I’m curious about the actual mechanics behind IQ testing. How do they turn your performance on various tasks into a single number? What’s the methodology?
Is it just about getting questions right or wrong, or is there more to it? How do they account for age differences? And how do they decide what the “average” score should be?
Just trying to understand the science behind it.
@CloverL Regarding age adjustments: Modern tests use age-stratified norms. A 7-year-old and 40-year-old taking the same test would have their raw scores compared to different reference groups - other 7-year-olds and other 40-year-olds respectively. This allows meaningful comparison across the lifespan. The WAIS-IV, for example, has separate norm tables for each age band. Without this, older adults would systematically score lower due to processing speed declines, even if their crystallized knowledge remains intact.
@Juan_San So a kid scoring 120 and an adult scoring 120 means the same thing?
@Gabby In terms of percentile ranking, yes - both are at approximately the 91st percentile for their age group. However, the absolute cognitive abilities differ substantially. The adult would outperform the child on most tasks, but both score similarly relative to their age-matched peers. This is why IQ is considered developmentally stable - if you’re in the 90th percentile at age 8, you’ll likely remain around the 90th percentile at age 30, even though your absolute abilities have matured.
IQ tests were literally designed for eugenics and remain culturally biased toward Western educated populations. They measure arbitrary puzzle-solving skills and call it “intelligence” while ignoring other aspects of real human intelligence and various cultures. The method seems more on pretending that number means something universal about human capability.
And I thought people who engage in online IQ sites would at least research before they make comments. What you just said is not entirely accurate. Alfred Binet created the first IQ test to identify students who needed educational support. Yes, the tests were later misused by eugenicists, but that’s not why they were designed.
Also, modern IQ tests have been revised extensively to reduce cultural bias. Are they perfect? No. But dismissing decades of psychometric research as pretending ignores genuine predictive validity for things like academic performance.
Does this mean IQ scores aren’t really comparing absolute intelligence across all humans? They’re just like relative rankings within age-matched groups?
Yes, in a nutshell, this is an accurate summary. And the reason for this is because there is no other way to quantify intelligence fairly and accurately. The relativity is what makes these tests reliable.
Yes, an IQ score of 120 means essentially the same thing for both a child and an adult, as modern, professionally administered IQ tests are age-normalized. A score of 120 falls within the “superior” or “above average” range, placing that individual in approximately the top 10% of their age-matched population. The percentile ranking is consistent.
It is worth noting that subtests in IQ tests are often weighted to reflect the importance of certain abilities, with the final score calculated from a weighted average of subtest scores. The weighting can vary depending on the test and the specific index being calculated, but it is common for some subtests to contribute more to the overall score than others to better estimate the general intelligence factor (g).
Scoring questions as right or wrong is just the first step in turning responses into an IQ score. (Actually, some tasks are scored based on speed of response, complexity of response, and other characteristics, but that doesn’t change the overall picture.)
The scores for each item are combined to form an overall score for each subtest. These are then combined to form a total score. (The combining is not as simple as averaging or adding up scores, but the process isn’t that different.) The overall score is compared to the overall scores for the norm sample, which is a representative group of the population the test is designed for. That is then used to produce an IQ score where 100 = average performance for the norm sample, values above 100 are for people who score higher than average, and values below 100 are for examinees scoring lower than average.