I don’t normally get my blog involved in controversial subjects such as politics or religion, but sometimes I feel the need to break this rule in the name of science. So here it is: in the great cream tea debate, which county applies the clotted cream and jam in the most popular order? Is it Cornwall with their jam first, then clotted cream on top? Or is it Devon, with the cream first and then a dollop of jam?
For readers outside the UK, clotted cream is an ultra thick dairy cream produced by heating normal cream in a water bath, then leaving it to cool and finally skimming off the clots that form as it cools. Residents of the Southwest of England felt that merely inducing heart disease by mounding this amazing product on a scone was insufficient, and therefore added a generous portion of strawberry jam (US: jelly) to the whole ensemble in order to make sure you get to dice with diabetes at the same time. Plus there’s a pot of tea because there is never any occasion that doesn’t require a pot of tea.
For a truly scientific study we would have required a large sample (a subset of individuals) from within the cream tea eating population of the UK which would allow us to draw conclusions about the likely opinion of the population as a whole.
In order to determine a useful sample size we would need to know:
- Population size – what is the total size of the cream tea eating population of the UK? This is tricky. We could start by taking the total UK population and removing the population of babies under a year old, lactose intolerant people gluten intolerant people, vegans, people with medical conditions precluding them from eating fat or sugar, people who are on a diet, people who inexplicably just don’t like cream teas. This all sounds like too much hard work so we’re going to assume either a very large population size or one which is unknown.
- Margin or error (confidence interval) – no sample is perfect, but what level of error do we feel is acceptable in our conclusion, perhaps a margin of error of +/-5%
- Confidence level – are we going to want to be 95% confident in our result (fine if we’re talking about cream tea preferences) or 99% confident (you’re going to want to be this confident in your results if your question has a life or death outcome, like a medicine trial). For a 95% confidence we would get a Z score of 1.96 to plug into our sample size calculation.
- Standard deviation – how much variance do you expect in your sample, 5 is usual.
Now we can put that all together to get: Necessary Sample Size = (Z-score)2 * StdDev*(1-StdDev) / (margin of error)2
((1.96)2 x .5(.5)) / (.05)2
(3.8416 x .25) / .0025
.9604 / .0025
385 respondents are needed
We didn’t have 385 cream teas or volunteers to sample them. We instead applied a technique which would be more likely to be
found in a tabloid newspaper than in a scientific journal, and had a woefully small sample made up of just four extremely biased individuals, all of whom have been raised in the Devon method of Cream Tea construction.
From our sample it was determined that the UK prefers the Devon technique of cream first and then jam, because although the flavour is identical, the cream first technique is more like the familiar ‘butter first’ method used for making toast or sandwiches, and therefore feels more correct.
The next job is to determine who is saying ‘scones’ correctly in our shamefully divided household where we both aim to raise the kids in what we each believe to be the correct pronunciation. Is it Matt who rhymes scone with bone, or me with my scone rhymes with gone? I feel my argument that the joke doesn’t work if you say scone like bone wins hands down. Here’s the joke, you can make up your own minds:
What’s the fastest cake?
(said like ‘it’s gone’)
Comments and opinions will be welcomed 🙂