Short answer
There is no single accuracy rating for online IQ tests. The question is whether a particular score interpretation is supported for a particular use. Online delivery alone does not settle that question; neither does adaptive selection, a model name or a displayed interval.
What to look for
The Testing Standards distinguish evidence for validity, score reliability/precision and fairness. Ask for the reference sample, evidence from repeated or independent measurements, comparisons relevant to the claimed use, and limitations. A reliable result can still support the wrong interpretation.
- Are the items and scoring rules described?
- Were item parameters estimated from appropriate data?
- Who was included in the norm sample?
- Does the evidence apply to this version and intended population?
- Are uncertainty and unsupported uses made clear?
AurorIQ's current status
AurorIQ now selects 25–35 items from a 120-item reviewed bank using a content-balanced, bounded Bayesian/EAP-like routing model. Difficulty and discrimination values are developer-seeded, not empirically calibrated. The reasoning index is proprietary; there are no representative population or age norms.
The engine retains posterior spread as an internal uncertainty signal. It does not include every source of error, and its empirical coverage is unknown. No numeric confidence interval or population percentile is displayed. A precise-looking number would not resolve the missing validation evidence.
Why a retake can differ
Answers, guessing, interruptions and familiarity can differ between sessions. A meta-analysis of retesting found practice-related score gains across cognitive ability tests, with results depending on study conditions. That does not quantify expected change on AurorIQ. Reusing the same small bank makes item familiarity especially relevant.
How to use the result
Take the exercise when you can read the instructions and focus without interruption. Record unusual conditions, and do not interpret a higher retake as proof that intelligence increased. Start with raw answers and a practical learning goal. AurorIQ has not been validated for diagnosis, educational placement, hiring, legal decisions or credentials.
Inspect the scoring method and validation status. If a result appears inconsistent with your answers, report the discrepancy rather than treating the score as authoritative.
References
- AERA, APA & NCME (2014). Standards for Educational and Psychological Testing. American Educational Research Association.
- Hausknecht, J. P., et al. (2007). Retesting in Selection: A Meta-Analysis of Coaching and Practice Effects for Tests of Cognitive Ability. Journal of Applied Psychology, 92, 373–385.
These sources support the specific research statements linked above. They do not establish reliability, norms or validity for AurorIQ.