ADHD Test Methodology

TestForADHD.com is a three-stage screening tool for a preliminary assessment of ADHD symptoms in adults. It combines a self-report questionnaire (45 questions across 5 areas, a 7-point scale, scored as POMP) with two short behavioral cognitive tasks: a Go/No-Go task inspired by the Conners CPT 3 paradigm and the classic Stroop test.

Why two kinds of measurement? Self-report scales are good at ruling ADHD out, but they produce many false positives: in clinical samples, the positive predictive value of most adult ADHD screeners falls below 20% (Harrison & Edwards, 2023). So besides asking how you describe your everyday functioning, we also measure how you perform on tasks that require sustained attention and response inhibition. This is a second, independent source of information: it shows whether the difficulties you report are also visible in task performance.

The test is an indicative screening tool. It does not diagnose, confirm or rule out ADHD, and it does not replace a consultation with a qualified professional.

How the test works

Questionnaire: 45 questions across 5 areas of functioning
Go/No-Go cognitive task: about 1 minute
Stroop test: about 1 minute
Results preview: free
Full report: paid, with AI analysis, recommendations and a bonus e-book

Both cognitive tasks are a standard part of the test and everyone completes them.

Part 1: Self-report questionnaire

Structure

The questionnaire has 45 questions in 5 areas. Together they cover the three symptom groups described in the DSM-5 criteria (inattention, hyperactivity, impulsivity) as well as the executive difficulties typical of adult ADHD:

Attention and Focus (10 questions): concentrating and sustaining attention
Impulsiveness (9 questions): behavioral control and decision-making
Hyperactivity (9 questions): physical activity and restlessness
Time Management and Organization (9 questions): planning and organizing tasks
Task Initiation and Completion (8 questions): starting and finishing things

The questions describe everyday situations (work, conversations, chores) rather than abstract concepts.

Answer scale

You answer each question on a 7-point scale. The two ends are labeled to match the question, for example from "very often" to "rarely" or from "never" to "always", so you can give a nuanced answer instead of a simple yes or no. Some questions are worded in reverse (a high answer means no difficulty); those answers are reversed before scoring.

Scoring

Each answer is converted to a percentage of the maximum possible score, POMP (Percent of Maximum Possible; Cohen, Cohen, Aiken & West, 1999): the answer least indicative of difficulties is 0% and the most indicative is 100%. An area's score is the mean of its questions, and the overall score is the mean of all 45 answers.

This is dimensional scoring, similar to the CAARS (Conners' Adult ADHD Rating Scales), where the total of all item scores counts rather than the number of answers above a cut-off. Every answer affects the result in proportion to its intensity, including answers that indicate no difficulty. The profile therefore reflects how strong the reported difficulties actually are, rather than a pass/fail outcome.

Result classification

The overall score falls into one of four groups. The thresholds are modeled on the CAARS score bands (typical / borderline / clinically significant) and mapped onto the 0–100% POMP scale. They are not population norms: each label describes how closely your answer profile matches ADHD symptoms, not a statistical probability of a diagnosis.

0–29%, Unlikely: the answer profile does not point to significant ADHD symptoms
30–49%, Possible: some areas show elevated scores
50–69%, Likely: a clear symptom profile; a consultation with a specialist is worth considering
70–100%, Highly likely: the results suggest significant difficulties; we recommend a diagnostic consultation

The same thresholds (30 / 50 / 70%) are used to describe each of the 5 areas in the report: typical, mild, elevated and marked.

Part 2: Go/No-Go cognitive task

Paradigm

The task is inspired by the Conners CPT 3 paradigm. Letters appear on the screen one at a time. You respond to every letter except X, by pressing the SPACEBAR on a computer or tapping the screen on a phone. When an X appears, you have to hold back your response.

About 75% of the letters require a response, which builds a strong habit of responding that has to be stopped on X. The task therefore measures inhibitory control and sustained attention, executive functions that research often finds to be weaker in ADHD.

