Monetary unit sampling options

For monetary unit sampling only, you can optionally specify one or more of the following:

Top stratum cutoff

Top stratum cutoff applies only to monetary unit samples that use fixed interval or cell sampling. Using top stratum cutoff, the data set is stratified into two ranges, using the interval as the default cutoff value for the strata. The bottom stratum contains all of the records whose amounts are less than the interval and the top stratum contains all records whose amounts are equal to or greater than the interval. All records in the top stratum are automatically selected for inclusion in the sample.

When sampling data, you have the option of specifying a cutoff that is different from the default interval value. You can also remove the top stratum cutoff completely by specifying a cutoff value that is large enough that none of the items fall into the top stratum.

For example, if you are sampling with an interval of $1,000,000, by default all items with an absolute value equal to or greater than $1,000,000 would be in the top stratum and automatically selected. If you specify 900,000 as the cutoff, an item with a value of $900,000 will fall in the top stratum and be automatically selected, even though its chance of selection would be only 90% using the sampling algorithm. Similarly, if you specify 10,000,000 as the cutoff, an item with a value of $5,000,000 will fall in the bottom stratum and have a proportionate chance of selection. While the probabilities may never exceed 1 because the value of this item is 5 times the interval, its chance of selection would nominally be five, indicating that it will likely be selected five times.

Subsampling

In some cases, one record in a table represents a number of separate transactions. For example, records that post totals from an accounts payable subsystem. You can use the Subsample option to test only one of the constituent parts of a record.

Note

Subsampling is only available for monetary unit sampling using field output.

Specifying Subsample causes ACL to display the actual dollar (cent) unit used to select that item to be included in the sample. For example, if the selected amount was $12,455, the actual dollar selected might be the 4,620th dollar of this amount. If Subsample is selected, $4,620 is displayed in the SUBSAMPLE column.

Top stratum cutoff items contain a $0 value in the SUBSAMPLE column because they are automatically included in the sample and no selected value is assigned to them.

Sampling without repeats

When selecting samples, ACL makes multiple selections from the data in a table. Depending on the sampling method, selected items might be set aside once they are drawn, or they might be put back into the stream to be potentially drawn again.

The default setting in ACL is to allow sampling repeats, or replacement. In sampling with repeats, selected items are returned to the stream, making them available to be selected again. As a result, the probability of reselecting these items remains the same for successive selections.

When sampling without replacement, or No Repeats, an item that has already been selected is set aside and cannot be drawn again. This changes the probabilities of selection for the items remaining in the selection stream.

Note

The No Repeats option is only available for monetary unit sampling. Record sampling always draws samples without replacement.

To compensate for the suppression of any duplicate sample items and to ensure that the full number of required sample items is selected, you can oversample. Oversampling is often used when the No Repeats option is used in conjunction with random and cell sampling.

Related concepts
Sampling data
About sampling types
About calculating sample sizes
About sample selection methods
About evaluating sampling errors
Related tasks
Sampling transaction records
Sampling monetary unit records


(C) 2013 ACL Services Ltd. All Rights Reserved. | Send feedback