Calculating Sample Size
It is advisable to use a sample size calculator for determining appropriate sample size for any generalizable sample. However, there are major differences between available utilities or applications.
Online sample size calculators
Online sample size calculators from reputable sources are accessible and easy to use. Available features vary, but most allow the user to adjust population size, confidence level, and confidence interval. However, not all online utilities allow the user to adjust the expected error. The major difficulty with practically all of these utilities is the distribution of variance that is used to make the calculation. Almost exclusively, online utilities use a basic t distribution for calculating sample size. This method is more appropriately used for normally distributed data. These online calculators should not be used for calculating sample sizes for audits, especially when the expected error is less than 30%.
The input variables for attribute sampling can easily be translated into terms used by the IDEA utility. Table 1C summarizes and compares the input variables needed for attribute sampling and how to translate these into the input variables that appear in the IDEA sampling utility (version 10). The value of tolerable error has to be calculated based on the desired confidence interval and the expected error level.
Table 1C – Adapting Attribute Sampling Variables for Common Software Sampling Utilities
| Typical Input Variables for Attribute Sample Size Calculation | Input Variables for IDEA Attribute Sample Size Calculator | |||||||
|---|---|---|---|---|---|---|---|---|
|
Input Variables |
Population Size |
Confidence Limit |
Confidence Interval |
Expected Error |
Population Size |
Confidence Limit |
Tolerable Error |
Expected Error |
|
Formula |
N |
CL |
CI |
EE |
N |
CL |
CI+EE |
EE |
|
Ex 1 |
500 |
90% |
5% |
10% |
500 |
90% |
15% |
10% |
|
Ex 2 |
500 |
90% |
5% |
15% |
500 |
90% |
20% |
15% |
|
Ex 3 |
500 |
90% |
10% |
20% |
500 |
90% |
30% |
20% |


