Standardized tests generally include at least some multiple-choice and true-false questions. These can be graded by computer, or by humans who do not understand the material in depth, as long as they have a list of the correct answers. One potential defect in such tests is that the test-taker can accidentally skip a line and then be marked wrong on material to which he or she knew the correct answer.
Standardized tests often include written portions as well; these are graded by humans who have use rubrics or guidelines as to what a good essay on the subject will be.
Some believe that standardized tests make it possible to compare the achievements of students from different schools, eliminating bias from grade inflation and the influence of schools' reputations on university admissions officers. Others contend that standardized tests reinforce bias in education because students whose families have access to enrichment opportunities do better on such tests than students from other parts of a society. Such tests, as objective as they try to be, are tools of culture and are rooted in whatever cultural or philosophical understanding gave them rise.
Standardized tests only generate useful information if each question generates a ratio of correct to incorrect answers. So if all(or most) test-takers get an answer correct, the test question needs to be eliminated from the test. Similarly, if all (or most) test-takers get a particular question incorrect, the test question needs to be eliminated. Standardized, objective tests create some meaning by comparing correct responses to incorrect ones.
Most frequently, tests are administered to people of similar ages or grades at particular times in development. This practice allows researchers to study correct to incorrect response ratios for 3rd graders (for example) over time or from region to region. The wide array of test-takers in a particular group generates data to establish normative classification for a group. Those data can serve as a guide for what that particular test finds as a normative response for that test. For that reason, many standardized tests are also called "norm referenced."
Most standardized tests are achievement tests and therefore have little predictive value for students, teachers or schools. They only generate good data for narrow skill sets or topics. However, many school systems use standardized testing as a screening tool, as a basis for curriculum or as a broad comparison between students. Standardized tests have limited value and need to be regarded as only one way to measure performance, potential or intelligence.
See also Standardized testing and public policy, education, alternative assessment