Be suspicious of maps, not of migrants

Do the maps tell you the truth or are they images to be used with extreme caution? The following demonstration is designed to sharpen your critical thinking skills. It deals with the so-called “migration crisis” in Europe, with the example of Syrian migration in 2015. judge for yourself…

R Packages

Package loading.

library("sf")
## Linking to GEOS 3.6.2, GDAL 2.2.3, PROJ 4.9.3
library("countrycode")
library("readxl")
library("cartography")

Data download, import & handling

Statistical data are from United Nations, Department of Economic and Social Affairs, Population Division (2017). It concern international migration between all the countries of the world at several dates. At this stage, the first step consists in selecting the year 2015 and the Syrian nationality and joining them with the basemap coming from the Natural Earth site.

Data sources

mapfile <- "http://www.naturalearthdata.com/http//www.naturalearthdata.com/download/50m/cultural/ne_50m_admin_0_countries.zip"
migfile <- "http://www.un.org/en/development/desa/population/migration/data/estimates2/data/UN_MigrantStockByOriginAndDestination_2015.xlsx"
popfile <- "http://api.worldbank.org/v2/en/indicator/SP.POP.TOTL?downloadformat=excel"

# download world basemap
download.file(url =mapfile, destfile = "data/countries.zip")
unzip("data/countries.zip", exdir = "data", unzip = "internal")
file.remove("data/countries.zip")

# Downlad statistical data
download.file(url=migfile, destfile="data/mig.xlsx")
download.file(url=popfile, destfile="data/pop.xls")

# import world countries
countries <- st_read(dsn = "data/ne_50m_admin_0_countries.shp", stringsAsFactors = F)
countries <- countries[,c("ADM0_A3", "ISO_A2","NAME_SORT","geometry")]
colnames(countries) <- c("ISO3", "ISO2","name","geometry")

# import migration data
mig <- data.frame(read_excel("data/mig.xlsx", skip = 15, sheet = "Table 16"))
mig$ISO3 <- countrycode(sourcevar = mig$X__2, origin = "country.name", destination = "iso3c")

# import population data
pop <-  data.frame(read_excel("data/pop.xls", skip = 3, sheet = "Data"))
pop <- pop[,c("Country.Code","Country.Name","X2015")]
colnames(pop) <- c("ISO3","name","pop2015")

# merge countries geometries and mig data
countries <- merge(countries[,c("ISO3", "ISO2","name")], mig[,-(1:5)], 
                   by.x="ISO3", by.y="ISO3", all.x=TRUE)

# merge countries geometries and pop data
countries <- merge(countries, pop[,c("ISO3","pop2015")],by.x="ISO3", by.y="ISO3", all.x=TRUE)

# modify col names with ISO3 codes
coln <- countrycode(sourcevar = names(countries)[7:238], origin = "country.name",destination = "iso3c")
names(countries)[7:238] <- coln
names(countries)[48] <- "GGY"

# Keep only syrians
countries <- countries[,c("ISO3","name","SYR","pop2015","geometry")]
colnames(countries)

# export the layer
st_write(countries, "data/countriesmig.geojson")

Study area

When we draw a map, an important step is to define the study area. Here, we choose the LAEA projection and a box focused on the european union.

# import
countries <-  st_read(dsn = "data/countriesmig.geojson", quiet = TRUE)

# Proj WGS84 => LAEA
crs <- "+proj=laea +lat_0=52 +lon_0=10 +x_0=4321000 +y_0=3210000 +ellps=GRS80 +units=m +no_defs"
countries <- st_transform(x = countries, crs = crs)

# Bounding box
extent <- c(2800000, 1340000, 6400000, 4800000)
box <- st_as_sfc("SRID=3035;LINESTRING(2800000 1340000, 2800000 4800000, 6400000 4800000, 6400000 1340000, 2800000 1340000)")

# plot
par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border ="white", lwd = 0.5)
plot(box, lwd=2, add=T)
layoutLayer(title="Study Area", sources = "Natural Earth, 2018", author = "Lambert & Bahoken, 2018",scale=NULL)

Cartographic template

In this analysis, we seek to represent Syrian migration in the European Union. Thus, we select the 28 countries of the Union for which we will represent statistical information. At this stage, a first cartographic template can be displayed

