Combining visualisations and text – What is possible?

As I mentioned yesterday I created OurWorldInData somewhat accidentally when I was doing research for a book on global development that I planned to write. I will still want to write this book. It will visually present the empirical evidence that I am showing on OurWorldInData but I want to add my narrative to this. The website is presenting the facts about the world the book is showing how I put these views together and what we can learn from this.
To achieve this I have to find a way of combing data visualisations with text. I have collected some example to learn from how others have done this. I came across some great work and I hope it can be an inspiration for others to create their own work. Also, I feel that there must be more out there. What am I missing?

Academic Journals – Columns to combine large and small visualizations

Journals often put a huge emphasis on visualizations. In fact publications in Science or Nature are often completely structured around a couple of very good visualizations that present the main finding.

The layout of these journals is mostly structured by 2 or even 3 columns. One advantage of this is that wide visualizations can be combined with smaller visualizations. The screenshot from a recent paper that shows that the global forest area is now increasing shows an example.

Screen Shot 2015-06-01 at 22.18.17
A screenshot of a paper in Nature Climate Change.

This is also this LaTeX template that creates this layout for you.

Tufte Style

Edward Tufte – widely respected as an authority in data visualisations – uses a similar layout in his books. He combines one wide main text column with visualizations that are either as wide as the entire page, or wide as the text column, or are small and fit next to the text. This also allows a combination of different sizes but I’m not convinced there is an advantage over the layout in academic journals.

Tufte photo0
An example of the layout used by Tufte. From: Edward R. Tufte – The Visual Display of Quantitative Information (2001)

There is also a LaTeX template for this and the physicist David McKay uses it for his absolutely fantastic (and free!) book on sustainable energy. So you can see for yourself if this layout convinces you.

Merging text and visuals

Some authors take visuals and make them part of the text flow.

A first example is Oliver Byrne’s classic edition of Euclid (freely available here).

euc-I-5A modern example are Randall Munroe’s cartoons.

Screen Shot 2015-06-01 at 22.47.02

Overlaying visuals with text

This style that is often used in infographics merges text and visuals by using the visual as a canvas on which text is highlighting and explaining some aspects.

thumb_IMG_3931_1024This example is from Sandra Rendgen’s very inspiring Understanding the World: The Atlas of Infographics.

Mixing text with interactivity

I’m pretty tired today because I spent way too much time with “Explorable Explanations” yesterday night..
These online documents allow you to manipulate visualisations which in turn manipulate the text. You have to go and see these if you haven’t. A great example is Brent Victor’s Ten Brighter Ideas, a text where every claim expressed in words is backed up by data that becomes visible on mouseover. He writes: “Digital documents aren’t subject to the constraints of paper. We should hold modern propaganda to a higher standard. By all means, be catchy, eloquent, passionate, and inspiring. But we must be able to dive through the pretty words to see the data and sources beneath.”

He also wrote “Explorable Explanations” with which he coined the term for this genre.

And one more by him: Scientific Communication As Sequential Art where he rewrote a paper on network theory published in Nature made it into an explorable explanation.
A list of many explorable explanations can be found at
Here is an explanation of OLS regression and here is an explorable explanation of how small differences in preferences lead to segregation.


Which beautiful and useful ways of combining text and visuals did you come across? Share it with a comment here.

Short history of OurWorldInData – and where to go from now

Short history: Why I built

About 4 years ago I started working on a book in which I wanted to take a long run perspective and show how the world is changing. There is a need for this since the journalism that informs most of us about how the world is changing is way too focused on current events. Journalists are reporting news – events that happen now – and the long trends that slowly but steadily shape our world are left out of the picture. ‘Industrial revolution happening right now‘ never was a headline.

A consequence of the focus on current events is that many have a very negative picture of how the world is changing – a negative event often happen in an instant (terrorist attack, industrial accident, or natural catastrophe) while most of the very positive events only move very slowly (child mortality or violence are declining over decades or centuries). The idea for the book was to show these long-term trends. I believe we should know these trends and understand what drives them so that we can learn from them and continue to build a better world.

To prepare the book project I started collecting empirical data and visualizations of data on everything that matters for our living conditions. I ended up with a lot of data: Now, after 4 years of data collection this is a database on more than 600 topics (air pollution, terrorism, dental health) and it includes probably around 10,000 visualizations on these topics.

