Do standardized tests really work?

Have you ever taken a test that was meant to tell you your intelligence? Maybe an IQ test?  Maybe you took the SAT, or the ACT.  How

People first started studying human intelligence over a century ago.  Charles Darwin's cousin, Francis Galton (1822-1911) was fascinated with measuring human traits.  When he learned of his cousin's discovery of natural selection, he wondered if it is possible to measure natural ability.  In 1884 he measured over 10,000 people on their abilities doing things such as measuring reaction time, sensory acuity, muscular power, and body proportions.  But he did not find any correlated results with this first study.  Even though this study wasn't successful he did give us some techniques we still use along with coining the phrase "nature and nurture."

Modern intelligence testing was started by Alfred Binet (1857-1911).  France passed a law that required all children to go to school.  France then needed to be able to identify children with special needs objectively.  Binet came up with the idea of mental age, where he assumed children develop intellectually at different rates so a "dull" child just performs like a younger child and a "bright" child performs like an older child.  He decided to take the average performance of a child who was 9 years old and give them work for a 9 year old, but if the child was "duller" he would give that child work meant for a 7 year old.

Binet's test was altered by Lewis Terman at Stanford University to become the Stanford-Binet test.  This test was then altered by a German psychologist, William Stern, to create the famous intelligence quotient, or IQ Test.

To measure an IQ test you take your mental age and divide it by your chronological age and then times it by 100 to get rid of the decimal point.  So if I am 21, and my mental age is 21, then my IQ score would be 100.

Modern testing is much different than testing used to be.  Two ways to now test intelligence are achievement tests, and aptitude tests.  An achievement test, tests someone on things they have already learned, whereas an aptitude test tries to predict someone's ability to learn a new skill.  So the final exam at the end of one of my classes would be an achievement test, but the SAT I took to get into college would be an aptitude test.

Psychologist David Wechsler created the most widely used intelligence test, the Wechsler Adult Intelligence Scale (WAIS), and a version for children, the Wechsler Intelligence Scale for Children (WISC).  The WAIS includes 15 subjects, some of which are: similarities (example, in what way are wool and cotton alike?), vocabulary (naming objects and defining words), block design (seeing a picture and being given blocks and being told to recreate the image with the blocks), and letter-number sequencing (repeating numbers, alphabetizing letters, etc.

Here are some examples of questions on the WAIS:

An accurate intelligence test needs to be standardized, reliable, and valid.

Standardized: 

With a standardized test the average score should fall around 100.  Using the normal curve, this is what Wechsler intelligence scores look like:

From 1918-1998 intelligence scores in every country studied were rising steadily.  However, very recently it has seemed to level off and even reverse in some cases.

Reliable: 

Reliability tests if the test yields consistent results as assessed by the consistency of scores on two halves of the test or on retesting.

The easiest way to test reliability is to retest people, if the two scores agree or correlate, then the test is reliable.  The WAIS and the WISC are very reliable tests with a correlation of +.9 while comparing testing and retesting.


Validity:

A test can be reliable but not valid.   For example, if you showed someone a picture of a blue triangle and asked them what the shape was, and they got it right, then you repeated the test a week later, it would be a reliable test, but not valid in measuring intelligence.  Validity is just the extent to which a test measures or predicts what it is supposed to.  To be valid the tests need to have content validity,  testing behavior of interest; an example would be like doing a list of everyday driving things on a drivers test.  It is better though, if the test also has predictive validity, where the test can also predict how someone will perform in the future.  But the sad fact is, aptitude tests are never as predictive as they are reliable.

___________________________________

How does aging affect our intelligence?  Crystallized intelligence is the accumulated knowledge such as verbal skills, vocabulary, facts, etc, and it increases up to old age.  Fluid intelligence however, is our ability to reason speedily and abstractly, and it begins decreasing in the twenties and thirties up until you're about 75 and then it decreases more rapidly.