Usually, a reviewer will review 100% of a document, however, it is possible to review just a sample of the document. This sample can be set at an elective interval between 10 and 100. It must always be a whole digit divisible by 10.
Review sampling is set for the entire project despite the fact that it is edited on a document. This means that if review sampling is set to 30% and there are 2 documents of 10 segments each, the documents may have 20% and 40% of segments to be reviewed, or 10 and 50.
The translator will still translate 100% of the document, and the reviewer will then review the selected amount; if set at 20%, the reviewer will then review 20%.
The error rate is also calculated on the basis of the sampled review on the following formula:
Error rate = (segments with one of the above error tag)/(total number of segments reviewed by the reviewer i.e. unlocked segments)
The error categories counting towards the error rate can be seen here (link to error rate page)
If the error rate is found to be above 5%, the project manager will then have the option of increasing the review sampling. It is not possible to lower a review sample percentage, it is only possible to increase.