If the elements of a sample are not independent, then the raw sample size may not be the best value to use when thinking about the data. The effective size of a sample when these factors are taken into account may be smaller than the raw sample size. For example, if we asked 100 random people attending a football match which team they supported, we could use this to estimate the popularity of the teams based on a sample size of 100. However, if we had instead obtained the sample by taking 50 people each from two coaches, then the effective sample size would probably be 2 rather than 100, as it is likely that everyone on a coach supports the same team.
Used on page 36