The Flynn Effect: Why IQ Scores Need Resetting

Picture a class graded on a curve. Every decade the students get a little stronger, and the teacher keeps moving the average back to a C. A student who gave the same answers in 1950 and in 2000 would get a lower grade the second time. IQ scores work much the same way, and that is why the tests keep getting reset.

A score of 100 only means "average compared with the group the test was set against." Change the group, and the same performance earns a different number.

What the Flynn effect describes

The Flynn effect is a large, long-running rise in intelligence test scores across many parts of the world during the 20th century. It is named for researcher James Flynn (1934 to 2020). The climb was not a single jump. Scores rose continuously and roughly in a straight line from the earliest years of testing to the present.

The gains showed up in many countries with long testing traditions, including other Western European countries, Japan and South Korea. That breadth matters. It points to a pattern shared across populations rather than a quirk of one national sample.

Some researchers have also suggested a mild reversal in developed countries beginning in the 1990s, sometimes called the reverse Flynn effect. Read that as a proposal and nothing firmer. The sources I used do not detail how it was measured in specific recent groups, and they do not describe the research methods behind the national gain figures.

Why tests have to be renormed

IQ scores are relative. When a test is standardized, the average is set to 100 with a standard deviation of 15 or 16 points. Your score tells you where you stand against that norm sample and nothing more.

Now add rising performance. When a newer group takes an older test, its average lands well above 100. At that point the number 100 no longer marks the middle of the population. It marks the middle of people who were tested years ago.

Publishers respond by renorming. One report describes resetting the mean to 100, which makes the test effectively harder, every 15 to 20 years. That report is a ScienceDaily summary from 2003, so the schedule it gives reflects practice at that time. Check current publisher practice before assuming it still holds.

Britain gives a concrete case. Children's average scores on Raven's Progressive Matrices rose 14 IQ points between 1942 and 2008. The usual summary is about three points per decade, or roughly thirty over a century. Use that as a handy rule of thumb. It is not an exact law.

What the gains are made of, and what might explain them

The gains are uneven. The biggest are on fluid, abstract-reasoning tasks such as Raven's matrices. The smallest are on crystallized knowledge such as vocabulary. So a single number can hide very different stories underneath. A child can score high on pattern puzzles and average on word knowledge, and the total blurs the two.

As for causes, the list of candidates is long: education, test familiarity, nutrition, a more stimulating environment, less infectious disease, heterosis, and less lead in gasoline. These are proposals. The sources do not pin down the exact causes, and no one factor is shown to account for the whole rise.

One source describes the effect as environmental rather than genetic. That is a single source's framing, so hold it loosely. Another note points a different way: cultural change may make parts of intelligence tests obsolete, which could explain some apparent reversals. If test content ages, a falling score might say something about the test and not only about the people taking it.

Why renorming has practical consequences

The same raw performance earns a lower score under newer norms. That means scores from different eras cannot be compared directly, and tests must be restandardized over time. A score from 1995 and a score from 2015 may look like neighbors on paper while measuring against different yardsticks.

The effects reach real children. A study of nearly 9,000 school psychologist assessments from nine U.S. school districts, covering students aged 6 to 17, found that students in the borderline and mild mental retardation range lost an average of 5.6 IQ points when retested on a renormed test. The phrase "mental retardation" is outdated and is no longer used in current practice. It appears here because the 2003 source used it.

The WISC shows the same thing. After it was renormed, the number of children with IQs of 66 to 70 recommended for special school programs for mild mental retardation tripled in the first five years of the new test, compared with the last five years of the old one. That figure covers one narrow score band, so do not stretch it to all children. Still, the lesson is plain. Children with the same ability can be classified differently depending on which norms were used, and a shift of a few points can decide whether a child is offered services.

How to read scores and what to ask

First, check which norm version a test used and when it was standardized before comparing scores across eras. The sources do not explain how to convert between versions, so this is about awareness, not arithmetic.

Second, if a score sits near a diagnostic cutoff, the norms matter a lot. A retest on a renormed test can lower it. A parent or teacher can ask plainly:

• Which norms scored this test, and when were they set?
• Would the child's classification change under the newer norms?
• Does the score reflect fluid reasoning or crystallized knowledge?

Two red flags are worth keeping in mind. Comparing raw scores across decades without accounting for norm updates can mislead. And treating renorming as a neutral technical step ignores its reported effects on special education placement.

One last caution. The sources do not cover current diagnostic criteria or the legal rules for special education eligibility. For those, you need the school or a qualified psychologist, and a rule of thumb will not do.