Here's a question that has started fistfights in psychology departments: is "smart" one thing or many things?
Think about someone you know who can ace any math test but can't read a room, or someone who can talk their way out of anything but freezes on a multiple-choice exam. Are those the same ability showing up differently, or genuinely different abilities? Your answer to that question is, secretly, a whole theory of intelligence — and psychologists have been arguing both sides for over a century.
And then it gets uncomfortable. Because once you build a test to measure this thing, the test starts making decisions about people: who gets into the gifted program, who gets the job, who gets labeled. A number that came out of a 90-minute test can follow someone for life. So this lesson is really two lessons braided together: what is intelligence, and can we measure it fairly — a question where the science and the stakes are both high. Let's measure your brain. Carefully.
In the early 1900s, Charles Spearman noticed something statistical: people who did well on one mental test tended to do well on all of them. Verbal, spatial, numerical — the scores clustered together. Using a technique called factor analysis (a statistical method that finds clusters of related items, identifying underlying "factors" that explain why scores correlate), Spearman argued that beneath all these specific abilities sits a single, general intelligence he called g (the g factor). On this view, g is the engine; specific talents are just g showing up in different domains.
Not everyone bought it. L.L. Thurstone ran his own factor analyses and concluded intelligence breaks into several independent primary mental abilities (such as verbal comprehension, numerical ability, spatial ability, and reasoning) rather than one master factor. The debate — one g or many abilities — never fully resolved, and most modern psychologists land somewhere in between: there's a real general factor, but also meaningfully distinct abilities riding on top of it.
The strongest pushback against g came from theories of multiple intelligences. Howard Gardner argued that intelligence isn't one number but several relatively independent intelligences — linguistic, logical-mathematical, spatial, musical, bodily-kinesthetic, interpersonal, intrapersonal, and naturalist. His evidence included people with brain damage who lose one ability while keeping others intact, and savant syndrome (individuals with limited general functioning but one spectacular skill, like lightning calendar calculation) — which is hard to explain if intelligence is a single thing.
Robert Sternberg agreed intelligence is more than g but organized it into a triarchic theory — three types:
A clean way to keep Sternberg straight: Analytical solves the problem on the test, Creative invents a problem nobody's seen, Practical handles the problem your boss just dumped on you.
Raymond Cattell split g itself into two flavors. Fluid intelligence (Gf) is your ability to reason, solve novel problems, and see patterns independent of prior knowledge — raw processing power. It tends to peak in early adulthood and decline with age. Crystallized intelligence (Gc) is your accumulated knowledge, vocabulary, and skills — everything you've learned. It tends to hold steady or even increase into old age.
Try This. Picture a 75-year-old retired English teacher and a 22-year-old competitive gamer. Who wins a crossword full of obscure vocabulary? Who wins a brand-new puzzle game with rules they've never seen? The teacher's crystallized intelligence dominates the crossword; the gamer's fluid intelligence dominates the novel puzzle. Same person can be "smarter" depending on which one the task demands.
Peter Salovey and John Mayer proposed emotional intelligence (EI) — the ability to perceive, understand, manage, and use emotions (your own and others'). It's not on a standard IQ test, and it predicts things IQ doesn't, like managing relationships and reading social situations. (Be careful: pop psychology inflated "EQ" into a cure-all. On the AP exam, EI is a specific four-part ability model, not "being a nice person.")
The whole testing enterprise started with a practical problem. In 1905 France, Alfred Binet and Théodore Simon were asked to identify schoolchildren who needed extra help. They built the first practical intelligence test around mental age — the level of performance typical for a given chronological age. A child who answered like an average 8-year-old had a mental age of 8, regardless of actual age. Binet was explicit that this was a tool for help, not a fixed measure of inborn worth — a caution later ignored.
Lewis Terman at Stanford revised Binet's test for American use, producing the Stanford-Binet. From William Stern's idea, the test reported an intelligence quotient (IQ) using the original formula:
IQ = (mental age ÷ chronological age) × 100
So an 8-year-old performing like a 10-year-old: (10 ÷ 8) × 100 = 125. An average child performs at their own age, giving exactly 100. This formula works for children but breaks for adults (a 60-year-old isn't twice as smart as a 30-year-old), so it's now historical.
David Wechsler fixed that. His tests — the WAIS (Wechsler Adult Intelligence Scale) and WISC (for children) — are the most widely used today. Instead of mental age, Wechsler used a deviation IQ: your score reflects how far you fall from the average of your own age group, expressed on a scale with a mean of 100 and a standard deviation of 15. Wechsler's tests also report separate scores (verbal, performance) rather than one number — a nod to Thurstone's multiple-abilities view.
