The reference uses quality-checked first completions
The Japanese version of BrainTypeIQ does not create an IQ scale by simply forcing raw scores to have a mean of 100. More than 2,000 completed test records were screened for response patterns, testing history and repeat attempts, the items used, and the scoring conditions of the current nine-task version. The conversion tables were built from 841 eligible first-time completers.
| Validation feature | Result |
|---|---|
| Split method | Age-stratified 70% / 30% |
| Number of repetitions | 100 |
| Mean overall IQ in validation participants aged 18–54 | 103.14 |
| Standard deviation in validation participants aged 18–54 | 14.41 |
| Largest remaining difference among the means of four age groups | 3.77 points |
The Japanese-version reliability analysis uses the same quality-checked participants.
Norming turns raw performance into a comparable score
Norming places a person's raw performance on a scale defined by a reference group. Twenty correct answers do not have the same meaning when the task difficulty or comparison group changes. A reported IQ of 100, a top-10% result, or an IQ of 130 therefore needs both a stated comparison group and a method for converting raw performance.
For the Japanese version of BrainTypeIQ, the scoring sequence is:
- Score each of the nine tasks.
- Use the raw-score distributions from the reference sample to place each task on a common scale.
- When age information is available, apply the observed age trend for the relevant cognitive domain.
- Combine task scores into the five domains, GAI, CPI, and overall IQ.
The scoring method describes how task scores are combined into the reported composites. IQ test measurement quality explains how norms differ from reliability and validity.
The age adjustment uses the adult core sample
The age adjustment was estimated primarily from the 739 participants aged 18 to 54. It was applied only to domains in which the Japanese-version data showed an age trend.
- Gc, crystallized intelligence, receives no age adjustment.
- Gf, Gv, Gwm, and Gs reflect the age trends observed in the data.
- There is no individual adjustment for sex, education, or country of residence.
- No age adjustment is applied when age information is unavailable.
Results can also be displayed for people under 18 and over 54, but the age adjustment was designed mainly from the 18–54 group. Intelligence and aging covers the broader ways cognitive abilities can change with age.
Split-sample validation tested stability across age groups
For each age-stratified split, the task conversions and age model were built on the training portion and applied to the unused validation portion. The analysis then examined the overall-IQ distribution and differences among four age-group means. The largest age-group difference reported in the opening table is not measurement error around one person's score; individual precision is estimated separately as an SEM of about 4.0 points in the Japanese-version reliability results.
What this reference supports
The IQ scores and percentiles are referenced to people who completed the Japanese version of BrainTypeIQ online. They are not rankings from a random sample of the entire Japanese population or scores based on WAIS norms.
For English-language results, BrainTypeIQ currently applies a provisional scoring model and is building a dedicated English dataset.