- Two people can share a mean RT with different scatter.
- Not from a website.
- Both inflate lapses.
- High-stakes labs have trimming rules.
- Often the early confusion fades.
- Flanker lite and number-Stroop lite.
Mean versus SD?
Two people can share a mean RT with different scatter. Labs sometimes analyse both. We show a hit count.
Is high variability ADHD?
Not from a website. ADHD research includes variability, among many findings. Do not self-diagnose from scatter.
Sleep and noise?
Both inflate lapses. A noisy café session is a context, not a trait.
Should I discard slow trials?
High-stakes labs have trimming rules. Our toys do not claim those rules. Do not invent a personal outlier policy to flatter a screenshot.
Does practice shrink variability?
Often the early confusion fades. That is still practice, not a new IQ.
What to try?
Flanker lite and number-Stroop lite. Notice that some trials feel clean and some do not. That feeling is not a clinical index.
❓ People also ask
What is construct validity?
Construct validity is the argument that a score reflects a named idea — working memory, not “being clever in general.” It is built from theory, patterns of correlations, and failed rival explanations. A 20-trial number-Stroop lite can feel like control. Feeling is not a validity coefficient. This site never reports a licensed IQ.
What is construct validity? →What is criterion validity?
Criterion validity asks whether a score lines up with something you care about that is not the test itself — a later grade, a supervisor rating, or a Wechsler index. Concurrent means measured at the same time; predictive means later. A visual 2-back lite has no published criterion table. It is not an employment screen and not an IQ.
What is criterion validity? →What is content validity?
Content validity is whether the item set covers the skill or knowledge you claimed. A spelling test with only the letter Q is a coverage failure. Shape-span lite samples a tiny visuospatial sequence. It does not cover “all spatial intelligence.” Experts, blueprints and reviews build content arguments. A weekend hackathon does not.
What is content validity? →What is a confidence interval?
A 95% confidence interval is a procedure: if you repeated the study under the same model, 95% of such intervals would cover the parameter. It is not the probability that this one interval magically contains the truth after you saw it. Licensed IQ manuals turn a standard error into a range around a scaled score. Digit-span lite does not. Treat any integer from this site as a toy count.
What is a confidence interval? →What is item response theory?
Item response theory (IRT) models each item: harder items need more of the trait for a 50% chance of success; discriminating items separate nearby people. Computerized adaptive tests pick the next item from that model. Pattern-matrices lite is a fixed original set. It does not estimate a theta with published item parameters. It is not an IQ.
What is item response theory? →