The story so far: The Australia Institute released a briefing note that contains some very interesting graphs. In the first instance it shows that Switzerland has a company tax rate of 8.5%. Then they appear to have arbitrarily excluded data from graphs while showing completely different data series to what they describe in the paper.
Today we look at their next graph.
Again please note the axes aren’t labelled and the y-axis has a maximum value of 60000.
The Australia Institute describes this graph as follows:
Figure 3 shows a strong relationship between living standards and the education attainment of the labour force.
Before we proceed to show what they do to fudge their data, let me show you what that graph looks like using data that the Australia Institute claim to have employed.
Yes, well. Again it looks like they have excluded everyone with a per capita GDP of more than $60,000. To be fair, they did previously drop out Luxembourg but again have given absolutely no reason for dropping any of the other economies. But wait … there is more.
When preparing the data it was apparent that there were a number of former Eastern Bloc countries; Estonia, Hungary, Poland, Slovak Republic, and Slovenia, that had high educational attainment levels but relatively low living standards. Figure 4 was presented to show the impact of excluding those countries.
Figure 4 shows that by excluding the former Eastern Bloc countries from the OECD data the relationship between education attainment and living standards is even stronger.
That is a game everyone can play – so what happens when you throw out those economies with relatively low educational attainment and relatively low GDP per capita? (and also Luxembourg – it doesn’t matter if you include or exclude it, the result is much the same, but the graph looks better if you exclude it).
Looks like the very opposite relationship to what the Australia Institute reported.
Tomorrow – why this flawed research is so interesting. (Some people have pointed out that there are interpretation issues at stake here too. Yes, I agree. For now I want to focus just on the actual data and graphs.)