A usable test needs three properties:
Standardization — administering and scoring the test the same way every time, then comparing each person's score against a pretested norm group. Standardized scores typically form a normal curve (the bell-shaped distribution): most people cluster near the mean (100), with fewer people toward the extremes. With a mean of 100 and SD of 15, about 68% of people fall between 85 and 115, and about 95% between 70 and 130.
Reliability — the consistency of a test. Does it give the same result each time? Checked by test-retest reliability (same people, two occasions — do scores match?) and split-half reliability (do scores on one half of the items match the other half?). A reliable test is repeatable.
Validity — whether the test measures what it claims to measure, and whether it predicts what it should. Types: content validity (do the items cover the actual domain?), predictive validity (does the score forecast future performance, like SAT predicting college grades?), and construct validity (does it measure the abstract trait it's supposed to?).
The one-sentence distinction the AP exam loves: reliability is consistency; validity is accuracy. A bathroom scale that reads 10 pounds too high every time is perfectly reliable (same answer every time) but invalid (wrong number). A test can be reliable without being valid — but it cannot be valid without being reliable.
Scores at the tails of the curve get labels. Intellectual disability is diagnosed not by a low IQ alone (historically around 70 or below) but also by significant limitations in adaptive behavior — everyday conceptual, social, and practical skills. The IQ number is never enough by itself. At the high tail, giftedness describes exceptionally high ability; Terman famously launched a longitudinal study of gifted children (see Spotlight).
Average test scores sometimes differ across groups, and how you interpret that gap matters enormously. The scientific consensus emphasizes environmental factors — unequal access to education, nutrition, health care, and enrichment — not innate group differences. Two specific concepts are essential:
Test bias — a test is biased if it measures something other than the intended ability for some groups, for instance by relying on vocabulary or cultural knowledge unfamiliar to a particular group. A biased item lowers scores for reasons unrelated to actual ability.
Stereotype threat — described by Claude Steele and Joshua Aronson (1995) — is the extra anxiety a person feels when they fear confirming a negative stereotype about their group, which can depress their performance on the very test where the stereotype is "tested." The threat is in the situation, not the person.
Two more must-knows. The Flynn effect is the observation (named for James Flynn) that average raw IQ scores have risen substantially across generations worldwide — strong evidence that environment (better schooling, nutrition, test familiarity) shapes scores, since genes can't change that fast. And heritability of intelligence — the proportion of variation within a group attributable to genes — is real but is regularly misused: heritability within a group says nothing about why two different groups (raised in different environments) differ on average. Keeping those two ideas separate is the single most important reasoning move in this section.
Steele & Aronson (1995) — Stereotype Threat.
Who & when: Claude Steele and Joshua Aronson, 1995.
Method: Black and White college students took a difficult verbal test (items drawn from the GRE) under different framings. In one condition, the test was described as diagnostic of intelligence/ability; in another, it was described as a non-diagnostic laboratory problem-solving task unrelated to ability. The researchers compared performance across race and framing, controlling for prior SAT scores so groups were academically matched going in.
Finding: When the test was framed as diagnostic of ability, Black students performed worse than matched White students. But when the same test was framed as non-diagnostic — removing the relevance of any racial stereotype — that performance gap shrank dramatically or disappeared. Simply changing the description of the test, not the test itself, changed the scores.
Significance: This is landmark evidence that situational psychological factors — the anxiety of potentially confirming a negative stereotype — can suppress measured performance. It reframes part of the test-score gap as a product of the testing situation rather than fixed ability, and it gives a concrete, manipulable mechanism (the framing) that environmental explanations of group differences had previously lacked. For the AP exam, Steele & Aronson = stereotype threat, 1995, framing manipulation, and "the threat lives in the situation."
Scenario 1. A company gives every applicant the exact same instructions, the same time limit, and scores answers with an identical answer key, then compares each applicant to a large pretested pool of past applicants.
Which test-construction concept is this? This is standardization: uniform administration and scoring, plus comparison against a norm group. (Note what it is not: standardization alone says nothing about whether the test is valid — whether it actually predicts job performance.)
Scenario 2. A new "creativity quotient" test gives Maria a score of 142 on Monday and 141 the next Monday — rock-solid consistency. But her score turns out to have zero relationship to whether she actually produces creative work.
What does this test have, and what does it lack? It has reliability (consistent across time — strong test-retest reliability) but lacks validity, specifically predictive validity — it doesn't forecast the real-world outcome it claims to measure. The reliable-but-invalid scale: same answer every time, wrong answer.
Scenario 3. Before a hard math exam, the proctor announces, "Historically, this exam shows a big gender gap." Several students who belong to the negatively stereotyped group report their minds going blank, and they underperform relative to their usual grades.
Which concept explains the drop? Stereotype threat (Steele & Aronson). The announcement made a negative stereotype relevant to the situation, and the anxiety of possibly confirming it consumed mental resources and depressed performance — even though nothing about the students' actual ability changed.
