Your IQ Is a Range, Not a Number: What Confidence Intervals Really Mean

What Is a Confidence Interval?

Every measurement tool has a margin of error. A bathroom scale that reads 155 pounds might be off by a pound or two in either direction. IQ tests work the same way, except the margin of error has a formal name: Standard Error of Measurement (SEM).

For major IQ tests like the WAIS-IV and Stanford-Binet, the typical SEM falls between 3 and 5 points. That means if you score 115, statisticians are 68% confident your "true" score lies between 110 and 120 (one SEM in each direction). At the 95% confidence level — the standard used in most psychological research — that range expands to roughly 105 to 125.

Ten to fifteen points of uncertainty. That single number on your score report is actually a band wide enough to span the difference between "above average" and "superior" on the standard classification scale.

This is not a flaw in the testing. It is a fundamental property of all psychological measurement. The validity and reliability of a test determines how tight that band is, but no test reduces it to zero.

Here is the math in plain terms. The SEM is calculated from two properties: the test's standard deviation (15 points for most IQ tests) and its reliability coefficient. A test with a reliability of .95 has an SEM of about 3.4 points. A test with a reliability of .85 has an SEM of about 5.8 points. The higher the reliability, the tighter the band — but it never vanishes entirely.

WAIS-IV subtest reliability coefficients illustrate this range. Full Scale IQ reliability sits at .96, giving a tight SEM of about 3 points. But individual subtest reliabilities range from .74 to .96, meaning some cognitive dimensions have substantially wider confidence intervals than others. Your Vocabulary score is measured more precisely than your Processing Speed score. Understanding this prevents misinterpreting small subtest differences as meaningful gaps.

Why Your Score Moves

If your "true" IQ is fixed, why do scores shift between testings? The answer involves both the measurement tool and the person being measured.

The Breit et al. 2024 meta-analysis deserves attention. A test-retest correlation of .76 is often cited as evidence that IQ is "stable." And at the group level, it is. But that .76 means roughly 42% of score variance is unexplained from one sitting to the next. For any individual, that leaves substantial room for scores to move — and that movement is not random noise you should ignore. It reflects real factors affecting real performance.

Fatigue is the most underappreciated factor. A full WAIS-IV administration takes 60 to 90 minutes. Your performance in the final 20 minutes is not the same as your performance in the first 20 minutes. Morning people score differently at 8 AM than at 4 PM. If you took a clinical versus online test under different conditions, the scores are not directly comparable without accounting for context.

Test anxiety creates a paradox: the more you care about the result, the more likely anxiety is to suppress your performance. Someone taking an IQ test casually will often score higher than the same person taking it when their career depends on the outcome. This is not because the casual version "inflated" their score. It is because anxiety consumed working memory resources that would otherwise go toward the test itself.

The testing environment matters more than most people realize. Room temperature, ambient noise, the comfort of your chair, whether you ate recently — all of these affect cognitive performance measurably. A 2-3 point swing from environmental factors alone is well documented.

The van Hoogdalem and Bosman 2024 finding reinforces this: IQ test validity holds robustly at the group level but breaks down for individual-level prediction. Your score is a useful data point, not a precise measurement of a fixed quantity. Treating it as the latter leads to exactly the kind of anxiety Bridget experienced — agonizing over whether you are "really" a 114 or a 121 when the honest answer is: you are somewhere in a range that comfortably includes both.

The Flynn Effect: When Norms Expire

There is another source of score variation that has nothing to do with you personally: the age of the test's norms.

The Flynn Effect — named after researcher James Flynn — documents a consistent rise in raw IQ test performance of approximately 3 points per decade on modern Wechsler and Stanford-Binet tests (Trahan et al. 2014 meta-analysis). This means the average person today performs better on old IQ tests than the average person did when those tests were normed.

This creates a practical problem. If you took a test normed in 2005 and scored 118, that number reflects your performance relative to the population as it was measured in 2005. The population has since shifted upward. Your score relative to 2025 norms might be closer to 112. The six-point difference is not trivial — it could change your percentile ranking significantly.

Some common myths about IQ testing stem from confusion about norming dates. People retake a test on a newer edition and believe their intelligence declined, when the test simply moved the goalposts.

Sackett et al. 2022 revealed a related precision problem in the research itself. For decades, the correlation between general mental ability and job performance was cited as r = .51. After correcting for statistical artifacts, the actual correlation dropped to r = .31. The research community had been overestimating precision. If the scientists studying IQ can be fooled by measurement artifacts, individual test-takers should be even more cautious about treating their score as an exact figure.

