Volunteering and health The doctor is also interested in the experiences of people from two minority ethnic groups among her patients. One group have lived in the area for many years, while the others are recent migrants who have lived in the area for only a year or so. Their ethnicity is recorded on their case notes. Each group is only a small percentage of her total patient list (both groups together make up about 5%, or 250), so her initial sample is likely to include just a few of each group. The doctor therefore decides to take an additional sample, made up of people from these two minority ethnic groups. In this case she selects a higher proportion of the total population, 1 in 2, to ensure that her sample is large enough to enable her to analyse her findings on the same variables of age and gender as her larger sample. However, when she comes to analyse her data, she cannot simply add the two samples together, because the samples have been drawn using different sampling fractions, and patients of the two ethnic groups had the chance of being selected in both samples. The doctor can analyse the two samples and present the findings for each sample separately, and the findings from the two samples can be compared using percentages rather than raw numbers. An alternative, or additional, approach would be to select two samples:
A: from the total population excluding the two ethnic groups – using a sampling fraction of 1 in 10;
B: from the two ethnic groups using a sampling fraction of 1 in 2. Each sample would then be analysed separately and compared using percentages. Then, to analyse both samples together as representative of the total patient list, the doctor would need to reduce the weight of sample B. To do this she would need to divide any results from sample B by 10/2 5 as the patients from the two ethnic groups were five times more likely to be selected than the other patients.
Note that the data gathered from all those included in sample B is used and the weighting is applied to it. Do not be tempted to simply select, for example, 25 out of the 125 patients from the small ethnic groups to include and exclude the remainder