I heard you want to write one of these “x is the best country to live in” articles. With a ranking of the best countries to which we should all move to. You know what’s perfect for that? An index is perfect for that! Let’s build one ourselves, to understand what they are:
First, we’ll define “good life”. In social sciences, indices are defined as composite statistics. So for each country, we can throw multiple variables together in a pot: Maybe “a good life” means to be healthy, to have enough money to spend and to have a good education? Then we would choose numbers for life expectancy, GNI per capita and the mean years of schooling. These are the variables the Human Development Index chose.
Second, we’ll stir our variables together. For each
pot country, we do some number magic (aka aggregating) to make one number out of three. And voila, we have an index. One single number for each country, that we can compare easily with other countries.
If we feel like it, we can add a third step and rank the countries according to the final score. That’s what I did for the three indices and the results from the one survey in this table:
We will find lots of “9999” entries in the table: They indicate countries for which specific indices were not calculated. We can see that none of the indices exist for every single country. Meaning, most rankings can’t tell us that “x is the best country to live in” (sorry), but only that country x is better than all the countries for which indices were calculated.
You may have noticed that this week’s Weekly Chart is a table. (In fact, it’s the first Datawrapper long table that we show in a Weekly Chart.) Why did I choose this chart type? It’s a combination of reasons:
- Tables can handle a lot of information, and that’s what we have: Four variables, for almost all countries on this planet. And yes, other chart types like the scatterplot are great for showing tons of data as well. But there’s more:
- Tables allow readers to sort number – and in our table, sorting is essential to compare how different indices rank countries differently.
- If our tables have a search option, readers can quickly find themselves in the data; comparing the variables for their company, their city, or – in our case – their country.
- And last but not least: A table shows missing values well (the many “9999”). If a value is missing in a bubbly scatterplot, we would probably not display the bubble at all. Here, the missing data makes an important point. It’s part of the statement I want to make about indices.
Massive tables can be a great choice for our data. They don’t communicate insights but can work well if we have lots of data that readers are supposed to dig into themselves and quickly find the data of the country they live in.
Datawrapper offers two tables: The long table and the short one. Are you confused about when to use which one? I wrote an Academy article that might help you with this decision. The summary: Use short tables if you want to make a point with the data; use a long table if you want readers to explore the data themselves. I’ll see you next week!
Here are two more of the most obvious problems with indices:
1. Indices are biased and arbitrary. To the problems of measuring all initial indicators in each country, we’ll add our personal bias when it comes to deciding which ones to select and how to weight them.
2. Indices are intransparent. Their name doesn’t tell readers what they entail (“Life expectancy in years” is a clearer-sounding measure than a “Where-to-be-born Index”) or how the indicators are weighted. ↩