Our version is much shorter than the clinical Conners CPT 3 (which takes about 14 minutes). It is a screening version, not a reproduction of the clinical test.

Parameters

Letter display time: 250 ms (as in the Conners CPT 3)
Interval between stimuli (ISI): 1100–1500 ms (random)
Probability of an X (No-Go) on each trial: 25%
Trials: 30 (one block); the first 2 are excluded as warm-up, so 28 trials are analyzed
Practice: 6 trials with feedback (not counted toward the result)
Total time: about 1 minute

Metrics

Omissions: no response to a letter other than X, an indicator of inattention
Commissions (impulsivity): responding to X, an indicator of difficulty with inhibition
Mean reaction time: response speed (mean of correct responses)
Reaction time SD: standard deviation, the spread of reaction times
Reaction time CV (SD / mean): response variability independent of overall speed. Increased reaction time variability is one of the most frequently replicated findings in ADHD research (Kofler et al., 2013, a meta-analysis of 319 studies)
d′ (d-prime): signal detection sensitivity, i.e. how well you tell "respond" letters apart from X

Metric classification

Three metrics (omissions, commissions and variability) are classified independently as typical, borderline or elevated. This way, a single clear signal (for example high impulsivity alongside good attention) is not averaged away.

Omissions: typical up to 5%, borderline 5–10%, elevated above 10%
Commissions: typical up to 15%, borderline 15–25%, elevated above 25%
Variability (CV): typical up to 0.20, borderline 0.20–0.30, elevated above 0.30

These thresholds are indicative, based on ranges reported in the CPT and Go/No-Go literature for adults (Conners, 2014; Bezdjian et al., 2009). They are not clinical norms.

Part 3: Stroop test

Paradigm

A color name appears on the screen (for example "RED"), printed either in the same color (a congruent trial) or in a different one (an incongruent trial, for example the word "RED" printed in blue). Your task is to pick the color of the letters and ignore what the word says.

The test measures interference control: the ability to suppress an automatic response (reading the word) in favor of a deliberate one (naming the color). It is the classic paradigm described by Stroop in 1935 and is still used in neuropsychology today. In ADHD research, Stroop findings are less consistent than reaction time variability (Lansbergen, Kenemans & van Engeland, 2007), so we treat it as a complement to the picture of executive functioning, not as a standalone indicator.

Parameters

Trials: 20 (each trial has a 50% chance of being congruent or incongruent)
Practice: 3 trials (all congruent, with feedback)
Response time limit: 2 seconds
Interval between trials (ISI): 350–600 ms (random)
4 colors: red, blue, green, yellow
Total time: about 1 minute

Metrics

Stroop effect: the difference in mean reaction time between correct incongruent and correct congruent trials (ms). The larger it is, the stronger the interference
Accuracy: the percentage of all trials where the correct color was chosen
Errors on hard trials: the error rate on incongruent trials, where the automatic response pointed the wrong way
Mean reaction time, congruent / incongruent: reaction time for each trial type

Metric classification

Two metrics are classified as typical, borderline or elevated:

Stroop effect: typical up to 150 ms, borderline 150–250 ms, elevated above 250 ms
Errors on hard trials: typical up to 5%, borderline 5–15%, elevated above 15%

These thresholds are indicative, based on Stroop effects commonly reported for adults. They are not clinical norms.

AI analysis

The personalized part of the report is written by a large language model (Google Gemini) following instructions we developed from the methodology described on this page. The model receives:

the question texts and your answers converted to POMP, the scores for the 5 areas and the overall score,
the Go/No-Go metrics (omissions, commissions, mean reaction time, CV, d′),
the Stroop metrics (Stroop effect, accuracy, errors on hard trials, mean reaction times).

The model does not receive your name or email address. Based on this input, it describes each area, compares the questionnaire results with the behavioral task results (noting whether they agree or diverge) and prepares recommendations.

