A good online self-check is accurate at the job it was designed for — describing the intensity of what you have experienced recently — and cannot do the job people usually hope it will do, which is decide what is wrong.
By Sam Roberts, MA Professional Counselling; Clinical Supervisor (ACA) — last reviewed 2026-09-05
Well-constructed self-checks are reasonably accurate at describing the severity of symptoms over a defined period, which is what they were designed for. They are not accurate at diagnosis, because a diagnosis requires history, context, alternative explanations and a qualified clinician weighing them together. Treat a result as organised information about your last two weeks, not as a verdict, and use it to decide whether a conversation is worth having.
Two properties matter. Reliability is whether the same person answering honestly on two nearby occasions gets a similar result. Validity is whether the score corresponds to the thing it claims to measure. Established questionnaires are tested against both, in large samples, before they are used clinically, and the good ones perform well on the narrow task of grading severity.
What that testing does not establish is diagnostic certainty for any individual. A questionnaire has no way of knowing that you have had four hours of sleep a night for a month because of a newborn, that you moved country in March, or that the low mood arrived alongside a thyroid problem. Those are exactly the facts that change what a score means.
This is why every responsible tool, including ours, uses ranges and descriptions rather than announcing a condition. The number describes intensity. The meaning comes from context.
Timing is the largest factor. Most instruments ask about a defined recent window, usually one or two weeks. Completing one during an unusually hard fortnight will produce a higher score than the same person would produce a month later — which is useful information, not an error, as long as you read it that way.
Honesty is the second. People routinely under-report, particularly on items about hopelessness or drinking, and particularly when they suspect where the answers lead. A quietly under-reported questionnaire is the one situation where the tool genuinely fails.
Who is answering is the third. Parent-completed checks about a child describe what the parent has observed, which is valuable and also partial; a teacher would report a different picture, and both are true.
The fourth is what is being ruled out. Sleep debt, thyroid function, medication, grief, iron levels and chronic pain all produce experiences that questionnaires score as mood or attention difficulty. None of them is visible to the questionnaire.
They are useful for noticing. Many people carry a vague sense that something has shifted without ever describing it specifically, and answering fifteen structured questions makes the shift concrete enough to act on.
They are useful for tracking. Repeating the same check after six or eight weeks of change — better sleep, fewer hours, a course of counselling — gives you a comparison you would not otherwise have.
They are useful for preparation. Clinicians rely on specific examples across time, and most people find those difficult to produce on the spot. Arriving with a completed self-check and its summary makes a first consultation substantially more productive.
And they are useful for lowering the threshold. For a lot of people in Singapore the self-check is the first thing they have ever done about a difficulty they have carried for years.
Read the description rather than fixating on the number. A moderate range means the pattern is affecting daily life often enough to matter, not that you have crossed a line into illness.
Ask the impact question that no score captures: is this getting in the way of work, sleep, study, parenting or your relationships? Impact is what usually determines whether support is worth arranging, more than the score itself.
Give it time if the fortnight was unusual. Retaking a check after a genuinely ordinary two weeks often produces a more representative picture.
And do not take five checks in one sitting. Fatigue degrades answers, and a scattered set of borderline results is harder to act on than one careful one.
Resist the urge to retake a check repeatedly until it produces a reassuring number. Small movements between attempts reflect ordinary day-to-day variation rather than real change, and chasing a better score turns a useful tool into another source of worry. Leave at least a few weeks between attempts if you want the comparison to mean anything.
Watch for the opposite reaction too. Some people read a lower-than-expected result as permission to dismiss a difficulty that has genuinely been affecting them for months. A questionnaire that misses something does not make the something imaginary; your own experience of interference outranks the number every time.
If the result matches what you already suspected, if the difficulty has lasted more than a few weeks, or if it is affecting your ability to function at work or at home, the useful next move is a conversation rather than another questionnaire.
A fifteen-minute consultation is enough to work out whether ongoing support is warranted, what kind, and whether anything medical should be checked first. It is not a counselling session and it does not commit you to anything.
This article is general information, not clinical advice, and nothing in it diagnoses anything. If what you read here matches your experience closely, that is a reason to speak to someone qualified rather than a conclusion in itself.