Sampling

An example would be where invoices up to or under $100 had little approval and those over $100 were dealt with more carefully. Picking a sample from all purchases payments to test could then be invalid. For example, you might pick only invoices under $100 (perhaps because there are many more of them) and obviously could not test controls. What you would have to do is to split the heterogeneous population (split at the $100 value point) and carry out two samples on the resulting two homogeneous populations. Eg Pick 30 <=$100 and do appropriate audit work on those and then pick eg 30 >$100 and do appropriate audit work to check the additional controls were operating.

If a company makes purchase of only particular item locally and the supplier became defunct and the company started making importations mid year, could the same thing be applied in a substantive approach technique?

Yes. There are really two sorts of purchase that need to be tested.

Thanks Mike! One last thing is the definition of tolerable error. It says that tolerable error is a monetary amount set by the auditor in respect of which the auditor seeks to obtain an appropriate level of assurancce that the monetary amount set is not exceeded by the actual misstatement in the popn. Could you plz explain this?

Let's say that true receivables are $100,000. A tolerable error of $2,000 might be set ie the auditor will be happy if receivables are stated from $98,000 to $102,000. So, whatever tests are done and extrapolation for the samples tested, the auditor needs confidence that the receivables are not more than $2,000 from the true receivables