Cognition, Language & Intelligence

This is Unit Seven of Psychology, Part Two: intelligence, and measuring it. Part One was about thinking. This part is about measuring the capacity to do it — and the first question is whether there is one capacity to measure or several. Four theories answer that differently, and the disagreement is not settled. There is a myth in here that is probably the most widely believed false claim in education. Matching instruction to a student's preferred learning style does not improve outcomes. The preference is real; the benefit is not there. The centre of this part is about what makes a test a test at all, and it contains the sharpest pair of entries in the course. A measure can be highly reliable and completely invalid — consistent results that do not reflect what was meant to be measured. And the entry after it explains why that combination is dangerous rather than merely wrong. And the unit closes with a procedure for reading a test report, which puts the score last — after four other things have been checked. By the end you will be able to do five things. One: say what each theory of intelligence claims. Two: say why the learning styles claim is a myth. Three: say what standardization establishes. Four: say why a reliable measure can be invalid. Five: read a test report in the right order.

Five entries. A definition, and then four theories that disagree about the most basic question available — whether intelligence is one thing that shows up in different tasks, or several separate things that happen to be measured together.

The definition.

Psych 0789 Intelligence

The capacity to learn from experience, reason well, and adapt to the demands of the environment. Three capacities in one definition, which is where the disagreement begins.

One factor.

Psych 0791 Spearman's General Intelligence

A general intelligence factor, proposed because performance across different mental tasks tends to correlate. The argument is statistical — the correlation is the evidence, and it needs an explanation.

Two kinds.

Psych 0793 Fluid and Crystallized Intelligence

Fluid intelligence is reasoning with novel problems; crystallized intelligence is accumulated knowledge and skill. The two follow different courses across a lifetime.

Three.

Psych 0795 Sternberg's Triarchic Theory

Analytical, creative and practical intelligences, only loosely related to each other. Loosely related is the claim — it is an argument against the single factor.

And several.

Psych 0797 Gardner's Multiple Intelligences

Several independent intelligences, including linguistic, spatial, musical and interpersonal. The furthest from a single factor of the four positions.

Three entries. Two capacities named on their own terms, and then the myth — which is worth stating carefully, because the thing being denied is narrower than people assume.

One capacity.

Psych 0799 Emotional Intelligence

The ability to perceive, understand and regulate emotion in oneself and others. Three operations, and the third is the one that is hardest to measure.

Two directions.

Psych 0801 Divergent and Convergent Thinking

Divergent thinking generates many possible solutions; convergent thinking narrows to the single best one. Most real work needs both, in that order.

And the myth.

Psych 0803 The Learning Styles Myth

Matching instruction to a student's preferred learning style does not improve learning outcomes. People do have preferences — what does not hold is that teaching to the preference produces better learning.

Three entries on the instruments. The quotient in the name is a historical artefact — it was originally an actual ratio, and it stopped being one, which is why the number needs the rest of this part to interpret.

Where the number came from.

Psych 0805 Mental Age and the Intelligence Quotient

Mental age is the age level of performance a person achieves, and the quotient originally expressed that as a ratio to chronological age. The name survived the arithmetic that produced it.

The first test.

Psych 0807 The Stanford-Binet

A standardized test producing an overall score and scores on separate cognitive factors. Both outputs matter, and only one of them gets quoted.

The second.

Psych 0809 The Wechsler Scales

Standardized tests reporting verbal and performance indices as well as a full-scale score. The profile across indices carries information the single number does not.

Four entries on the properties any measure has to have. Uniform administration, a reference distribution built from a real sample, and then the two properties everything is judged on — which are independent of each other, as the next topic makes uncomfortable.

Uniformity.

Psych 0811 Standardization

Administering and scoring a test in a uniform way, and establishing scores from a representative sample. Representative is the load-bearing word — a score means what the sample makes it mean.

The reference.

Psych 0813 Norms and the Normal Curve

Norms are the distribution of scores from the standardization sample, and intelligence scores are scaled to a normal curve. The score is a position in a distribution, not a quantity of anything.

The first property.

Psych 0815 Reliability

The consistency of a test's results across occasions, forms, or scorers. Three kinds of consistency, and a test can have one without the others.

The second.

Psych 0816 Validity

The extent to which a test measures what it claims to measure. A separate question from consistency, and a much harder one to answer.

Two entries, and this pair is the sharpest thing in the unit. A measure can be highly reliable and completely invalid — and the second entry explains why that specific combination survives scrutiny that a plainly bad measure would not.

The combination.

Psych 0817 Reliable but Not Valid

A measure can be highly reliable while being invalid, producing consistent results that do not reflect the intended construct. It gives the same wrong answer every time, which looks exactly like precision.

And why it survives.

Psych 0818 Why Consistency Persuades

Consistency is easier to demonstrate than validity and more persuasive to observers — which is what makes a reliable invalid measure dangerous rather than merely mistaken.

Five entries on what the scores do across groups and over time. One of them is a rise across an entire century that no theory of fixed capacity comfortably explains — and one is a performance effect produced by nothing but what a person fears being taken to confirm.

Across a century.

Psych 0820 The Flynn Effect

A substantial rise in average intelligence test scores across the twentieth century in many countries. Large, sustained, and too fast for any genetic account.

One end.

Psych 0822 Intellectual Disability

Significant limitations in both intellectual functioning and adaptive behavior, with onset during development. Both criteria, and the second one keeps the definition from resting on a test score alone.

The other.

Psych 0824 Giftedness

Generally defined by scores at the far upper end of the distribution, often with additional criteria. Generally and often — the definition is less settled than the other end.

An effect on performance.

Psych 0826 Stereotype Threat

A reduction in performance when a person fears confirming a negative stereotype about their group. Nothing about the person's capacity changed — the testing situation did.

And the two influences.

Psych 0828 Heredity and Environment in Intelligence

Scores show substantial heritability within populations and remain strongly responsive to environment. Both at once — and Unit Two's warning about reading heritability applies directly here.

Two entries closing the unit. A procedure that reads a test report in a deliberate order, with the score last — because four other things determine what the score means. Then the capstone, which returns to Part One's material.

Watch a report get read properly.

Psych 0830 Reading an Intelligence Test Report

Checking the standardization sample, the index profile, the reliability interval and the validity claim before reading the score. Everything in this part, arranged as an order of operations.

The capstone. One problem, four ways.

Psych 0832 Solving One Problem Four Ways

Running one problem through an algorithm, a heuristic, working backwards and divergent generation, and accounting for the differences. Part One's strategies, put in direct competition.

Five things you can now do, and Unit Seven is complete. You can say what each theory of intelligence claims, and that they disagree about whether there is one capacity or several — from a single general factor to several independent ones. You can say why the learning styles claim is a myth: preferences are real, and matching instruction to them does not improve outcomes. You can say what standardization establishes — uniform administration and scores from a representative sample — and that a score is a position in a distribution. You can say why a reliable measure can be invalid, and why that combination is dangerous: it gives the same wrong answer every time, and consistency is easier to show than validity and more persuasive. And you can read a test report in order — sample, profile, reliability interval, validity claim, and only then the score.