Which statistic would be most affected by introducing bias into the sample of reviewed documents?

Precision is defined as the ratio of relevant documents retrieved to the total number of documents retrieved. When bias is introduced into the sample of reviewed documents, it can significantly distort this ratio. This bias may result in irrelevant documents being included in the sample or relevant documents being overlooked, which directly impacts the precision calculation.

When the sample is biased towards certain types of documents or categories, the chance of incorrectly classifying non-relevant documents as relevant increases, which lowers precision. Since precision reflects the quality of the results produced by the sample, any bias that alters the distribution of relevant versus non-relevant documents will thus have a profound effect on this statistic.

On the other hand, recall, which measures the ratio of relevant documents retrieved to the total number of relevant documents available, may not be as directly affected by sample bias in terms of its calculation structure, though it could still be influenced in a broader context. Elusion rate and total number of documents do not directly relate to the bias in the same manner, as elusion rate refers to the missed relevant documents and total number simply counts the documents without reflecting their relevance. Therefore, precision is the statistic that would be most affected by introducing bias into the sample of reviewed documents.