Part II Collecting and analysing data To compare government spending on health, you can divide total spending by the population. This is called normalization. Now you can compare government spending on health per capita. Another way of normalizing values is to use percentages. For example, in determining the maximum available resources for fulfilling ESCR, you might want to compare tax revenues between countries. One way to do this is to normalize tax revenue as a percentage of both countries’ GDP (e.g. to say that the first country collects a relatively high amount of revenue, whereas the second country does not). HELPFUL TIP If you end up working with a very large dataset, there may come a point when you “outgrow” your spreadsheet. If this happens, consider using other database software such as SPSS, Stata or R. Unlike a spreadsheet, which is designed to be able to be “read” on the screen, the way data is stored in a database is often completely hidden from the user. This enables abstract, complex ways to store larger amounts of data and gives the user more flexibility in how to use it. That said, databases are a much more technical way to store and analyse data. As such, you may need to work with experts like statisticians, programmers and designers. 8.4. INTERPRETATION AND HOW TO AVOID COMMON MISCONCEPTIONS While simplification is required to understand what the data means, when you are simplifying and presenting graphical evidence it is crucial to stay as close as possible to the “full story”. 8.4.1. Correlation is not causation In general, it is extremely difficult to establish causality between two correlated observations. There are several reasons why common-sense conclusions about cause and effect may be wrong. For example, you might want to report on the relationship between education and health. Using a scatter plot, you have average years of schooling on one axis and average life expectancy on the other. The scatter plot reveals that the two variables are highly positively correlated: higher average years of schooling are associated with higher average life expectancy. But can it be said that higher average years of schooling causes higher average life expectancy? No. There could be a number of reasons for this association: • Higher average years of schooling may cause higher average life expectancy. • Higher average life expectancy may cause higher average years of schooling. • Higher average years of schooling and higher average life expectancy are consequences of a common cause, but do not cause each other. The common cause may be higher average income, for instance, or a more equitable distribution of income. • There is no connection between average years of schooling and life expectancy. The correlation is coincidental. Chapter 8: Analysing data: A short introduction to working with spreadsheets | 93