These are the validation gaps standing between a promising tool and a credible one. Each study is self-contained and could be run by a single lab or graduate student with IRB access.
1
Convergent validity against established batteries
Do our tests measure what they claim to measure? The most direct way to find out is to administer our battery alongside gold-standard instruments (WAIS-V, D-KEFS, Conners CPT-3) to the same participants and correlate the results.
What you'd do: Recruit 80 to 100 adults, administer both batteries in counterbalanced order, and analyze correlations across matched subtests. We provide the digital battery, raw data export, and co-authorship.
What it would prove: That our browser-based implementations produce scores consistent with lab-administered equivalents. A correlation of r ≥ .60 across domains would establish the platform as a viable screening instrument.
2
Test-retest reliability
If someone takes the same tests two weeks apart, do they get similar scores? Without temporal stability, individual results are meaningless and the platform can't be used for tracking change over time.
What you'd do: Administer the full battery twice to 50+ participants at a 2 to 4 week interval. We handle recruitment infrastructure if you provide the participant pool. ICC analysis across all subtests.
What it would prove: That scores are stable enough to reflect real cognitive ability rather than noise. An ICC of ≥ .80 is the standard threshold for clinical utility.
3
Clinical discrimination for ADHD and dyscalculia
Can the platform distinguish between people with documented ADHD or dyscalculia and neurotypical controls? This is the study that would make Neuropsych useful as an actual screening tool.
What you'd do: Recruit three groups of 30+ participants each (ADHD-diagnosed, dyscalculia-diagnosed, and neurotypical controls). Administer the battery and run ROC analysis to determine screening thresholds.
What it would prove: Whether the platform can reliably flag individuals who should seek formal clinical evaluation. An AUC of ≥ .80 would support its use as a first-pass screening instrument.
