Need help - Sample size issue 2558

  • Hi all,
    I have a question:
    A key control is quarterly and each control consists of checking let’s say 20-25 reports (these are checked by the control owner). He does not fill out an sheet where he confirmed controlling these sheets.
    Assuming that the sample size in a quarterly control is 2 and if the yearly population is 30-150, then 30 samples.
    Do I have to control 2 spreadsheets (2 quarters), 50 spreadsheets (25 spreadsheets *2) or 30 spreadsheets??? Or a different sample size?
    What is the correct sample size?
    Pls help,

  • For each answer I will spend a virtual cake. 😄

  • There is no right answer to this, and each of the suggestions you make on sample size I have seen in practice. That said …
    … I would have thought you could operate the control such that you can justify the sample being 2. The key here would be to operate the control in so that you evidence that all 20-25 spreadsheets/reports have been checked each quarter. The control you are then testing is the one that monitors that the detailed activity has happened.
    This would be easier 😉

  • If these spreadsheets were account reconciliations, you would likely test a sample of 2 of each reconciliation for the year.
    While I agree that one control is a checklist to ensure that these spreadsheets are being reviewed, I worry that a sample size of 2 would not be adequate to catch issues. I would argue that you test to ensure that the checklist was being completed, but also check at least one sample of each spreadsheet (report) on an annual basis to ensure that it is being controlled properly.
    As Denis mentioned, the sample size can be any of the suggested totals. As always, judgment is involved in choosing your sample size. Ultimately, your sample size should be based on risk. Do you have excellent entity-level controls in place? How material are these spreadsheets to the financial statements? How complex are they? All of these figure into your ultimate risk and your sample size.

Log in to reply