maps, charts, matrices, etc.). Choosing a theme to map means highlighting a specific case among others. The first and most important decision is to clearly identify the subject to be presented graphically.

The digital revolution has made social science statistics more accessible, more bountiful and of better quality (despite shortcomings in certain areas). The complexity of the phenomena dealt with in this atlas means that they need to be assembled, compared and connected them, bearing in mind the statuses and rationales of those who produced them. One must also accept the fact that maps and charts only show a compromise, at a given point in time, between a research question, more or less verifiable data, and choices as to how to represent them.

So-called official statistics (produced by national statistics bureaus, ministries, government agencies, etc.), taken from national censuses, national population counts and surveys, are the most abundant. The quality of this data in terms of exhaustiveness, comparability, historicity, topicality, and so on, depends on the administrative, technical and conceptual capacities that states are willing and able to bring into play. These may be amnesiac, unavailable, even falsified, particularly in autocratic states; or mediocre and lacking reliability in poorer states where civil registries are incomplete. Furthermore, bureaucrats who produce data do not all enjoy the same guarantees of independence, which influences the quality of the results, especially where migration, employment and unemployment, and religious or ethnic affiliation are concerned. Lastly, some data are the result of administrative operations by state actors; functioning, in this case, more as management tools than knowledge instruments.

International organizations (IO) collect, compile, harmonize and publish a considerable body of statistical data produced by their member states. They also conduct their own studies, calculate estimates and projections, and construct composite indicators. These are a standard reference, but the rationale behind their production is strongly tied to their objectives (rankings, evaluations, etc.). IOs also produce summary reports on various themes, on their own or in conjunction with other inter-state organizations such as the OECD and regional development banks. Abundance, however, does not solve the cartographer’s basic problem, which is to manage to understand and chart a world of continuous transnational flows using mostly state-produced information.

The major NGOs (Doctors without Borders, Reporters without Borders, Amnesty International, Greenpeace, Transparency International, etc.) regularly produce data, analyses and reports that are made widely available online and that serve as a basis for their advocacy activities. Transnational by definition, they have fewer constraints, but their activist aspect must always be factored in. Data put out by certain research centers are also standard references in certain areas of study such as poverty or conflict studies.

Private actors, particularly businesses, are often in competition and thus do not readily reveal their data. For instance, revenue figures for global corporations are collected by private agencies and available in the trade press, but they can be difficult to cross-check with other sources. Other information can be gleaned from annual activity reports designed for shareholders, the publication of which is mandatory for companies quoted on the stock exchange. These reports are interesting in what they reveal about the organization and the company’s own logic, but there is often a dearth of visualized data. Opacity prevails as regards global financial flows, which moreover are extremely mobile and partly illicit. Lastly, many statistics that can serve as a basis for gainful activity are only available for a fee. Little is known about illegal actors (illegal migrations, tax havens, mafias, trafficking, informal or forced labor, etc.), although the global risks these invisible segments of the world economy are beginning to be studied more closely by certain UN agencies such as the United Nations Office against Drugs and Crime (UNODC) and the International Labor Organization (ILO).

Scales of time and space

The continuous flow of information that individuals are immersed in leaves little room for taking account of longer timeframes that would shed light on events. Used to illustrate current events, figures set to images are too often isolated from their context. The sources are generally not referenced or incomplete and not necessarily crosschecked. Especially in economics, rates of change from one year to another are often highlighted, leaving aside absolute values and the main orders of magnitude. Paradoxically, an excess of information often produces amnesia, and failing to contextualize the present within broader trends does not help to convey complexity. It is equally important to strive for multiscalarity both in terms of social space as well as time.

Even in the event that a temporally coherent series that holds true for all countries permits diachronic and synchronic comparisons, the result can prove quite unsatisfactory. Indeed, often significant economic, demographic and social inequalities among regions within the same state are smoothed over by sometimes misleading national averages, particularly for large countries. It is only possible represent social-spatial diversity in cartographic reasoning and projection by using often fragmentary subnational data that are poorly connected and difficult to compare from one country to another. Only the European Union, with Eurostat, has a finely-tuned and coherent statistical apparatus. Large federal states supply data about their federated entities, but national nomenclatures do not coincide. Lastly, for the world’s major conurbations, there is very little reliable and comparable data. In all, the widening gap between the abundance of data and the complexity of processes underway shows that using old or incomplete tools to approach a globalized, global and local (glocal), transnational world involving multiple actors is increasingly difficult. Researchers, cartographers, teachers, experts, journalists, students and users of the present resource are urged to proceed with vigilance and critical detachment.

Seeing the world: forms and points of view

Cartographic projection is a technique for transcribing a spherical reality – the Earth globe – onto a flat surface – the map. For mathematical reasons, projection cannot at once accurately represent distances, angles and surface areas. Each projection therefore selectively distorts one of these dimensions. The position of the map’s center also influences its interpretation, because it gives greater importance to the areas placed in the middle.

This combination of projection and centering causes considerable discrepancies in the “shape” of the world. For instance, the Mercator projection distorts surfaces closer to the poles, thus exaggerating the size of Northern hemisphere countries. The “plate carrée” (“square plate”) projection, cylindrical as well, represents surfaces more accurately. Conversely, some projections concentrate discontinuities and distortions in the oceans to preserve the shape of the continents. This is true for the Waterman and Fuller projections, which, by separating the continental land masses, make it easier to trace flows, for the Goode projection, which divides the continents into vertical lobes, and the Atlantis projection, asymmetrical around the Atlantic Ocean. Lastly, the Bertin 1953, Natural Earth and Mollweide projections offer interesting compromises for what is known as thematic cartography, combining accuracy of continental surfaces and shapes with a compact design.