Initially this data collection was just my preparation to write a good book. But when I started working in Oxford it was Tony Atkinson‘s idea that it would be useful to present the empirical evidence I collected freely available for everyone on the web. We came up with the plan to visualize the data and put these visualizations in context – why do these changes matter for us?, what are the limitations of the data?, what is driving these long-term changes? how are the trends interlinked? This is how I ended up building

For a long time it was my side project that I did in addition to my research on economic inequality. This changed when I told David Hendry about this. He was enthusiastic about it from the start and supported my work on this website. At his research program (EMoD) he gave me the freedom to split my time between research on inequality and the construction of the website. David was immensely helpful: He is one of the world’s top experts on time series econometrics and therefore the very best expert to have on a project that is all about a long run perspective on how things are changing. And he is doing empirical research on a huge range of topics (from the link between productivity and wages (VoxEU here) to research on climate change (together with Felix Pretis) so that he is informed about the many different topics that are covered at OurWorldInData.

It was also together with David that I applied for external funding for OurWorldInData. We applied at the Nuffield Foundation – an institution that is very successfully funding empirical research that is focusing on well-being. The Nuffield Foundation has a great track record of supporting research that matters for the public by getting the work done in research institutions out to a larger public.

Where are we now

The London based Nuffield Foundation gave us a grant to expand OurWorldInData over the course of this year and I’m very thankful for this opportunity. It allows me to devote more time on this project and crucially it means that I’m not doing the work alone anymore: now has a team.

The new team

A month ago I started working with two very talented new colleagues. Lindsay Lee, who is helping to expand the content of the website and Zdenek Hynek, who is building a new framework for storing and presenting the data.

Lindsay is currently doing a MSc in Applied Statistics at Oxford and from September onwards she will continue to do a Master of Public Policy here. She just started working for OurWorldInData – in addition to the preparation for her statistics exams (!) – and she is already doing fantastically useful work for the project. I think she is such great help because she combines very strong quantitative skills  (she is a statistician) with a very good understanding of what the data tell us and why the data matter (she is also a public policy student). Lindsay will continue to work on the project and I’m very much looking forward to continue to work with her. We are planning to add a lot more content together – particular on the empirical evidence of how health is changing around the world.

And did you see the great example of how to combine visuals and text that she chose for the background of her website? She really is a great fit for the project.


Zdenek is a London based developer with very strong skills in the technologies that are used for visualizing data on the web. He knows his JavaScript, d3.js, php, SQL, – but don’t take my word for it, have a look yourself: Here are two past projects of Zdenek that I like a lot: visualizes the data on crime in the Czech Republic. It is a great example for the kind of job where visualization is useful: yes, the data has been available for some time, but it is only now after you build a useful tool that this data can actually tell you something.

ecotrust Canada is an example for Zdenek’s skills to design clean, useful websites. I also like the way that maps are used in this site to highlight the work that the ecotrust is doing.

One of biggest problems of the current technical framework of OurWorldInData is that each visualization has to be done individually: A spreadsheet with the data has to be prepared and then you have to write a page of html and javascript to visualize the data stored in this file. This means a lot of manual work to add one visualization and it is also very cumbersome to update figures with new data.

Zdenek is building a system that will eliminate this tedious, not-scalable process. He is essentially creating two tools: A first tool that allows us to upload data into one central SQL database (as opposed to the current system of individual unconnected csv files). And a second tool with which we can then pull any of the data from this database and visualize it in an interactive chart.

Next week we are planning to use this new system for the first time! Zdenek’s job is absolutely fundamental for the future of this web publication and it has been fantastic to work with him.


This is what is happening here in our office in Oxford – do you have additional ideas for our work? Are you missing anything on OurWorldInData? Do you have ideas or requests for our work? – Please tell us via the commentary section below!

Usage of OurWorldInData

Many more people than I thought access the website, here is a map that shows from which countries people access OurWorldInData:

Screen Shot 2015-06-01 at 00.00.52

Every country in the world with the exception of Eritrea!

And the media also makes more use of it than expected. Here I have a regularly updated list of the media coverage of OurWorldInData.

Inequality or Living Standards: Which Matters More?

For many years inequality in income and wealth received little attention in public debate and was only a minority interest in the economics profession. GDP per capita was widely considered to be a satisfactory indicator of economic prosperity. Yet, inequality has now become the focus of remarkably wide-ranging attention, from Davos and the State of the Union address to academic journals across a variety of disciplines. Thought-provoking research by Tony Atkinson and Thomas Piketty on an increased concentration at the very top of the income and wealth distribution over the last century has played a major role in moving inequality towards the centre of political and academic debate. As the chart below shows the rapid rise of top income inequality in the English speaking countries – and the more modest rise in continental Europe and Japan – from about 1980, in sharp contrast to the decline seen over the previous 40 years. These trends bring to the fore the question how inequality can be addressed, and a set of concrete proposals aimed at doing so is put forward in Tony Atkinson’s new book.



