A Global Model:
Which countries are faring well at keeping kids alive based on their formal healthcare system resources?
To create Orb Media’s global healthcare ranking, our chief data scientist Heather Krause developed a unique model for comparing “expected” child mortality with “actual” child mortality. To develop this model, she identified and analyzed several key variables.
The first two things she needed were the child mortality rate and healthcare worker rate of each country. By “child mortality rate,” she means the estimated number of children who die before turning five years old, per 1,000 live births. Healthcare worker rates were calculated in much the same way—here, she is talking about the number of formally trained doctors and nurses per 1,000 people. She also needed to know how much each country spent on healthcare each year (whether in the form of government programs, private schemes, or a combination of both).
Once she had that information for 160 countries, from Afghanistan to Zimbabwe, she was ready to dive into a more complex investigation. She examined data from 2010-2019, using the combination of financial expenditures, the number of formally trained staff, and other factors to develop the rate for “expected child mortality.”
She then compared this expected rate against each country’s actual child mortality rate—which left us with a raw difference of approximately 5.5 percent between the two. However, she also needed to calculate the proportional difference between the two figures to be able to compare countries against each other more accurately
With a list of both raw and proportional differences, she was ready to create our tiers. Our team settled on a five-tier system, with “1” designating the top performers and “5” designating those who performed below expectations. These tiers can be viewed here. For details of the data and help customizing it to your own story, contact us.
We’re excited to support your work. In return, we ask that you:
2. Credit Orb Media’s work (or other graphic sources) including our original data analysis and key findings as appropriate.
3. Share your expected publishing date and link (or PDF if appears in print only) with Orb so we can aggregate, promote and learn from original reporting worldwide. In the future, we’ll pass our algorithmic and framing learnings from the story’s collective performance on to you.
4. This package was published in concert with other media organizations during the week of August 10. However, its evergreen elements remain available for your future use.