Solution to question 1.2
Strictly speaking the children were not a random sample of 'children' or even '5-6 year old children'. They were a systematic sample of children; all the 5-6 year old children in Wick had a 100% chance of being picked and all other children had a 0% chance of being picked. The population that was sampled was 5-6 year olds in Wick and so, strictly speaking, the results can only be generalised to that population.
In practice with epidemiological studies we sometimes have to stray from purely random selection of subjects. We then need to make informed judgements about how far we can generalise our results and also how far our statistical analysis is compromised. In this case we might well feel that we could make use of this information to make assumptions about the effects of defluoridation on other small towns in Northern Scotland with similar social compositions, some people might feel there is justification for further generalisation. Note that it the resposibility of people making further generalisations to justify them.
