For the implementation of inventory sampling, various extrapolation methods are permitted. These can be categorized into testing methods and estimation methods. While testing methods such as the homogeneous sequential test only allow for a general quality assessment of the reliability of the book inventory (as “acceptable” or “not acceptable”), estimation methods – for example, stratified mean estimation – enable an evaluation of the quality of the inventory records based on specific indicators (relative sampling error and percentage deviation between book and estimated values).
All procedures imply that the results of an inventory sampling must not exceed defined tolerance limits for these quality indicators. For instance, when using an estimation method, the relative sampling error must not exceed 1%, and the percentage deviation between the book value and the estimated value must not exceed 2%.
Testing methods such as the homogeneous sequential test may be applied with very small sample sizes but generally place higher demands on the quality of the underlying inventory records. They are typically used successfully in fully automated inventory management systems, where only a few organizational deviations occur. Estimation methods, on the other hand, can also be applied when deviations between recorded and actual stock levels are expected – as long as these deviations remain within a defined range.
Stratified mean estimation has proven to be a practical and targeted extrapolation method under all conditions (different numbers of items, varying expected inventory quality, etc.). When its parameters are tailored to the specific warehouse, it yields reliable estimations and can often reduce the total effort required for conducting the inventory sampling to an absolute minimum.
Applying multiple testing and estimation methods simultaneously, and then selecting the one that yields the most favorable outcome, is fundamentally not permissible.