Sense about Science ? equipping people to make sense of science and evidence
The Making Sense of Statistics guide is not a lesson in statistics. It provides the questions to ask and identifies the pitfalls to avoid, to help us get behind news stories that use statistics - stories like “Diabetes drug raises death risk by 60pc”, “Gender pay gap still as high as 50%” and “Polls puts Tories up to 7% ahead”. Dismissing all statistics as just ‘lies’ does not help us get to grips with a story. By working through the points in the guide we can find out what they really mean. The four main points to look out for:
1. Remember, when statistics are quoted they are just the answer to the particular question that was asked. The first step to understanding the statistic is knowing what this question was and how it was asked. We can then ask where the results came from (for example, a survey, a trial, administrative data or a projection), how the samples were chosen and how the figures were analysed.
2. The results of studies are commonly captured in a single figure, but this figure will not represent everything that the study might have found. The common pitfalls to be aware of are: there is more than one type of average, extreme values might not be very likely, and big and small numbers are difficult to comprehend without the context. Most of us don’t use millions and billions in our daily lives so by dividing them by the number of items they relate to we can make large numbers more meaningful.
3. A mathematical association, even if statistically significant, is not a certainty that one thing is causing another. Finding out what the confidence interval is for a result can give us an idea of how sure we are of the conclusions we have drawn. Confidence intervals give the scale of potential uncertainties in counting, measuring and observing data.
4. There are many alarming newspapers headlines about risk, its increase or decrease in relation to a particular factor. To understand the importance of any increase or decrease we need to know both the absolute and relative change and how large the risk was to begin with.