The sampling type you choose depends on which sampling bias is appropriate for the type of data you are working with. Sampling bias is the chance, or likelihood, that a particular record or item will be selected.

## Monetary unit sampling

Monetary unit sampling is typically used when auditing assets and expenses. This type of sampling uses the absolute value of a field to determine which records are selected for inclusion in the sample and therefore biases higher valued items, both positive and negative. The probability that any given record will be selected is directly proportional to the value of the specified field, unless the item is affected by top stratum cutoff.

With monetary unit sampling, the total absolute value of all the amounts in the selected field is treated as a stream of dollars, expressed in cents. Each dollar, exclusive of the individual amounts, has an equal chance of selection. Therefore, a \$1000 item is 1000 times more likely to be selected than a \$1 item. This selection type creates a bias toward greater dollar items and is the standard tool for detecting overstatements.

## Record sampling

Record sampling, also called transaction sampling, is typically used in control testing. This type of sampling treats each record equally, using a nominal value of 1. This results in an unbiased sample that is not based on the values in a record. Each record has an equal chance of being selected for inclusion in the sample. Because of this equality, there is a significant probability that very large monetary transactions will be overlooked with this type.

With record sampling, the data set includes all of the records in the table and the data is treated as a stream of records. Consequently, a \$1000 item and a \$1 item have an equal chance of being selected. This selection type is ideal for testing binary (yes/no) conditions, and detecting understatements on quantities.