Sampling is in essence an estimation of tax compliance. Taxpayers may be either in a position of underpayment, overpayment or zero payment. There is a direct correlation between the level of accuracy of the sample and the accuracy level of the underlying documents. After providing a short explanation of a statistical sample, this article examines whether sampling should be allowed and, if so, under what conditions, as sampling is an estimation method and does not provide exact information on the tax position of the taxpayer. Discussed are the role of the allocation of the burden of proof, the type of tax, the principle of fair play, the influence of the confidence interval, and the risk of inaccuracy.
Separate discussions are offered on the use of sampling for the determination of interest and penalty payments, and on the issue of extrapolation of sampling results. The author supports the use of sampling agreements between taxpayers and tax authorities, because it clarifies the terms under which the sampling will take place and will likely increase understanding of the process by the taxpayer and, thus, his acceptance of the outcome.Intertax