How the intelligence profile is measured

Measured, not guessed.

Four timed tasks on your phone. Compared with AINIQ users, error margin shown. Figures refresh every night.

What is measured

Pattern reasoning
Gf · inductive reasoning
3 × 3 matrix, missing cell, 8 options
Series
Gf · serial reasoning
Next member of a series, 5 options
Figure rotation
Gv · visualisation
The same object turned, 4 options
Digits backward
Gwm · working memory
A digit sequence backwards, length 2–8

Pattern reasoning: Carpenter rules (constant, progression, distribution, combination, subtraction). Seven distractors, each breaking one rule.
Series: arithmetic step, alternation, period and letter–number interleaving.
Figure rotation: objects of 7–10 cubes. Wrong options are mirror images or other objects; a mirror image is never correct.
Digits: random digits without repeats or long runs.
Every generator checks that exactly one option is correct. Tasks are drawn by difficulty level, the same task is never shown twice, and the correct answer never leaves the server.

The comparison group is AINIQ users, not the population. Bands from 100 users, percentiles from 500. Error margin 90 %.

Correlations between modules
What is measured	Series	Figure rotation	Digits backward
Pattern reasoning	Loading…	Loading…	Loading…
Series		Loading…	Loading…
Figure rotation			Loading…

Loading…

Live figures are compared with these limits. Publication did not wait for them; the figures accumulate from use.

Criterion	Limit
Internal consistency per module	≥ 0.80 (pattern reasoning ≥ 0.85)
Overall reliability	≥ 0.90
Correlations between modules	0.30–0.60, all positive
Device difference (phone vs desktop)	\|d\| ≤ 0.20
Retest correlation	overall ≥ 0.80, modules ≥ 0.65
Task quality	point-biserial ≥ 0.15; hit rate 10–95 %

A percentile places you among AINIQ users, not in the population.
Below 100 runs a module’s error margin rests on a literature assumption (reliability 0.75).
The first hundred or so users per module also see unpruned tasks.
The result is not a diagnosis or an assessment of fitness for work or study, and it never appears in team views.

We say	We do not say
Measured, not guessed.	A single IQ number
Error margin shown.	“Accurate”
Method published, figures update.	“Scientifically proven” without own data
Compared with AINIQ users (n shown).	Compared with the population
Your Twin adapts its explanations to your profile.	“You improve your intelligence or memory”
For self-knowledge.	Diagnosis, hiring, aptitude