Are IQ tests reliable?

Question

Some of the more common types of IQ tests are:

1. Stanford-Binet Intelligence Scale, Fifth Edition (SBIS-V) (Ages 2–90+)
   An update of the SB-IV. In addition to providing a Full Scale score, it assesses Fluid Reasoning, Knowledge, Quantitative Reasoning, Visual-Spatial Processing, and Working Memory as well as the ability to compare verbal and nonverbal performance.

2. Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV) (Ages 6–16-11)
   An update of the WISC-III, this test yields a Full Scale score and scores for Verbal Comprehension, Working Memory, Perceptual Reasoning, and Processing speed.

3. Woodcock-Johnson III Tests of Cognitive Abilities (Ages 2–90+)
   This test gives a measure of general intellectual ability, as well as looking at working memory and executive function skills.

4. Cognitive Assessment System (CAS) (Ages 5-17)
   Based on the “PASS” theory, this test measures ‘Planning, ‘Attention, ‘Simultaneous, and ‘Successive cognitive processes.

5. Wechsler Adult Intelligence Scale (WAIS) (Ages 16-89)
   An IQ test for older children and adults, the WAIS provides a Verbal, Performance, and Full Scale score, as well as scores for verbal comprehension, perceptual organization, working memory, and processing speed.

6. Comprehensive Test of Nonverbal Intelligence (CTONI) (Ages 6–18-11)
   Designed to assess children who may be disadvantaged by traditional tests that put a premium on language skills, the CTONI is made up of six subtests that measure different nonverbal intellectual abilities.

7. Universal Nonverbal Intelligence Test (UNIT) (Ages 5-17)
   Designed to assess children who may be disadvantaged by traditional tests that put a premium on language skills, this test is entirely nonverbal in administration and response style.

8. Kaufman Assessment Battery for Children (KABC) (Ages 2-6 to 12-5)
   This test measures simultaneous and sequential processing skills, and has subscales that measure academic achievement as well.

–Adapted from Table 1 of "The Testing Process"   [cached image]

There have been many IQ tests used over the years, and many different proposed scales to evaluate the results. Many tests have been revised and edited in attempts to provide more accurate results. A general outline of this can be found here.

There is, of course, some criticism surrounding the reliability of IQ tests...

There are sites like this which quote sources such as Walter Lippmann who said that

"We cannot measure intelligence when we have not defined it" –[source]

There are other studies which suggest that the outcome of an individual IQ test is largely dependent on the motivation of the person taking it. Or at least that individual motivation to succeed introduces a variable into the evaluation process which is not accounted for by those determining results.

The theory of multiple intelligence seems to assert that there are levels of intelligence which are not measured by standardized IQ tests. Howard Gardner is quoted as saying:

In the heyday of the psychometric and behaviorist eras, it was generally believed that intelligence was a single entity that was inherited; and that human beings – initially a blank slate – could be trained to learn anything, provided that it was presented in an appropriate way. Nowadays an increasing number of researchers believe precisely the opposite; that there exists a multitude of intelligences, quite independent of each other; that each intelligence has its own strengths and constraints; that the mind is far from unencumbered at birth; and that it is unexpectedly difficult to teach things that go against early 'naive' theories of that challenge the natural lines of force within an intelligence and its matching domains. (Gardner 1993: xxiii)

Questions:

1. Has the term "intelligence" been defined well enough that it can be objectively measured?

2. If so, do IQ tests reliably determine an individual's intelligence? Or are they too subjective to be considered accurate?

3. Is there scientific evidence to support claims of being able to accurately assess individual intelligence as related to real-world outcome?

4. Or, are IQ tests at best a ballpark estimate of individual intelligence?

Answer 1

Good Summary:

A good summary of the literature on Intelligence can be found in the article "Intelligence: Knowns and Unknowns" (FREE PDF). You will find references for the claims I make below and further discussion. It was based on an American Psychological Association Task Force and represents propositions that received "unanimous support of the entire Task Force". It addresses each of your questions.

Below are just some general observations regarding your points.

Has the term "intelligence" been defined well enough that it can be objectively measured?

• This is open to debate. There are competing definitions.
• Good intelligence tests certainly measure something. And that which is measured is one of the best predictors of education outcomes, and work performance, and is also a good predictor of many other important outcomes.

If so, do IQ tests reliably determine an individual's intelligence?

• IQ tests are reliable in that longitudinal studies have shown that test scores are very stable over long periods of time. Thus, there is good evidence that what is measured is a trait.
• IQ tests do not determine "intelligence"; theory would suggest that some factor in the individual determines test scores. Whether you label this factor "intelligence", depends on definitions and is open to debate.

Or are they too subjective to be considered accurate?

• Intelligence tests are not a subjective measure, in that the score derived from an intelligence test is generally objectively determined by standardised scoring processes.
• Deciding what intelligence test scores truly mean is still an open question.
• However, plenty of research has explored the correlates

Is there scientific evidence to support claims of accuracy?

• What do you mean by "accuracy"?
• There is plenty of evidence that Intelligence tests are useful in predicting important real-world outcomes.

Or are IQ tests at best a ballpark estimate?

• Estimate of what?