There is a nearly infinite range of possibilities to choose from in centering a cartographic projection. Certain parameterizations produce a world that is difficult to recognize, so conditioned are we to our Europe-centered, North-up view. It is a difficult habit to break. Depending on the projection, decentering can be performed on the longitude (sliding East-West), the latitude (flipping North and South), rotation or all three at the same time. Khartis, the online cartography tool, makes it easy to play with these distortions and points of view.

Various cartographic projections

Seeing time: charts and diagrams

Data visualization in the form of charts and diagrams proves just as effective as maps. They provide alternatives when spatialization is incomplete, non-existent or even impossible, or else when one wants to simplify information into categories. Charts are highly suitable for showing the chronological dimension.

The vertical scale should be carefully thought out. In addition to the base value, where the zero axis is sometimes truncated, it is important to differentiate between arithmetic and logarithmic progression. The former, the most common, shows a succession of peaks whereas the latter depicts a change using the slope of a curve. The arithmetic scale often produces an impression of exponential change without visualizing the low values; the logarithmic scale transcribes a more accurate evolution from one date to another and makes it possible to compare trends even when values are very low.

Sets of curves are suitable charts for comparing chronological profiles. Curves are sorted vertically to bring out similarities in time (start, peaks, troughs and end).

Vertical scale graph and collection of curves

Sources: Louis Johnston and Samuel H. Williamson, “ What Was the US GDP Then ?”, measuringworth.com; US Department of Homeland Security 

Types of data

Maps and diagrams are constructed images combining graphic elements, symbols and colors. They are therefore partly subjective. Nevertheless, they are the result of a scholarly approach that relies in particular on the nature of the data. Two main families of data can be identified by the relationships between values.

Quantities or numbers (static or dynamic: population, people flows, market capitalization, GDP, etc.) are sorted by increasing or decreasing order, as they describe a proportional relationship. These quantities can be compared to another, in which case they describe relative quantities (percentages, rates, densities, yields, etc.). They are also ordered, but often grouped into categories that have a hierarchical relationship. This partition – or discretization – is not done at random. Statistical methods (mean, standard deviation, median, etc.) or manual methods (observed thresholds) are used. These methods are varied and produce different maps. For maps to be comparable it is vital that they use the same type of discretization.

Qualitative data can be ordered (in logical order, by date, etc.). They are then processed in an identical manner to relative, or non-organized quantities (presence/absence of an NGO in a country, membership of a regional organization, etc.), and have relationships of difference or resemblance.

(Carto)graphic transcription

New trends in cartography developed thanks to progress in the semiology of graphics beginning in the 1950s-1960s, driven by Jacques Bertin’s pioneering research. In a clear departure from the preceding period in which topographical (or survey) maps and regional sketch maps predominated, Bertin synthesized a set of rules making it possible to process and transcribe information visually. His research on visual variables combined with the “graphic primitives” (points, lines, areas) revolutionized cartography in such a way as to finally produce maps “to look at” (using symbols and colors) and not “to read” (with an abundance of text and figures). Although graphic semiology was subsequently adapted, enriched and sometimes rivaled with the emergence of new tools and practices (geographic information system, computer graphics, dynamic cartography, dataviz, etc.), it remains a robust and essential visual grammar.

Graphic choices should be made in coherence with the data represented. Proportionality, order and difference call for different graphic responses. Mixing them up makes the image hard to read at best, at worst inaccurate, or even manipulative.

Proportionality and order

The proportional relationship between absolute quantities is rendered by varying the size of points and thickness of lines. Relative quantities already classified are represented by color gradients, organized from lightest to darkest (monochrome). The visual order adheres to the order of the data. A visual break in the color gradient can serve to highlight a particular phenomenon (for instance, a shift from positive to negative change).

Comparing these two types of maps considerably enriches reflection. The two images can be juxtaposed or superposed. In the case of a population, the total numbers show the size of one country compared to another, whereas values per 100 inhabitants represent the intensity of a phenomenon within each country (making it possible to compare countries of very different sizes).

Absence of order, difference or typology

The use of different color tones does not indicate a hierarchy; it only expresses difference or resemblance. Size, gradient, color, shape and orientation of the pictograms are often combined to emphasize the separation of the various elements so that the superposition of qualitative data does impair legibility.

These methodological considerations are far from exhaustive; we invite the reader to systematically cast a triple critical gaze on the maps – scrutinizing their message, design tricks and the data used – to see beyond the impression of a reflection of reality that they convey at first glance.

Visualizing proportionality, order and difference

Sources: IDMC; BP; European Commission; The Electoral Commission and https://data.gov.uk; AidData, Hanban and Chinese Ministry of Foreign Affairs; World Bank and World Inequality Database; compilation of various German atlases and based on G. Duby, Grand Atlas historique, Paris, Larousse, 1997; WHO and Population Division - United Nations.

To quote this article

" Representing the World " World Atlas of Global Issues, 2019, [online], accessed on Mar 15 2021, URL:
https://espace-mondial-atlas.sciencespo.fr/en/topic-introduction/article-0A03-EN-representing-the-world.html

References

  1. Bertin, Jacque
  2. Brunet, Roger, La Carte, mode d’emploi, Paris/Montpellier, Fayard/Reclus, 1987.
  3. Khartis, Sciences Po – Atelier de cartographie, https://www.sciencespo.fr/cartographie/khartis
  4. Lambert, Nicolas and Zanin, Christine, Manuel de cartographie. Principes, méthodes, applications, Paris, Armand Colin, 2016.
  5. Lévy, Jacques, Poncet, Patrick and Tricoire, Emmanuelle, La Carte, enjeu contemporain, Paris, La Documentation française, 2004.
  6. New Challenges for Data Design, edited by David Bihanic
IntroductionWorld, Space, TimeContrasts and InequalitiesRich and Poorback to top