One person, one dot

Versión en Español

The dot map represents an increasingly used tool for visualising the spatiality of social data. Emerged in the 19th century by the research performed by John Snow (not the one from Game of Thrones), this technique has been converted into one of the more effective tool for exploring spatial concentration patterns of different phenomena. In this post, we will analyse the application of this technique to the study of ethnic segregation.


Fig.1. The Chicago Boundaries project, William Rankin (2009)

PR1 - Rankin

(Click on the image to access the respective webpage)


The ethnic segregation maps


In 2009, a professor of history of science at the Yale University, William Rankin, employed a dot map to represent the spatial distribution of ethnicity and income in Chicago. Using aggregated data at the census block level from U.S. 2000 census, Rankin generated a map where each dot represented 25 individuals of a certain ethnicity or income group. The results provide us with an instantaneous view of a phenomenon increasingly salient in American society: class and ethnic segregation.

In 2013, Dustin Cable, demographer and data scientist, reproduced the same technique to generate the U.S. “racial” dot map. Based on the method developed by Rankin, he expanded the coverage to the entire territory and increased the resolution from 1:25 to 1:1, meaning that each dot would correspond to one individual, something around 308.7 million dots, according to 2010 U.S. census. His “racial” dot map has provided us with a fine-grained, detailed and visually pleasant (by its excellent use of colour) of the ethnic segregation observed in U.S. cities.


Fig.2. The “Racial” dot map, Dustin Cable (2013).

PR1 - Cable

(Click on the image to access the respective webpage)


Cable’s project was reproduced, in the same year, for the UK and Toronto. In 2015, the same was done for Brazil, by a team of the Nexo journal. The Brazilian example provides us with an overview of a territory divided by the colour of the skin, with black (“pretos”) and brown (“pardos”) predominating in the north and a mostly white south. Nonetheless, when the user increases the zoom, it reveals the existence of a great deal of hybridity in metropolitan cities, with varying patterns according to different regions of the country.


Fig.3. Melbourne Ethnicity Dot Map, Monash City Science Lab (2015).

PR1 - Monash

(Click on the image to access the respective webpage)


City Science, a lab based in the Monash University in Melbourne, Australia, developed the same map in 2015 focusing one city. The reduced spatial coverage allowed them to add more interaction to Cable’s original map, making possible to select all ethnic groups or just one. This was unthinkable for the huge volume of information included in the previous examples. The main advantage of this functionality is to clearly highlight where each group is concentrated, as well as its relative density.

Despite less visually appealing, The New York Times has gone one step further by generating a similar map, but with much more interactivity. This was achieved by the establishment of a dot for each thousand individuals. Two main aspects must be highlighted. The map presents a predefined zoom of U.S. most important cities, allowing the user to navigate directly to them. Besides, and this is its more substantial contribution, this is a scale sensitive visualization. Each ethnicity percentage is aggregated and presented according to the level of zoom selected: census tracts when city-level zoom; and counties when a supralocal scale is employed.


Fig.4. Mapping Segregation, The New York Times (2015)



(Click on the image to access the respective webpage)


Why these ethnicity dot maps are interesting?


The interest of this kind of map is explained by five determinant factors:

 Density mapping. Dot maps are excellent instruments to identify denser or sparser zones, as well as to point out those areas where clustering of cases exists.

Many scales, one map. This feature allows exploring geographic patterns of distribution of different ethnicities in distinct scales: from the entire country to a block.  Alternative scales (or zoom levels) provide us with different perspectives or images of the phenomenon under scrutiny. Besides, it makes explicit those biases related to the selection of the level of analysis. This occurs because, at a more aggregated level, we can just observe predominant colour patterns. At an intermediate scale, some colour nuances can be perceived, while in a maximum zoom individual dot colours can be identified.

Multiple categories in the same area (internal heterogeneity). Many groups can be represented in a same area unit at the same time by just one map. This feature converts a dot map in one of the most efficient visualization methods according to the ink-data criteria, employed by Edward Tufte in his Visual Display of Quantitative Information, to assess the informative quality of scientific graphics.

Discrete phenomena into continuous surfaces. This type of map presents dots in a spatial continuum, from denser ethnic concentration to sparser zonas where the existence of a certain group is reduced to just few dots. This kind of visualization relativizes and reformulates the notion of frontier, from its formal or administrative conception to other closer to the empirical distribution of the studied phenomenon.

Oversight and detail simultaneously. Its potential to reveal heterogeneity also allows the emergence of quite complex synthetic pictures on how urban space is composed. We are able to identify not just zones with denser population, but also areas in which there are more (or less) segregation, the dominant ethnic group and, in hybrid zones, what combination among groups establishes their personality and specificity.


Precautions concerning the use of this method


Despite all advantages mentioned before, it must be highlighted that some precautions must be taken when it comes to dot maps, since some key details can endanger the entire work. We present three here:

Cautious use of colour. The number of colours must be limited or thought in terms of degree, depending on the type of the variable being used: categorical or continuous. Our brain is not able to distinguish among a large number of colours. We also should avoid the use of derivative colours. For instance, in a map where White are represented in red, Black using yellow and Hispanic in orange, those neighbourhoods where a hybrid between white and black would be perceived, at certain scales, as predominantly orange (and, thus, interpreted as Hispanic).

More dots, less interactivity. The larger the area covered by the map, the lesser the capacity of interaction with users at a one individual, one dot scale. As one can observe in The New York Times and Melbourne examples, the options are either reducing the analysis to one city or readjusting the scale of representation to 1:1000 individuals.

Knowledge production/Knowledge diffusion ratio. What is the original contribution of this kind of map to the study of urban segregation? If we focus the studies of big metropolises and their residential dynamics, the answer to this question would be: none. Most of these cities are far from covered. Nonetheless, we should answer yes when we refer to all other urban (and rural) areas that receive much less attention from urban studies. In this case, a comprehensive map, as the ethnicity dot map of an entire country, besides revealing segregation patterns in medium and small localities, can also add a holistic perspective to the phenomenon under scrutiny, a very rare thing to occur in this type of study.


Application to other social phenomena


The major contribution of this map is put probably on its applicability to other social phenomena. The fundamental idea of representing discrete phenomena into continuous surfaces shows enormous application potential in diverse social sciences fields. Some examples can be observed in Rankin’s map of income or the map of jobs, but the method can be applied to many other events. This technique can also be employed to fiscal revenues or expenditures, crime or voting behaviour. Anyway, the potentiality of a dot map is evident and its combination with resources that allow some interactivity, as we could experience in the NYT and Melbourne examples, enables us to take full advantage of this active tool of knowledge building.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s