Step 3 – Preparing a Sampling Plan

Once a sampling approach has been selected, the next step is preparing a detailed sampling plan. Audit teams that decide to use a generalizable sampling approach should consult the section Key Elements of a Sampling Plan – Generalizable Sampling. Audit teams that decide to use a purposeful sampling approach should consult the section Key Elements of a Sampling Plan – Purposeful Sampling.

Key Elements of a Sampling Plan – Generalizable Sampling

Before starting the audit’s examination phase, the scope, methodologies, and objectives are documented and reviewed with the audited organization. If generalizable sampling is to be used during the audit to gather evidence, then it is suggested that a sampling plan could be developed and, if appropriate, discussed with the auditees before the examination phase. A good generalizable sampling plan should cover the following seven key elements.

Key element 1: Population analysis

Where generalizable sampling is concerned, there is usually some form of administrative database with a reasonable amount of quantitative data that can serve as a sampling frame. This data can also be used to perform basic population analysis. The plan should provide an assessment of the data quality.

The plan could also include histograms and bar charts that adequately describe the distribution of data and potential sources of heterogeneity. If this analysis indicates that there is enough heterogeneity to be a concern, it needs to be documented. If needed, the plan could contain a strategy to segment the population into more homogenous groups.

Key element 2: Objectives of sampling

The plan could indicate that this is a generalizable sampling approach and explicitly state what aspects of the sample are being extrapolated to the population. (Rarely is there a single type of error or deviation being measured.) This section of the sampling plan lists what population parameters are being estimated.

Key element 3: Level of accuracy desired

The plan could describe the level of accuracy desired for each segment of the population being sampled. Auditors should determine whether this level needs to be low, medium, or high for each segment of the population based on the expected quality of controls in place and the impact if risk is materialized. This determination also has to be contextualized based on other evidence available that in part mitigate the risk of inadequate controls. In some circumstances, sampling may be providing supplementary evidence to other audit procedures and this may also affect the level of accuracy needed.

Key element 4: Established expected error level

Although performance auditors may be able to rely on the experience of other auditors (internal or financial audits) on similar topics and in the same entity they plan to audit, they will often have to deal with a limited availability of empirical information that can be used to predict the level of error. In such instances, auditors should choose, and document in their plan, a level of error that is a high estimate and for which their audit recommendations would not change if the level of observed error were higher than expected. If the sample is insufficient to serve the purposes intended, it may be necessary to extend the sample.

Key element 5: Reportable findings

Before making any observations, auditors could state in their plan what conclusion would be appropriate for various levels of error that might be observed and reported.

Key element 6: Sample size calculation

Auditors could calculate an appropriate sample size for each segment of the population that has been defined and that they intend to sample from. The plan could indicate the parameters used to determine sample size and the specific software or method of calculation used. More information on how to determine the sample size for a generalizable sample is in Appendix 1.

Key element 7: Sampling method

The plan could describe the method of sampling for each segment of the population that has been defined and that the audit team intends to sample. The plan could indicate whether the method of sampling is simple random, systematic random, proportional stratified, non-proportional stratified, or some other method of sampling. (More information on unbiased methods of selection for generalizable sampling is in Appendix 2.) Auditors could be specific with respect to how random selection is accomplished and ensure that the method used can be substantiated as unbiased. Sampling software does offer log files that can substantiate unbiased sampling (i.e., the software keeps track of the random seed and all steps taken to determine the sample). Ideally, another person should be able to repeat the sample selection by using the same seed, software, and population. Otherwise, auditors could ensure that more than one person witnesses the selection process and is willing to substantiate that the chosen method was conducted appropriately.

• PreviousOther Factors to Consider When Selecting a Sampling Approach
• NextKey Elements of a Sampling Plan – Purposeful Sampling