AP Psychology 2.8: Intelligence and Achievement

Psychologists have never fully agreed on what intelligence is or how to measure it, and the tests they built have carried the assumptions, and sometimes the biases, of the people who wrote them.

One long debate is about its shape. People who do well on one kind of mental task, such as defining words, tend to do well on others, such as number puzzles. Some researchers take that as evidence of a single general intelligence, called g, underneath every kind of thinking. Others point to people who are strong in one area and ordinary in another, and argue that intelligence is made of multiple abilities that can develop separately.

The first formal intelligence tests gave each child a mental age: the age of typical children who answer the same questions correctly. The intelligence quotient (IQ) divided mental age by chronological age, the child's actual age, and multiplied by 100. Today, IQ scores are often used to help identify students for educational services.

Those early scores were also used to shut people out. In the United States they were used to limit access to jobs, military ranks and places in schools and colleges, and to argue for limits on immigration. Some early test developers supported the eugenics movement described in 1.1 Interaction of Heredity and Environment.

Every psychological test, an intelligence test included, has to follow sound principles of measurement to be useful.

A test is standardized when it is given using the same procedures in the same kind of setting for everyone: the same instructions, the same time limit, a similar quiet room. Without that, a score might reflect the conditions rather than the person.

A test is valid if it measures what it is designed to measure. Construct validity asks whether it truly measures the quality it names; a "logic" test packed with long, tricky wording may really be measuring reading. Predictive validity asks whether scores forecast what the test is meant to predict, such as whether an entrance test score goes with grades a year later.

A test is reliable if it gives similar results each time it is given. Test-retest reliability is checked by giving the same people the test twice, some time apart, and comparing the scores. Split-half reliability compares each person's score on one half of the items, such as the odd-numbered ones, with their score on the other half.

Consistent is not the same as accurate. A bathroom scale that always reads five pounds heavy is reliable, and it is still wrong.

How a test is given can change a score. Stereotype threat is the worry, during a test, of confirming a negative stereotype about a group you belong to. The worry takes up attention and can pull a score down. Stereotype lift pushes the other way: knowing that another group is negatively stereotyped can give your own score a small boost. Both make scores less fair, so researchers try to build socio-culturally responsive tests, whose questions, instructions and setting draw on experiences shared across backgrounds and avoid calling up stereotypes.

Reading a score needs the same care. IQ scores vary more within a group than between groups. Inside any group, scores spread from low to high, and that spread is far wider than the gap between two groups' averages, so a group's average says little about any one person in it.

Poverty, discrimination and unequal schooling can lower the scores of individuals and of whole groups, so a lower score can reflect a person's circumstances rather than what they could learn. And whoever reads a score brings their own personal and cultural biases, which can shape what they think the score means and what they expect it to predict.

Average IQ scores have risen across much of the world over the past century, enough that test makers keep having to reset what counts as an average score. This is the Flynn effect. The rise came far too quickly to be genetic. It is linked to changes in how people live, such as higher socioeconomic status, better health care and better nutrition.

Not every test at school is an intelligence test. An achievement test measures what someone already knows, like a final exam on a history unit. An aptitude test tries to predict how someone will do in the future, like a test that forecasts who will succeed in a training course.

Beliefs matter too. A student with a fixed mindset believes intelligence is set from birth, so a low grade feels like proof of a limit and more effort seems pointless. A student with a growth mindset believes ability can grow with experience, so a low grade points to what to work on next. That difference can affect achievement: students who see ability as changeable tend to keep working after setbacks, and the extra work can show up in their grades.