Mapped Data

Introduction

Recently I had a discussion with a colleague trying to display some data on a map. Inadvertently the conversation turned to the concept of displaying data on a map, and the apparent impact it has on the people seeing the data. We have both previously worked at a workplace that once felt like it had become as obsessed with data on maps as people are obsessed with AI these days. It was impossible to have a discussion about data visualisation without the topic of maps and displaying the data on a map. Thankfully that period didn’t last too long, but it lasted long enough to get me thinking about what it was that made data on maps so compelling. My discussion with my colleague reminded me of that, and I thought that might be an interesting tidbit for a hot-take.

Real maps

Throughout time we as a human species have become obsessed with mapping out the world around us as indicated by the existence of maps throughout time, and for a wide array of purposes. From sketched makes to explain a travel route. Land map that registers one’s holdings and aims to settle disputes. Sea maps aimed at charting currants and travel times, or warning of dangerous features. There are maps showing paths and roads and just about any feature in the real world. And that is one of the cornerstones of a map. It serves to represent something of the real world.

source: By Zde - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=69695100

Possibly the earliest human map from 25000BC

Yes, today we also have maps of non-real things such as org structures or digital pathways, but I would argue that these are map-like derivatives, compared with the origin of maps. Intended to fill a similar role within the abstract, as maps do in the real world. It is also worth mentioning that maps only are representations and models of reality, not true reality. However, they have become so ubiquitous for all human beings, that as we think of geographical distances, you are likely to imagine them on a map surface. These days that is likely to be a Google map or Apple map surface.

Source: By Chinese (image enhanced by contributor) - http://geog.hkbu.edu.hk/GEOG1150/Chinese/Catalog/am31_map1.htm, Public Domain, https://commons.wikimedia.org/w/index.php?curid=3723658

Source: By Pedro Reinel - Bayerische Staatsbibliothek, Munchen, Public Domain, https://commons.wikimedia.org/w/index.php?curid=4973845

It is further tempting to think that the craft of making a map in the antique was a marvellous and challenging task to do. As they had to map out the land while being in it. The age of satellites had mostly made this a solved problem, and we no longer had a need for cartographers. That is far from the truth. Today the field of cartography is still going strong. But I am no cartographer and make no claim of being able to do the field justice.

I would however like to take you on a tangent from our tangent to talk shortly about the remarkable challenge it is to map our not quite round (The world is surprisingly lumpy) earth unto a flat map. There is competition for who can make the best 2D representation of the world on a map, that gives the most accurate representation of both size, location and distances. It i a map that is truly fascinating to see, and I feel everyone needs to have seen it at least once.

Source: http://www.authagraph.com/projects/description/【作品解説】記事01/?lang=en

There is also a double spherical one, that supposedly is even better that can be found here: https://www.scientificamerican.com/article/the-most-accurate-flat-map-of-earth-yet/

Maps make data real

While this is an immensely interesting topic we are on quite a tangent of discussion. We can however begin to anticipate the power of compulsion that lies within maps. So let us leave the world of maps behind, and return to the topic at hand Data that has been layered on top of a map.

For a long time, I was used to seeing data presented in two-dimensional tables, containing columns and rows. If one wanted to draw a summary or aggregate information you could find a graph with a X, Y and sometimes even a Z axis. These would generally be labelled according to the information in the data that you wished to display. You might have grouped data together or displayed them next to each other for comparison. If you were stuck in particularly theoretical fields such as statistics or mathematics, you would be more or less used to seeing whole data sets represented as formulae of distribution coefficient/variables. These representations are perfectly adequate at providing the skilled reader with all the information that they want and need out of the data. And as long as the data that is used to populate these graphical Denis plays of maps with data on them. I am comfortable to say that there strictly speaking is no additional information in seeing the data on the map, than in the table.

But that is not how it feels, and how it works. And that is because there is more data. There is our internal model of the world. Supplying and enhancing our understanding of the data, by bringing it from the world of the abstract and disconnected realm of numbers, directly into our sense of reality. Putting a mental tag on the information. The act of seeing the data on the map somehow becomes more real, and more relatable. And this is a phenomenon that works with everyone regardless of numerological literacy because everyone has a mental map in their head.

Maps make data weak

So, data on maps are miraculous, should we always use maps to display data? Unfortunately no. There are several challenges when representing data this way. And it is frustratingly easy to get this wrong, leading to people making the wrong conclusions about the data. That probably is I suspect why I often see this display in dubious company when used to convince people of something that is not quite right. I don’t say they all are wrong, just that they are tricky to get right. Their visceral impact is one of the hardest aspects. As everyone carries with them their personal model of the world it is hard to know what association a person has to a geographic location, and that might unconsciously alter their interpretations of the data.

Due to the geographic boundaries, it becomes hard to make arbitrary comparisons between data, and thus be certain that claims made about the data are valid. While the Office for National Statistics (ONS) the UK national statistics office does a very good job in their mapping it still serves to point out an example. Imagine two cities on opposite sides of the UK. Let’s take Glasgow and Cardiff. How do they compare to each other in relative income? How are outliers displayed?

Source: https://www.ons.gov.uk/visualisations/dvc1370/

A somewhat worse example on the same topic from the BBC. For those not versed in UK geographics this picture shows aggregate data that spans an enormous difference in economic, industrial, and social factors, condenses to a single number, split into 12 regions. All of Scotland and the Hebrides are a single area. This illustrates that the BBC had a story to tell here. And the map is made to match.

Source: https://www.bbc.com/news/business-53628115

While colour graded maps make it possible to estimate trends and relative differences at an aggregate level. It is very hard to get absolute numbers or to verify any claims about absolute numbers. How would you verify any claims of population size, or relative size in this plot? Can you verify based on the map alone that the American population is 13 times larger than the Swedish? Or find the population of Bangladesh? Mind you, this map was made deliberately to make it confusing to illustrate just the point I am making now.

Source: http://visualizingrights.org/kit/charts/bubble-map.html

Closing thoughts

I have for some time now lambasted the use of maps and their challenges. And I might come back to the topic in more depth at another time. But, That is not to say that data maps are only bad, or only good. They are a tool like any other. However, it is important to think about the weaknesses of maps for displaying data. Especially keeping in mind the visceral impact that that has especially on non-technical readers.

So the short of it. Data on maps are compelling because they connect something intangible with something that is “real” and that everyone has a relationship with. That grants Data on maps enormous storytelling power. These powers need to be used with caution. And as a reader of data on maps, think carefully about what the story that the map wants to tell, and if that story is the whole story. If something is interesting, take the time to go looking for the data behind the map.