Immunisation 2/5 Correlation and Causation
This is the second of five posts designed as a teaching mini-scheme about the controversy surrounding the MMR vaccination; it is partly inspired by the recently published work by Brian Deer. Please note I feel, quite strongly, that MMR is safe and highly desirable (albeit underused in the UK right now). This is my effort to provide colleagues with the tools (and printable resources etc) to provide good information. I suppose you could see it as immunising them against bad science such as that recently published in the Sunday Express. At the end of the five posts I will put the ideas and resources together, informed by any and all comments and feedback, into one downloadable scheme.
2 Correlation and Causation
Ask students to explain ‘links’ between things that happen at the same time:
- boys growing facial hair and starting to produce sperm
- height and shoe size
- final exams and sunny weather (in the UK, more or less)
- ice cream sales and deaths by drowning at the beach
- star sign and academic success
Alternatively use the card sort and ask them to find the pairs of linked variables. You could perhaps add in one or two pairs that are causatively linked.
Printable: correlation cardsort as pdf.
Start by defining correlation and causation. It might be worth going back to tricky examples from the starter, pointing out when two variables are controlled by a third. (Flow charts show this nicely). Explain why observational studies can’t easily distinguish between correlation and causation, but that a link is often interesting and may help to suggest a hypothesis. Examples such as the British Doctors Study (more detail at the MRC) might be useful. Individual facts about this study could be used to produce a ‘murder mystery’ activity.
Recap definitions of variables (independant, dependant and control) and elicit ideal experimental design. Discuss limitations, especially with health-related research (compliance, ethics, comparison to placebo etc). We try to look for a pattern once other factors are excluded, then investigate links without endangering patients. Cue debate about value/ethics of animal trials!
A few weeks back Matt Parker produced a fantastic bad science/stats trap – explained here in the Guardian – suggesting a link between mobile phone masts and fertility. It is of course simple – we put mobile phone masts where there is demand. Most students should be able to suggest that this is because that’s where the people are. I suspect the correlation is enhanced by the fact that demand is linked even more closely to the section of the population aged 15-45, who are both more likely to have mobiles and have babies. He’s archived the data as an Excel file, available through his website StandupMaths. Students could plot (some of) this data and then try to explain the pattern.
Give students a simplified version of the original claims about MMR, according to Wakefield and authors. Challenge students to suggest (1)possible other factors/explanations (2)best ways to investigate this possible link. What would they have done if they had been involved in medical science at the time? What would they have recommended to parents/GPs? Instead of a simple ‘yes’ or ‘no’, it can be useful to have students show their feelings on an ‘opinion line’. This allows more subtlety, for example by letting them position themselves between ‘Cancel all vaccines’ and ‘Don’t even bother researching a possible link’ – both unhelpful extremes.
Why is this funny?
Above cartoon is of course from the wonderful xkcd.
Some of these ideas and activities are introduced in immunisation2 saved as powerpoint.
As before, please let me know of any ideas, suggestions, improvements – through the comments here, by email or via Twitter.
Filed under: badscience, biology, printables, science, students, teaching | Leave a Comment
Tags: Brian Deer, Correlation and dependence, Matt Parker