(All charts in this text are taken from the INET supported web publication OurWorldInData written by Max Roser. Our World In Data shows how living conditions around the world have changed over the long run.)

Insight into top income shares and summary measures of inequality such as the widely used Gini measure helps to recognise inequality as a central indicator of national economic progress. Nevertheless, these measures do not tell us about living standards for individuals at different parts of the distribution. The top income shares are pre-tax, and both the top shares and Gini measures are expressed in relative terms and therefore do not reflect actual consumption possibilities. In order to gain insight into income levels as well as its distribution we should look at the evolution of living standards at different points of the distribution. The most direct way to capture living standards is to focus on household income expressed in real terms (that is, taking changes in price levels into account to reflect purchasing power) and also adjusting for household size and composition to take into account that the standard of living attainable with a given level of household income depend on the number of persons in the household. Income has limitations as a measure of living standards, but is more satisfactory than alternatives such as consumption (which could be financed by depleting savings or running up debt) or subjective measures of well-being that are influenced by many other factors.

Increasing income inequality with and without rising living standards

Focusing entirely on income dispersion can blur differences in the evolution of living standards, as can be illustrated by looking at the contrasting experience of the USA, Australia and Greece over recent decades in terms of inequality – as captured by the Gini measure – and real disposable incomes for households across the income distribution. As the chart for the USA shows and has been widely discussed, rising inequality and increasing shares going to the very top have gone together with stagnation in real incomes for the rest of the distribution for much of the period from 1980. In fact, living standards for the middle and below households only increased substantially between the mid-nineties and 2000. For these households, their income levels in 2013 were lower than in 2000.

Income inequality and growth of living standards across the income distribution in the USA since 1974

While Australia has also seen substantial increases in income inequality as is evident from a rise in the Gini and top income shares, the evolution of real incomes has been quite different from the American experience. There are significant increases for households in the middle and lower part of the distribution.

Income inequality and growth of living standards across the income distribution in Australia since 1981

Greece in the period up to the economic crisis saw rapidly rising living standards across the distribution. The Gini index has been decreasing or stayed flat even in the crisis. This can hardly be interpreted as good news for middle and below households, however, as can be seen from the calamitous collapse in living standards.

Income inequality and growth of living standards across the income distribution in Greece since 1974

Also looking at developments within other OECD countries over the past 30 years illustrates the perils of a sole focus on inequality. The UK, for example, has seen both periods of rapid real income growth for those in the middle and lower parts of the distribution and periods where their incomes have seen little or no increase. While this is associated with how GDP evolves, GDP per capita is an unreliable indicator of how these living standards have changed, even when seen together with top income shares or the Gini inequality measure. Thus, if real incomes and living standards of middle and lower-income households are at the core of our concerns, reliable direct measures of how they are changing over time and how policy is likely to impact on them need to be front and centre.

It is an important step forward that inequality is increasingly recognized as a central concern not just from the point of view of fairness but as representing a real threat to economic performance and social cohesion. To get the whole story, though, it has to be looked at not in isolation but together with the way living standards evolve. It is this combination that is of such central importance to people’s lives of people and needs to be at the core of research, debate and policy-making.

Is income inequality rising or falling?

How are the benefits of economic growth shared across society? Much of the current discussion assumes that income inequality is rising, painting a gloomy picture of the rich getting richer while the rest of the world lags further and further behind. But is it really all bad news?

The reality is complex, but by looking at recent empirical data we can get a comprehensive picture of what is happening to the rich and poor.


Let’s start with the share of total income going to that much-maligned one per cent. Reconstructed from income tax records, this measure gives us the advantage of more than a century of data from which to observe changes.

The blue line in the left hand panel below shows the long-term trend in the US. Prior to the Second World War, up to 18% of the all income received by Americans went to the richest 1%. Following the war, the share of the top 1% dropped substantially, increasing again in the early 1980s until it returned to its pre-war level. This U-shaped long-term trend for top income share is not unique to the US; several other English-speaking countries shown in the left hand panel followed the same pattern. After a decline in the past, inequality is now on the rise.

But it’s not a universal phenomenon. The right hand panel shows that in a number of equally rich European countries, and in Japan, things developed quite differently. Just like in the countries on the left, the income share of the rich reached a low point in the 1970s, but then, rather than bouncing back up to previous levels, it remained flat or increased only modestly, giving us an L-shape on the graph. Income inequality has decreased drastically since the beginning of the 20th century so that today a much smaller share of total incomes is paid to the very rich.