Scenario 4. A grandmother destroys her grandkids at crossword puzzles and Trivial Pursuit but takes much longer than they do to figure out the rules of a brand-new board game.
Which distinction does this illustrate? Crystallized vs. fluid intelligence (Cattell). Her accumulated vocabulary and knowledge (crystallized, which holds with age) wins the word and trivia games; the grandkids' faster novel-problem reasoning (fluid, which peaks young) wins the unfamiliar game.
Reliability vs. Validity. The big one. Reliability = consistency (same result repeatedly); validity = accuracy (measures and predicts what it claims). Mnemonic: a broken clock is reliable twice a day — reliably wrong. A test can be reliable but invalid; it can't be valid without being reliable. If a question stresses "same score every time," think reliability; if it stresses "measures the right thing / predicts the outcome," think validity.
Fluid vs. Crystallized. Students flip them. Fluid = on-the-fly reasoning about new problems, peaks young, declines with age (flows away). Crystallized = stored knowledge, holds/grows with age (crystals accumulate). Link "fluid" to "novel/fast" and "crystallized" to "knowledge/experience."
g vs. Multiple Intelligences. Spearman's g says intelligence is fundamentally one general factor. Gardner and Sternberg say it's several relatively independent intelligences. Don't blend them into "intelligence is one thing with parts" — they're competing claims. Spearman = one engine; Gardner/Sternberg = several engines.
Standardization vs. "normalization" / norm group. Standardization means giving and scoring the test uniformly and comparing to a norm group; it's a procedure. It does not mean "making scores form a normal curve" by manipulation, and it does not mean the test is "normal" or fair. Standardized ≠ unbiased.
Four-choice MCQs in current AP format. Answers and explanations in section (h).
EBQ format. Using the three sources below, write a response that makes a defensible claim, supports it with evidence from at least two sources, and explains your reasoning using course content. You will be scored on: Claim (0–1), Evidence from 2+ sources (0–3), and Reasoning & Application of course content (0–2), for a total of 7 points.
Prompt: Using the sources, develop an argument about the following question: To what extent do situational and environmental factors — rather than fixed inborn ability — shape performance on intelligence and standardized tests?
Source A
Steele & Aronson (1995). Black and White college students, matched on prior SAT scores, completed a difficult verbal test drawn from the GRE. In one condition the test was described as diagnostic of intellectual ability; in another it was described as a non-diagnostic problem-solving task. Finding: In the diagnostic condition, Black students scored significantly lower than matched White students; in the non-diagnostic condition, that gap shrank substantially or disappeared. Changing only the description of the test changed performance.
Source B
Flynn (cross-national archival analysis). The researcher compiled standardized IQ test results from many countries across several decades, comparing each generation's performance against the original norming samples. Finding: Average raw IQ scores rose by roughly 3 points per decade over the 20th century, with the largest gains on tests of abstract, fluid reasoning. The gains occurred far too rapidly to be explained by genetic change in the population.
Source C
Intervention/enrichment study (summarized). Children from under-resourced backgrounds were randomly assigned either to an intensive early-childhood educational enrichment program (providing nutrition, cognitive stimulation, and language-rich instruction) or to a no-program control group. Finding: Children in the enrichment program showed meaningfully higher average IQ scores than control children when tested years later, though some of the advantage narrowed over time. Random assignment makes a causal interpretation of the program's effect reasonable.
Question: Develop an argument about the extent to which situational and environmental factors shape test performance. Use evidence from at least two of the sources and explain your reasoning with psychological concepts from this unit.
Claim. Situational and environmental factors substantially shape performance on intelligence and standardized tests, meaning such scores reflect more than fixed inborn ability. (Claim — 1 pt)
Evidence. Source A supports this: when a difficult verbal test was described as diagnostic of ability, Black students matched on prior SAT scores performed worse than White students, but when the same test was framed as a non-diagnostic problem-solving task, that gap shrank dramatically or disappeared — so performance changed with the situation, not the test content. Source C adds converging evidence: children randomly assigned to an intensive early-childhood enrichment program later scored meaningfully higher on IQ tests than control children, and because assignment was random, the environment can be interpreted as causing the score difference. Source B further shows that average raw IQ scores rose about 3 points per decade across the 20th century — far too fast for genetic change — pointing to environmental drivers like improved schooling and nutrition. (Evidence — 3 pts: specific findings used from three sources; two would suffice for full credit)
Reasoning & Application. These findings cohere through course concepts. Source A is a textbook demonstration of stereotype threat: the anxiety of potentially confirming a negative stereotype consumed cognitive resources and depressed measured performance, showing the score gap was partly a feature of the testing situation rather than ability. Source B is the Flynn effect, which undercuts a purely genetic account because genes cannot change across a few generations; rising scores must reflect environmental change. Critically, even if intelligence shows high heritability within a group, heritability describes variation inside a group and cannot explain average differences between groups raised in different environments — so Sources A and C, which manipulate the situation and environment directly, carry more weight for the between-group question. Together the sources support a strongly environmental interpretation while still leaving room for ability to matter. (Reasoning & Application — 2 pts: applies stereotype threat, Flynn effect, and the heritability-within-vs-between distinction)
1. (B). Spearman's factor analysis revealed that mental-test scores positively correlate, which he attributed to a single general factor, g. (A) is Gardner; (C) is Cattell; (D) is Salovey & Mayer.
