Wednesday, 15 April 2015

The Cause-Effect Conundrum

We’re in a General Election campaign (in case you hadn’t noticed!). Which means it’s open season for silly statistics and dubious claims in support of one cause or another. The advice here is not to take these claims at face value. Can they be justified by cause and effect? “Correlation does not mean causality” is the often-quoted phrase.

A classic example: Accident & Emergency waiting times have increased over the last five years, since the Coalition came to power. But that in itself does prove that the present government has caused this increase. It’s possible, but a host of other factors may be responsible as well, or instead.

A wealth of spurious (and amusing) correlations can be found with a quick Internet search. For instance, the US murder rate almost exactly matches the trend line for MS Internet Explorer’s market share! Fortunately, no-one suggests the two are connected. More intriguing though is the match between national chocolate consumption and Nobel Prize-winners, illustrated below:
(Credit: New England Journal of Medicine)

Here there is a plausible – but arguable – connection. Chocolate has been shown to improve cognitive function not just in individuals but in wider groups. So the fact that Switzerland tops the table on both counts might be no coincidence.

Such theories – for that is exactly what they are – can be described through a Theory of Change (ToC). The diagram here illustrates a very basic ToC to explain how training can help unemployed people to find work. Not rocket science, but note that:
a)      It’s expressed in terms of trainees’ experience, and
b)      The job is not and end in itself, but a route to more personal outcomes

In this case the ToC is obvious; in the chocolate example, and indeed for A&E waiting times, it’s a lot more problematic. Proving a ToC – understanding what actually makes the difference – is not always easy, but getting people involved in discussion is a start. Is it plausible based on their knowledge and experience? Can it be tested? What other comparisons do we have? ToC’s feature in many evaluations, as a start point to understanding what causes the changes we see – or hope to see in future. The more we can understand and apply these cause-effect principles, the better.

PS: Sweden comes 2nd in the Nobel Prize-winners list, and Alfred Nobel was Swedish. Coincidence?

No comments:

Post a Comment