UEcode <- c("AUT", "BEL", "BGR", "CHE", "CYP", "CZE", "DEU", "DNK", "EST", "ESP", "FIN", "FRA",
            "GRC", "HRV", "HUN", "IRL", "ISL", "ITA", "LIE", "LTU", "LUX", "LVA", "MKD",
            "MLT", "NLD", "NOR", "POL", "PRT", "ROU", "SWE", "SVN", "SVK", "GBR")

UE2015 <- countries[countries$ISO3 %in% UEcode,]

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, 
     lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))

plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
layoutLayer(title="Cartographic Template", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

Syrian refugees towards Europe

Then, we design a map representing the number of Syrian migrants in the countries of the European Union. On this map, we can see that the 3 most welcoming countries are Sweden, Germany and France.

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, 
     lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))

plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")

propSymbolsLayer(x = UE2015, var = "SYR",
                 inches=0.5, col="#91591D",legend.frame = T, 
                 legend.pos = "topright", legend.title.txt = "Number of persons")
labelLayer(x = UE2015[UE2015$SYR > 10000,], txt="name", halo=T,cex=1)
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

In share of pop

If we consider the number of Syrians hosted in proportion to the population, the main host countries are Sweden, Norway, Denmark, Cyprus and Malta.

par(mar = c(0,0,1.2,0))
UE2015$rate <- (UE2015$SYR / UE2015$pop)*10000
var <- UE2015$rate 
breaks <- getBreaks(v = var, nclass = 6, method = "quantile")
cols <- carto.pal(pal1 = "blue.pal", n1 = 3, pal2 = "red.pal", n2 = 3)
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
choroLayer(x = UE2015, var = "rate", breaks = breaks, col = cols, border = "grey40",
           add = TRUE, legend.pos = "topright", legend.title.txt = "Nb of syriens\nper 10 000 inh.",
           legend.values.rnd = 2)
labelLayer(x = UE2015[UE2015$rate > 7,], txt="name", halo=T,cex=1)
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth, World Bank, United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

What about sizes?

What would happen if the size of the circles were changed?

Small

With small circles, the map suggests that the number of Syrian migrants arriving in Europe is very low.

par(mar = c(0,0,1.2,0))
size <- 0.1
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", inches=size, col="#91591D",legend.frame = T, legend.pos = "topright", legend.title.txt = "Number of persons")
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

Medium

par(mar = c(0,0,1.2,0))
size <- 0.5
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", inches=size, col="#91591D",legend.frame = T, legend.pos = "topright", legend.title.txt = "Number of persons")
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

Big

With large circles, the map suggests that the number of Syrian migrants arriving in Europe is very substantial and can become problematic.

par(mar = c(0,0,1.2,0))
size <- 1
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", inches=size, col="#91591D",legend.frame = T, legend.pos = "topright", legend.title.txt = "Number of persons")
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

What about colors?

“Mere color, unspoiled by meaning, and unallied with definite form, can speak to the soul in a thousand different ways.” (Oscar Wilde). Actually, colors in cartography can suggest many things…

Peacefullness

The color green is the symbol of balance and happiness. It is a colour that builds confidence.

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#9dbadb", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", inches=0.5, col="#bbe5b7",legend.frame = T, legend.pos = "topright")
layoutLayer(title="PEACEFULNESS", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

Warning

The red is the color of all controversies. It evokes danger and aggressiveness, but also love and passion. Red never leaves you indifferent.

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="black", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", inches=0.5, col="red",legend.frame = T, legend.pos = "topright")
layoutLayer(title="WARNING", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

What about words?

Sur une carte, les mots sont très important. Changer un seul mot peu changer complètement le point de vue, l’intention de la carte.

Welcome

Here, the words “refugees”, “exiles” and “welcome” turn the map into a map in favour of welcoming Syrian migrants.

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", col="#91591D",legend.frame = T, inches=0.5, legend.pos = "topright", legend.title.txt = "Number of Syrian exiles welcomed\nin European countries")
layoutLayer(title="REFUGEES WELCOME", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

Not Welcome

Here, the words “invasion”, “illegal” turn the map into a map against the reception of Syrian migrants.