One lesson to take away from this empirical research is that there is reason to believe that we can do something about inequality. If there was a universal trend towards more inequality it would be in line with the notion that inequality is determined by global market forces and technological progress, where it is very hard (or for other reasons undesirable) to change the forces that lead to higher inequality. It is dangerous to believe that there is a unanimous trend to higher inequality, as this encourages the belief that growing inequality is inevitable.

The reality of different trends suggests that it is not global forces that shape the distribution of incomes but the country-specific institutional and political framework. Therefore it is crucial to understand the institutional settings that allowed some countries to achieve economic growth without returning to the old levels of top income inequality. A major step in this direction is the forthcoming book from the inequality researcher Sir Tony Atkinson in which he makes concrete proposals on how to reduce inequality, based on the insights from periods in which inequality decreased.


The data on top income share do not measure the share of income that reaches the pockets of the rich, but rather the gross income before taxes are paid. Yes, the rich do avoid paying taxes, and top marginal income tax rates were higher in the past, but progressive taxation still does a great deal to narrow the gap between rich and poor: in the US, 37% of the total sum of income tax is paid by the top 1%, while less than 3% is paid by the bottom 50%. The redistribution means that the incomes of the poor are higher after taxes (because of transfer payments such as pensions, child benefits, and unemployment benefits) and the incomes of the rich are reduced after taxes (due to generally progressive income tax rates).

This difference between pre-redistribution ‘market incomes’ and eventual disposable income is shown in the next chart. Redistribution through taxes and transfers reduces inequality considerably.

In this chart inequality is measured with the Gini index, an inequality measure that not only looks at the top of the income distribution but captures the whole distribution.
We can see that market income inequality in the UK, the US, and France is fairly similar (Gini between .5 and .52) – but there are big differences in how much these countries reduce inequality by redistribution. Inequality in both market and disposable income are steadily increasing in the US, and compared to similarly rich countries the US redistributes comparatively less.

These data are based on household surveys, a shortcoming of which is that they under-report top incomes. Likewise, the shortcoming of the top income measure is that it is necessarily silent about what is happening within the bottom 99% of the distribution. Taking the ‘top 1%’ chart and the ‘market income vs disposable income’ charts together allows us to understand how inequality has developed. In the UK, the pre-tax share of the top 1% has been rising continuously since the late 1970s, but disposable income for all earners followed a very different trajectory, with inequality increasing rapidly in the 1980s but not changing much since then. If anything, income inequality has actually fallen in the UK over the last 25 years. In summary, the incomes of the poor in the UK are growing as fast as the incomes of the rich, apart from the top 1%, whose incomes are racing away.


Before we turn to the global income distribution I want to shift the focus from Europe and North America down to South America. In all of the South American countries shown in the following chart, income inequality has fallen since the year 2000. It is shown in the chart for the bigger South American countries (and it can be studied for the other countries on here). Rapidly falling inequality in South America that lifts millions out of poverty is a huge success, and demonstrates once again that there is not one simple answer to whether inequality is rising or falling within individual economies.


We’ve looked at the changing income distribution within countries but what about the global picture? The chart below shows income estimates for the world in the year 1820, adjusted, as always, to take account of inflation and price level differences between countries. Two centuries ago no country in the world had a life expectancy over 40 years, and even in relatively rich countries like England and France the poor were so malnourished and weak that they were effectively excluded from the labour force. The share of the world population living in absolute poverty was estimated to be around 90%.

Over the next 150 years some countries achieved economic growth while others remained poor. While Europe and the European offshoots in North America and Oceania grew rapidly, most of Asia, most of Latin America and all of Africa remained poor. The consequence of this was a hugely unequal world. The bimodal (“two-humped”) blue line for 1970 shows the world income distribution of a planet clearly divided into rich and poor countries.

Since the 1970s the world has changed. The circle of countries achieving economic growth includes much of Asia, Latin America, and for the last two decades also includes Africa. The consequence of this is that global poverty is falling faster than ever before – the share of the world population living in poverty has decreased from over 50% in 1981 to 17% in 2011.

Because of rapid growth in formerly poor countries, the world income distribution has changed dramatically. The bimodal distribution of very high inequality across countries in 1970 has changed into a unimodal distribution of lower inequality today. The latest research on this question shows that the Gini index for global inequality has fallen from 72.2 in 1988 to 70.5 in 2008 (the last year for which we have data).

While there is no reason for complacency, it’s a fact that global poverty is falling faster than ever before, with the share of people living in poverty decreasing from more than 50% in 1981 to 17% in 2011. There is still a long way to go to improve living standards for the world’s worst-off, but we can take a clear and heartening message from the data: global income inequality is falling and global poverty is in decline.


If you are interested in long-term trends of living standards around the world follow me on Twitter where I share many data visualisations of long-term trends from my web publication Our World In Data.