What the report includes

Overall score (POMP) and classification (unlikely / possible / likely / highly likely)
Go/No-Go results: classification of 3 metrics (omissions, commissions, reaction time variability), a reaction time histogram and details (mean reaction time, SD, d′, number of hits)
Stroop results: Stroop effect, accuracy and errors on hard trials, plus mean reaction times and accuracy for congruent and incongruent trials
An AI summary that compares the questionnaire results with both cognitive tasks
A description of each of the 5 areas with its percentage score and severity level
Personalized recommendations for areas scoring 30% or more
The option to print the report or save it as a PDF
Bonus: the "ADHD Guide" e-book (PDF, in English). The download link is emailed together with the notice that your report is ready

Limitations

The test is a screening tool, not a diagnostic one. Even clinical computerized tests are not accurate enough on their own: a meta-analysis of continuous performance tests found a pooled sensitivity of 0.75 and specificity of 0.71 (Arrondo et al., 2024), and for the QbTest the figures were 0.78 and 0.70, with the authors recommending it only as an adjunct to a full clinical assessment (Bellato et al., 2024). An ADHD diagnosis is made by a psychiatrist or psychologist based on a clinical interview.
The cognitive tasks are short (28 analyzed Go/No-Go trials, about 7 of them with an X on average, and 20 Stroop trials). This is a trade-off between convenience and precision: a single mistake can noticeably change a percentage.
Results can be affected by your emotional state at the time, lack of sleep, fatigue, medication or other substances, and by distractions during the test.
The test does not account for co-occurring conditions (such as depression, anxiety or mood disorders), which can produce symptoms similar to ADHD.
Reaction times measured in a web browser include a constant delay of several tens of milliseconds (about 80 ms on average in Anwyl-Irvine et al., 2021), which depends on the device and browser. Within-person comparisons (such as the Stroop effect or reaction time variability) are robust to this, but absolute reaction times cannot be compared directly with laboratory norms.

When to see a specialist

If your result is "likely" or "highly likely", or if you recognize ADHD symptoms that significantly affect your daily life, we recommend consulting:

a psychiatrist (formal diagnosis and medication, if appropriate)
a clinical psychologist (neuropsychological assessment)
a psychotherapist (cognitive behavioral therapy is a well-documented form of support for people with ADHD)

Scientific foundations

The test methodology draws on the following instruments and studies:

Conners CPT 3: the paradigm that inspired the Go/No-Go task (Conners, 2014)
CAARS: the reference for the result classification bands (Conners, Erhardt & Sparrow, 1999)
ASRS v1.1: the WHO Adult ADHD Self-Report Scale (Kessler et al., 2005)
POMP: Cohen, Cohen, Aiken & West, 1999 (Percent of Maximum Possible as a standardized scale metric)
Harrison & Edwards, 2023: accuracy of self-report methods in identifying adult ADHD (Journal of Attention Disorders)
Arrondo et al., 2024: meta-analysis of continuous performance tests (sensitivity 0.75, specificity 0.71; JAACAP)
Bellato et al., 2024: meta-analysis of the QbTest (sensitivity 0.78, specificity 0.70; JCPP)
Kofler et al., 2013: meta-analysis of reaction time variability in ADHD (319 studies; Clinical Psychology Review)
Bezdjian et al., 2009: the Go/No-Go task, with omissions and commissions as separate indicators of inattention and impulsivity
Stroop, 1935: the color-word interference effect
Lansbergen, Kenemans & van Engeland, 2007: meta-analysis of Stroop interference in ADHD (Neuropsychology)
Anwyl-Irvine et al., 2021: precision of reaction time measurement in web browsers (Behavior Research Methods)

The methodology is updated as new scientific evidence becomes available.

Privacy

All data is processed in accordance with the GDPR. Your results are available only to you, through a personal link, and we do not share them with other people. To generate the AI analysis, we send your answers and task results without your name or email address. See the Privacy Policy for details.