2. (B) High reliability, low validity. Consistent repeated scores = reliability; failure to measure/predict the intended outcome = low validity. The reliable-but-invalid case (the scale that's always wrong by the same amount). (A) reverses the terms; (C) and (D) are contradicted by "consistent."
3. (C) 125. IQ = (MA ÷ CA) × 100 = (10 ÷ 8) × 100 = 125. (A) inverts the ratio; (B) would require MA = CA; (D) is too high.
4. (B) Sternberg. The triarchic theory = analytical, creative, practical. (A) Gardner = eight multiple intelligences; (C) Spearman = g; (D) Cattell = fluid/crystallized.
5. (C) Crystallized and fluid intelligence. Accumulated knowledge that holds with age = crystallized; speed on novel problems that peaks young = fluid. (A), (B), and (D) describe unrelated constructs.
6. (C). Wechsler introduced the deviation IQ, scoring people relative to their own age group (mean 100, SD 15), which works for adults. (A) is the old Stanford-Binet formula it replaced; (B) is false (Wechsler reported subscores); (D) is the opposite — his tests are highly standardized.
7. (C) 68%. On a normal curve, about 68% fall within one SD of the mean — here 85 to 115 (100 ± 15). (A) 34% is half of that band; (D) 95% is two SDs (70–130).
8. (B) Stereotype threat. Framing the test as ability-diagnostic and making the group stereotype salient depressed performance for matched students — the situation, not ability, changed. (A) is contradicted (scores were systematically lower, not random); (C) Flynn effect is across generations; (D) is the biological-determinist trap the data refute.
9. (B). Steele & Aronson manipulated whether the test was described as diagnostic of ability and watched performance change accordingly. (A) is heritability research, not this study; (C) describes test bias, a different concept; (D) is Terman's gifted study.
10. (B). The Flynn effect = rising average raw IQ performance across generations. (A), (C), and (D) describe other phenomena; the Flynn effect specifically is the generational rise.
11. (B) Test bias. The test measures culture-specific vocabulary rather than the intended reasoning ability for that group, systematically distorting their scores. (A) reliability is about consistency, not fairness; (C) stereotype threat is a situational anxiety effect, not a property of the items; (D) is a scoring method.
12. (B). On a normal curve, ~95% fall within two SDs (70–130) and ~68% within one SD (85–115). (A) swaps the two; (C) is false (about 2.5% score above 130); (D) is false (only ~2.5% score below 70).
13. (A). Tightly clustered scores along the diagonal across two administrations indicate strong test-retest reliability (consistency); a random scatter indicates poor reliability. (B) the data show nothing about validity/accuracy; (C) bias isn't shown by retest consistency; (D) is unsupported.
14. (B) Significant limitations in adaptive behavior. Diagnosis requires both a low IQ and deficits in everyday adaptive skills — the number alone is insufficient. (A) describes giftedness; (C) and (D) are unrelated.
15. (B). Heritability describes variation within a group and cannot be extrapolated to explain average differences between groups raised in different environments — the central reasoning safeguard of this unit. (A) and (C) overcorrect (both genes and environment matter); (D) misuses the Flynn effect, which actually supports environmental influence.
### EBQ Rubric (7 points)
| Component | Points | Criteria |
|---|---|---|
| Claim | 0–1 | 1 pt for a defensible, arguable claim that takes a position on the extent of environmental/situational influence. No point for restating the prompt or a non-directional statement. |
| Evidence | 0–3 | Up to 3 pts for accurately using specific findings from the sources to support the claim. Must use at least two sources for full credit; citing only one source caps this at a lower score. A source letter without its finding earns nothing. |
| Reasoning & Application | 0–2 | Up to 2 pts for connecting the evidence to the claim and applying course concepts (e.g., stereotype threat, Flynn effect, heritability within- vs. between-group, test bias, validity). Labels without explanation, or a biological-determinist leap, do not earn the points. |
| Total | /7 | |
---
PsyIQ · Lesson 13 of 30 · Unit 2: Cognition. This is an EBQ lesson (3-source evidence-based question, 7-point rubric). MCQ and EBQ practice modeled on the redesigned (2025+) AP Psychology exam. Not affiliated with the College Board. AP is a registered trademark of the College Board. Content pending external psychology QC.