A good IQ test follows a standardized process. It is administered and scored according to established criteria designed to control everything in the test environment except the test-taker's answers. The examiner should be a professional educator who strictly adheres to procedure. The rules also include where the test is given and what people other than the examiner are present in the room. Content of the test is also standardized. IQ tests measure verbal and non-verbal performance and abilities such as reasoning, perception and memory.
According to Jonathan Rich, Ph.D., on the Psychological Testing website, the reliability of an IQ test is based on whether test scores are consistent over time. For example, if a subject scores below average one week and well above average the next week, the test is not reliable. Adult test scores are considered more reliable than the test scores of children. Children, especially very young children, are more susceptible to factors such as fatigue, hunger or whether an examiner puts them at ease.
An IQ test is valid if evidence shows that it succeeds in measuring what it set out to measure. Because of a strong correlation to factual data such as academic grades and career success, IQ tests are proven indicators of how well students will perform in school. The most widely administered IQ tests --- the Stanford-Binet Intelligence Scale, the Wechsler Adult Intelligence Scale and the Wechsler Intelligence Scale for Children --- are considered valid and reliable instruments for measuring human intelligence.
Norm-referenced IQ tests compare an individual's score to others in the same age group who are taking the same standardized test. For an IQ test to be norm-referenced, it first must be administered to a large group of individuals that will serve as the statistical basis for comparing the test scores of peer groups. On the Stanford-Binet and Wechsler IQ tests, the mean IQ is 100 and points are added or subtracted for any deviation above or below 100.
A longstanding controversy surrounding IQ tests is the question of cultural bias --- whether the tests favor individuals with a Caucasian background. According to the National Center for Fair and Open Testing, while tests are reviewed for "obvious biases" and statistical techniques are used to reduce bias, unconscious bias can creep into the test format, content and administration. Psychologists, including Dr. Jonathan Rich, generally acknowledge that it may be impossible to create a completely culturally unbiased test. Despite this flaw and weaknesses in measuring abilities such as creativity, Rich believes IQ tests "do a good job of predicting academic success," according to his website, Psychological Testing.