What Is the Margin of Error on an IQ Score?
The IQ Town Team
Psychometrics & test design
The margin of error on a full IQ score is usually about 3 to 5 points either way. A reported 120 often means 'somewhere in the mid-teens to mid-120s, and we are reasonably sure.' Treating 120 and 124 as different people is a mistake.
The band around the number
Psychometricians report a confidence interval. A common one is 95%: if you could test the person many times without practice effects, about 95 of 100 scores would fall in that band. On a good full-scale IQ, the band is tight — often on the order of 5 points either side, sometimes tighter. On a 15-minute test it is looser. How IQ is measured is the scoring story; this is the humility that belongs next to it.
Why a precise test is still fuzzy
Items are a sample of all the items that could have been written. Attention flickers. A lucky guess and a missed easy item both happen. The test is reliable enough that your rank will not swing from average to gifted. It is not reliable enough to rank you against someone four points away. Two different tests add another layer, because they are not the same sample of items.
One item is a sample. So is one score.
Try one example before you start the full test. Instant feedback — no account required.
All blips are blops. Some blops are blups. Which conclusion must be true?
Read your result as a band
The free test gives an estimate. Hold it loosely, the way a confidence interval asks you to.
How to use the band
If a cutoff is 130 and your interval is 124–134, you are not 'not gifted.' You are near a line, and the line is administrative. If you retest and move seven points, check practice effects before you announce growth. Take the free test as an estimate with an even wider unspoken band, because it is shorter than a WAIS.
Say it as a range
'Around 115, give or take 5' is the adult sentence. 'I am a 115' is the sentence that will pick a fight with the next test.Frequently asked questions
What is the margin of error for an IQ test?
On a full modern scale, a 95% confidence interval is often about 3 to 5 points either side of the score. Shorter tests are less precise.
Is a 5-point difference meaningful?
Usually no. It sits inside ordinary error. Look for gaps that clear the interval, or for a pattern across subtests, before calling it a change.
Are very high scores less accurate?
Yes. Fewer people and fewer hard items mean the top of the scale is fuzzier than the middle.
Get the estimate, then give it room
About 10 minutes, unassisted.