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", col="#91591D",legend.frame = T, inches=0.5, legend.pos = "topright",legend.title.txt = "Number of Syrian illegal immigrants")
layoutLayer(title="THE BIG INVASION", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

If we combine color, size & words…

If we combine colors, size and words, we can design two completely different maps with contrasting messages. It should be recalled that these two maps strictly follow the rules of graphic semiology and have been made from the same data.

:-)

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="#9dbadb", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", col="#bbe5b7",legend.frame = T, inches=0.1, legend.pos = "topright", legend.title.txt = "Number of Syrian exiles welcomed\nin European countries", legend.style="e")
layoutLayer(title="REFUGEES WELCOME", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

:-(

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border =NA, lwd = 0.5, xlim = c(extent[1], extent[3]), ylim = c(extent[2],extent[4]))
plot(UE2015$geometry, col="black", add=T, border="white")
propSymbolsLayer(x = UE2015, var = "SYR", col="red",legend.frame = T, inches=1, legend.pos = "topright",legend.title.txt = "Number of Syrian illegal immigrants")
layoutLayer(title="THE BIG INVASION", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

What about geographical extent?

So far, we have considered Syrian migration only to the countries of the European Union. But was that relevant? Let’s change the geographical scope…

extent <- c(2800000, 300000, 8200000, 4800000)
par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border ="white", 
     lwd = 0.5, xlim = extent[c(1,3)], ylim = extent[c(2,4)])

plot(UE2015$geometry, col="#f7ceb2", add=T, border=NA)

plot(countries$geometry[countries$name=="Syrian Arab Republic"], col="#91591D", add=T)
text("EUROPE",x = 4156169, y=2845867, cex = 1.5)
text("SYRIA",x = 6896775, y=1333590, cex = 1.5)

layoutLayer(title="Syria & Europe", sources = "Natural Earth, 2018", author = "Lambert & Bahoken, 2018",scale=500)

… and reproduce the map.

migrants <- countries[countries$ISO3 != "SYR",]
fixmax <- max(migrants$SYR, na.rm=T)
par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border ="white", 
     lwd = 0.5, xlim = extent[c(1,3)], ylim = extent[c(2,4)])
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")

propSymbolsLayer(x = migrants, var = "SYR",
                 inches=1.3, col="#91591D", 
                 legend.pos = NA, fixmax=fixmax)

labelLayer(x = migrants[migrants$SYR > 100000,], txt="name", halo=T,cex=1.3)

layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

The map speaks for itself. The main receiving countries for Syrian migrants are not in Europe but in neighboring countries. If there is a migration crisis, it is therefore not in Europe. And the map is even more meaningful if you add 7.6 million internally displaced persons in Syria.

interne <- 7600000
countries$SYR[countries$ISO3=="SYR"] <- interne
deplaces <- countries[countries$ISO3 %in% c("SYR","CAN"),]

par(mar = c(0,0,1.2,0))
plot(st_geometry(countries), col = "#e5e1d5", border ="white", 
     lwd = 0.5, xlim = extent[c(1,3)], ylim = extent[c(2,4)])
plot(UE2015$geometry, col="#f7ceb2", add=T, border="white")
propSymbolsLayer(x = deplaces, var = "SYR",
                 inches=1.3, col="#fc951650", border="#fc951650", lwd=0.1,
                 legend.pos = NA, fixmax=fixmax)

propSymbolsLayer(x = migrants, var = "SYR",
                 inches=1.3, col="#91591D", 
                 legend.pos = NA, fixmax=fixmax)

labelLayer(x = migrants[migrants$SYR > 100000,], txt="name", halo=T,cex=1.3)
text("+7.6 million internally displaced persons in Syria",x = 4843332, y = -100931.1, cex = 1, pos = 1, col = "#8e4708")
layoutLayer(title="Syrians refugees in Europe, 2015", sources = "Natural Earth & United Nations, 2018", author = "Lambert & Bahoken, 2018",scale=500)

In summary, be suspicious of maps, not of migrants.

Françoise BAHOKEN & Nicolas LAMBERT

Dec. 12, 2018