Audits by HMRC Charities - statistical sampling
8 Statistical sampling
8.1 Where the number of documents supporting a repayment claim is large we will use statistical sampling techniques to consider the overall accuracy of the claim. This will help to ensure that the time spent carrying out the audit is proportionate for both the charity and for us.
8.2 These techniques involve:
- identifying all relevant 'populations' within the tax claim being reviewed;
- taking a representative sample of relevant records from each population - the number of records sampled will depend on the size of the population and is fixed by a standard statistical formula;
- selecting individual records at random;
- reviewing the sample records and establishing the extent of any errors;
- calculating the error rate for the sample (i.e. the proportion of the amount claimed for which supporting records were not provided);
- applying the error rate for the sample to the relevant population as a whole and using it to compute an estimate of the tax repaid with respect to that population that cannot be substantiated.
8.3 If you are uncertain about any aspect of this procedure, please ask to discuss it with your auditor before the review.
9 Sampling populations
9.1 Usually, all the donations relating to a single claim will constitute a single 'population'.
9.2 Where donations are processed in significantly different ways, there may be a case for treating them as distinct populations. For example:
- donors giving more than a particular amount may be treated in a different way;
- donations relating to a particular sponsored event may be handled differently;
- different branches of a charity may have different record keeping procedures;
- some donations may be handled by a third party using different processes.
9.3 Where some donations are processed in significantly different ways, you should advise the auditor before the review visit.
9.4 It is possible that the review itself may reveal unexpectedly significant variations in the treatment of different groups of donations. In such circumstances, another review may be necessary. The auditor will discuss this with the trustees and/or senior management for a larger charity.
10 Application of results to earlier periods
10.1 If we identify errors in relation to a particular donor population, we will assume that the same error rate is likely to apply to all similar claims for the same population.
10.2 If earlier claims were handled in a significantly different way to the one reviewed in our audit, please discuss this with the auditor – ideally before the review visit.
10.3 It may, for example, be the case that:
- records for earlier claims are maintained on paper instead of computer or
- a member of staff identifiably responsible for flaws in the preparation of a particular claim was not involved in the preparation of other claims.
10.4 We may need to look at a claim (or part of it) separately if we agree that the sample used is not representative for that claim.
