Chapter 6

Use these questions to check what you have learned throughout the chapter.

In the beetle experiment in section 6.2, the design using only one forest of each type (and sampling 50 trees in each) was rejected. Can you explain why?

Imagine that you are interested in interviewing final year students in schools in your area to learn how they decided whether to go to university and which university to choose. Bearing in mind the issues raised in section 6.2 on subsampling, decide how many schools you should visit and how many pupils you should interview in each. (Imagine that there are 60 schools in the area. It takes 5 hours to arrange each school visit. Interviewing a pupil takes 15 minutes.)

Would you interview all the final year students at each school, or would you sample? Justify your decision.

If you could only sample 25% of the pupils at each school, how would you select this sample?

If a researcher were studying schoolchildrens’ perception of racial intolerance in a particular city, they might use stratified sampling by ethnic group. This should allow them to collect representative data from all ethnic groups. In contrast, by simple randomly selecting from the school rolls they might find that (by chance) their random sample means that they have poor data or no data on ethnic groups that make up only a small proportion of the general population, but whose views might be particularly interesting. Can you think of analogous situations in another area of life sciences, where stratified sampling might be preferred over random sampling?

In the example in section 6.4.3, can you see any way that asking the doctor to give the questionnaire out to every 10^th patient would produce a biased sample?

Can you think of a hypothesis where the interesting biological question depends on whether there is an interaction between two factors, and a factor and a continuous variable?

Oxford University Press, Online Resource Centre, Chapter 6.