Decolonizing Data

When implementing our DEI values, we should evaluate everything, including why we choose to include some data points and not others when we are highlighting social issues. In particular, if we are talking about the United States, we need to stop comparing ourselves mainly to European nations to the exclusion of a more global approach. Here is one example of how that might look.

In a regenerative economics course I participated in recently, this graphic was offered as a talking point about empowered participation through creating economic prosperity and improved outcomes. It was specifically used to illustrate points about economic inequity and health outcomes in the United States.

My first reaction at seeing this graph was something like “Yikes! How is the US that bad?”.

I don’t deny that we have a lot of issues to address in the US and most of it is urgent. However, I usually take data where the United States is only compared against smaller countries, and mainly smaller, European countries, with a grain of salt. It was the outliers that caught my attention, and also that the USA was presented as “the worst.”

Yes, it’s true that income equity and health outcomes in the US could be improved. But I also know there are nations that struggle even more than we do that weren’t even a part of the graphic. If we always rank ourselves as “last,” especially compared mainly to European nations, then we are never going to see that there are other countries, that are not even on the chart, that we can work with, and possibly all help ourselves together.

The things that stood out to me and that made me want to find out more about the data behind the graph were:

  • The three largest countries in the world population-wise are: China, India and the United States. Neither China nor India were included on this graph.
  • The most diverse countries in the world are mainly in Africa, none of which were included in this graph
  • By land mass, the five largest countries are Russia, Canada, China, US, and Brazil. Only the US and Canada were included
  • Of our North American neighbors, Mexico is left out
  • Japan is the only Asian country included, Israel was the only country in or near the middle east that was included and no South American countries were included

I think it’s a good question to ask why. What bias do the researchers that put this graphic together have? I love data, so I decided to find out more.

First, I found the source of the graphic. It was from a British nonprofit called “The Equality Trust.” Here is the original graphic created by The Equality Trust:

So, first of all, this graph was originally generated by a British nonprofit for the purposes of comparing Britain to other European countries. This makes sense because, at the time they published it, Britain was still part of the EU and, even now, is still geographically part of the broader European community. When I found the graph on the Equality Trust website, I noticed that the y-axis on the original graph had been re-labeled for the regenerative economics presentation. I also noticed that Singapore was an outlier that had gotten cut off from the original graph on the Equality Trust website. Keeping Singapore on the graph makes the US look not quite as bad as the rest of the pack, so I felt like we were getting somewhere if we were going to use this graph to talk about the US. Last, the graph from The Equality Trust cited a source and year, which was 2009. I don’t know when the y-axis label changed from “Infant Deaths per 1000 Live Births” to “Health Outcomes Index,” or how / why Singapore got cut off, why the source and date got cut off, but those little niggling feelings had become red flags for me.

While this graph may have made sense for the British nonprofit to use as a way to bolster its own mission in the United Kingdom, among other European nations, it seemed less and less appropriate to use it for any kind of comparison for the United States. In addition, the cited data source was nearly 15 years old. Given how much we have been through as a whole planet in just the past 4 years, I wanted to look at more current data.

In my search for more current data, the first thing I learned was that the United States’ Center for Disease Control (CDC) counts “infant” deaths from gestation to about 10.5 months. Most countries stop at about six weeks. Any graph that compares infant mortality in the US to any other country is going to be off. So, to then take that statistic and relabel it as a general “health outcomes index” is not acceptable in my opinion. I don’t think this was intentional, but I also felt it needed to be changed to a broader health outcome index, not just relabeled.

I set out to find a more general “health outcome index” and came across the Legatum Health Index, from Prosperity.com. The Legatum Health Index takes into account several health-related outcomes over the entire lifespan, not just infant mortality rates. I believe that this is the index that the economics course instructors thought that they were referring to originally.

On the income equity side, it turned out that during the past 15 years, researchers switched from reporting “income inequality” to reporting “income equality”. That way 100% would mean total income equality. The index I used to update the graph in 2024 was the “GINI” index from WorldEconomics.com, which had been inverted since 2009 in order to switch from “inequality” to “equality”. The numbers on the x-axis are percentage of income equality in the given nation. Not all nations reported for the same exact year, but they are all more recent than 2009. The range of years for the income equality data is 2021 – 2023.

I should also note that the data analysis tool I used was Google sheets which is notoriously bad for generating charts. Google Sheets absolutely refused to go from high to low on the x or y axis, so we are both switching from income inequity to income equity and going from a chart that mixed and matched “high to low” and “low to high” on its axes, to a chart that only uses low to high on both axes.

When I generated my own graph from more recent data, for the same countries in the original graphic, a few things came to light:

First, Japan was misplaced even on the original graph from The Equality Trust. Unless Japan has undergone major shifts from income equity in 2009 to income inequity in 2024 (which I don’t think happened), Japan should have been an outlier on the same side of the graph as Singapore, as it is on this graph.

The USA, while still an outlier when compared to mainly European nations, was less so when we use a health outcome metric that is more evenly reported among all countries, eg: Legatum Health Index vs. Infant Mortality only. The graph really flattens out when you use a metric that all countries report more consistently. This tells me that, while the US has plenty of work to do, it wasn’t as bad as the original graph made it out to be and there were enough other “outliers” close to us that we could look to for inspiration. Especially the outliers that have the same or less income equality — but have better health outcomes! This was really a revelation for me: we need to find out what countries like Japan and Singapore are doing to get better health outcomes with similar income equality levels.

Even with all of these differences in data and graphing, the original point that higher income equality is associated with better health outcomes still remained strong. I am not arguing that point. I am arguing for using a wider lens in order to get a truer picture of where the US stands globally, and finding better sources for potential first steps and potential partners.

Since I still felt that this graph was too Eurocentric to make any conclusions about the United States, let alone find any solutions or paths forward to do something about it, I decided to keep going. I also included data from other nations based on the criteria I mentioned at the beginning of this article: population, diversity, land mass, and geographic proximity.

I added:

  • South Africa and Nigeria because they are two of the most diverse countries in the world
  • Mexico, in order to have 100% representation of our North American continent with our two closest neighbors
  • Central and South American nations, plus Middle-Eastern nations in addition to Israel, and more nations from Asia
  • India and China for large population size
  • Russia for its land mass

None these things were represented in what is now looking like a very tight cluster of small, European nations from the original graphs. I took a wider lens to see what other countries we could look to for examples. It makes a difference when you have a larger land mass to deliver health services across and a bigger, more diverse population to serve.

Yes, we sill have work to do, but by choosing a more inclusive set of nations, it doesn’t look unsurmountable anymore. Constantly comparing ourselves to smaller countries with less diversity is probably not going to help or reveal good first steps.

While its good to know that humanity is actually capable of producing excellent outcomes it’s pretty overwhelming to start out with “how can we become more like Belgium?” when we simply can’t become that much smaller geographically or population-wise.

Finding practical solutions and good first steps will bear out more readily if we compare ourselves with, and align ourselves with, countries more like our own. Instead of constantly ranking ourselves poorly against smaller, less diverse countries, we need to give ourselves a realistic view of what is possible and what works for countries with similar levels of equality and inequality.

It’s also high time we stopped with the “first world” comparisons and start looking at the countries we are actually contemporary with. There are so many nations that we have things in common with that we could both learn from and help raise up, if we just start including them in the conversation, and including them on the graphs that we choose to represent “how the world is.”


For higher quality resolution graphs, please download the Decolonizing Data PDF:

References and Sources (visited March – June 2024):