What Your Range Actually Means for Your Career

This is where confidence intervals stop being abstract and start mattering for your decisions. Linda Gottfredson's research on occupational intelligence thresholds demonstrates that IQ requirements for jobs represent floors, not ceilings. You need to be above a certain level to succeed in a role, but being further above that threshold offers diminishing returns.

The more important finding: the middle 50% of workers in any given occupation spans 15-20 IQ points. Accountants range from roughly 105 to 125. Software engineers range from roughly 110 to 130. Physicians range from roughly 115 to 135. The spread within any profession is wider than most people's confidence intervals.

This means that if your confidence interval overlaps with the typical range for a career you are considering, you have the cognitive capacity to succeed there. Obsessing over whether you are at the bottom or middle of that range is wasted energy. A score of 125 and what that range means opens different doors than a score of 100, but a 115 versus a 120 rarely changes the practical picture.

With 76% of organizations now using skills-based assessments in hiring — roughly half of which include cognitive ability tests (TestGorilla 2025), understanding what your score range actually means has practical stakes. If you are screened out because your score was 108 on a day you were fatigued and anxious, and your true range extends to 118, that is not you failing — that is the measurement capturing the wrong slice of your ability. Understanding this distinction helps you advocate for yourself and strategize around cognitive assessments in hiring.

How to Read Your Score Report

A good score report should give you more than a single number. Here is what to look for and how to interpret it.

Your IQ subscores and what they mean for your career path matter more than your composite number. A person with strong verbal reasoning and weaker processing speed has a fundamentally different cognitive profile than someone with the reverse pattern — even if they share the same Full Scale IQ. The confidence interval around each subscore helps you distinguish real cognitive strengths from measurement noise.

When comparing scores across testings, account for the confidence interval overlap. If your first score was 114 (CI: 109-119) and your second was 121 (CI: 116-126), those intervals overlap at 116-119. The scores are not statistically different. You did not "get smarter." You measured the same underlying ability from two different angles and got results that are consistent within the margin of error.

This is exactly what happened with Bridget. Her two scores looked different — a seven-point gap feels significant when you are staring at a report. But once she understood that both scores sat comfortably within the same confidence band, the anxiety dissolved. She stopped asking "which number is real?" and started asking the more useful question: "What does this range tell me about which careers fit my cognitive profile?"

That shift in framing — from fixating on a digit to understanding a range — is the difference between using your score as a source of anxiety and using it as a decision-making tool.

When Retesting Makes Sense

Understanding confidence intervals also clarifies when a retest is worth your time and when it is not.

If your two scores fall within the same confidence interval, retesting will not tell you anything new. You are sampling from the same distribution. The third number will land somewhere in the same range, and you will have three data points all saying the same thing.

Retesting makes sense when you have reason to believe your initial score was depressed by situational factors — severe illness, extreme anxiety, sleep deprivation, an unfamiliar testing environment. In those cases, a retest under better conditions can give you a more representative sample of your ability. Our guide on when and how to retake an IQ test covers practice effects and optimal timing in detail.

Retesting also makes sense if you took a test with outdated norms and want a score benchmarked against a current population. A more recently normed test will give you a more accurate picture of where you stand today.

But if you are retesting because you want a higher number — because 114 does not feel as good as 121 — that is the single-number fixation talking. Your range is your range. A third test will not change your underlying cognitive ability. It will just give you one more data point scattered within the same band.

The Permission to Act

The deepest value of understanding confidence intervals is not statistical. It is psychological.

When you believe your IQ is a single, precise number, every decision feels fragile. Am I smart enough for this career? What if my real score is at the bottom of my range? The number becomes a gatekeeper, and you need it to be exactly right before you will move forward.

When you understand your IQ is a range, the question shifts. Instead of "Am I exactly smart enough?" you ask "Does my range overlap with what this career requires?" And for most career decisions, the answer is clearer than you expected. The scoring methodology behind modern IQ tests was designed to inform decisions at this level — ranges and thresholds, not to the single point.

Bridget did not need to know whether she was a 114 or a 121. She needed to know whether her cognitive profile — somewhere in the 107-129 range at the 95% confidence level — was compatible with data analytics. It was, by a comfortable margin. She started the career transition that same month.

Your score is a useful tool. Your range is even more useful. And neither one defines the upper limit of what you can achieve — it just tells you where the cognitive floor sits. Everything above that floor is effort, strategy, and choice.