Reading comprehension is the understanding of read material, not the understanding of prior or background knowledge that the reader has. Thus, validated reading comprehension tests asses a reader's comprehension of a passage or passages in topics with which he is not already proficient. Often, passages address topics or concepts with which students have some familiarity so readers understand the terms or vocabulary used. However, the passages should present information or content that is new in order to accurately assess comprehension. Although there can be no way to assure that all test takers have limited prior knowledge of a topic, passages are chosen based on evidence that students at a particular grade level have little training in the subject.
Evidence of better-than-chance performance is used in the construct validation of reading comprehension assessment. This kind of evidence may be gathered in different ways. For instance, in multiple choice exams, an exam may repeat a question or slightly rephrase the question to test if the test-taker provides the same answer to the question to glean if the test-taker is guessing. Alternatively, the test may include more multiple choice options (five to six choices, instead of two to three) to reduce the test-taker's chances of guessing correctly.
Since reading comprehension is a mental process, it is critical that reading comprehension exams determine that mental processes are at work. Thus, evidence of the construct validation of a reading comprehension assessment might include evidence that readers (test-takers) actively think and use the mind to respond to questions. The easiest way to gather evidence of such is to include free response or open-ended questions, rather than multiple choice or true/false questions. These kinds of questions compel readers to "think aloud" and to demonstrate meaningful comprehension of a passage.
Inference is the mental process by which a reader applies logic or rationality to answer a question based on the passage. Evidence of questions based on inference support and validate reading comprehension exams, since inference requires readers to accurately and meaningfully comprehend material in order to apply it. Sometimes, general questions based on the content in a passage can be correctly answered if the reader re-visits the passage to look for the correct answer. However, these kinds of questions fail to gauge reading comprehension. Questions that require the reader to infer more aptly assess comprehension.