Audit Manual

Sampling Guidelines

Background

The extent of testing a sample of controls or transactions may vary depending on a variety of factors including complexity, population size, transaction frequency, importance, type of activity (manual or automated), and level of comfort desired from a test. Additionally, many sampling methodologies are available including statistical (e.g. systematic/interval or dollar unit) and non-statistical (e.g. judgmental) techniques. The software tool, ACL, is available for use for systematic and dollar unit statistical sampling methods.

Systematic Sampling

  • A random approach of selecting items at intervals. The first item in the selection process must be picked at random. Often used in financial auditing to test for understatement.
  • Example: Choose a sample of 40 items from a population of 2,000 unnumbered documents. Divide the population of 2,000 by the sample size of 40, making the sample interval 50. Start the selection with a random number of say, 15, which is less than the sample interval of 50, and the sample items would be the 15th item, the 65th item, the 115th and so forth - every fiftieth item until 40 items are selected.
  • Suggested documentation of sampling method: Selected a random sample of 40 items, using the Systematic Sampling method, by choosing every 50th item, beginning with the random number of 15.

Dollar Unit Sampling

  • A unique statistical approach based on a probability proportional to size. The probability of any one item being selected for detailed verification is proportional to the size of the item. Often used in financial auditing to test for overstatement.
  • Example: The probability that M&O account transaction No. 1 ($10,000) is selected is 10 times greater than that of M&O account transaction No. 2 ($1,000).
  • Suggested documentation of sampling method in workpaper: Selected a random sample of X documents, using the Dollar Unit Sampling method, based on the materiality identified in planning and a 95% confidence level.

Judgmental Sampling

  • A nonrandom approach of selecting sample items based on the auditor’s reasoning or suspicions. Often used to select examples of deficiencies to support the auditors’ contention that the system is weak.
  • Can not statistically extrapolate sample results to the entire population
  • If no errors, may be able to say that there is no basis for examining the population further or for suspecting any material error.
  • Example: Select X expenditure transactions from the travel account from the fiscal year and determine that there was documented authorized approval. The outcome will provide the auditor with adequate assurance that transactions are properly approved.
  • Suggested documentation of sampling method in workpaper: Selected a judgmental sample of X travel expenditure transactions based on amount, description, and date.

Although statistical sampling provides the best assurance of obtaining accurate results, it can be time-consuming and difficult to apply. As a result, most large auditing firms developed and follow non-statistical sampling guidance, based on statistical sampling methodologies. The guidance below is derived from PricewaterhouseCooper’s publication, Sarbanes-Oxley Act: Section 404, Practical Guidance for Management. Sample sizes proscribed in this publication are based on the number of times a transaction occurs and the amount of comfort needed to be derived from the test based on the importance and complexity of the transaction.

Given the ease of use of applying these guidelines, the acceptance among the major accounting firms, and the consideration of the risk of the transactions, these sample size guidelines are appropriate for use in our internal audits of controls and transactions.

Manual Controls and Substantive Testing

The minimum sample sizes for use when testing manual controls and transactions that operate annually, quarterly, monthly, weekly, daily, or multiple times per day in an audit of controls or transactions are as follows:

 

Frequency of Control

Number of Items to Test
Low Mid High
Annual 1
Quarterly 2
Monthly 2                  to                   5
Weekly 5 10 15
Daily 20 30 40
Multiple times per day 25 45 60

The number of items to test is based on the level of comfort desired for an activity as determined by the risk assessment process for each engagement and should be documented in the testing. Items that are complex or where controls are weak would require a high level of comfort. Transactions that are routine in nature and do not have a history of findings would require a low level of comfort.

Automated Controls

For an automated control, selecting one item for testing is generally sufficient. However, in order to reduce our sample size to one, we must have tested general computer controls and inquired about explicit override policies and procedures that may impact the functioning of the automated control.

Small Populations

The guidelines above were established for populations in excess of 200. For smaller populations, smaller sample sizes may be considered where coverage may be obtained. Coverage of 50% of the number of transactions or the total dollars in the population would be an acceptable alternative to sample sizes listed above.

Documentation

The selection of sample sizes should be documented in the workpapers. The following items should be documented each time a sample is chosen:

  • population
  • sample size
  • sampling unit
  • sample design (e.g., random, haphazard, systematic)
  • definition/explanation of an error
  • errors detected and/or error rate

Exceptions

Alternative sampling methodologies (i.e., variations from these guidelines) should be discussed with the Engagement Manager and Director of Audits as part the understanding meeting to ensure adequate testing for the engagement is completed.

Note: The sampling methods are based on information outlined in Sawyer’s Internal Auditing, The Practice of Modern Internal Auditing, 5th edition, Chapter 11 on Sampling.