What an online depression test is actually measuring

Nearly every reputable online screener is a repackaged version of one validated instrument, usually the PHQ-9. It asks how often, over the past two weeks, you have experienced nine specific symptoms — low mood, loss of interest, sleep change, fatigue, appetite change, feelings of failure, concentration trouble, being noticeably slowed or agitated, and thoughts of self-harm. Each answer scores 0 to 3. The total lands between 0 and 27.

A score of 10 is the conventional cut point because that is where the instrument’s ability to separate depressed from non-depressed populations is best balanced. Below it, the test misses too little; above it, it flags too much. Ten is the compromise, and it is a compromise, not a law of nature.

If you want the score bands and what to do with each one in more depth, the full PHQ-9 guide covers the instrument specifically.

The four ways online screening gets it wrong

1. It cannot see the medical causes

This is the big one. Hypothyroidism, anemia, low vitamin B12 or D, untreated sleep apnea, and a long list of common prescriptions — including some blood pressure medications, hormonal contraceptives, and corticosteroids — all generate fatigue, low mood, poor concentration, and sleep disruption. A person with an underactive thyroid can score 18 on the PHQ-9 and not have depression at all. No questionnaire can order bloodwork.

2. It cannot detect bipolar disorder

Someone in a bipolar depressive episode scores exactly like someone with unipolar depression, because the depressive symptoms are identical. The difference lives in the history — past periods of elevated or irritable mood, sharply reduced need for sleep, racing thoughts, uncharacteristic spending or risk-taking. The PHQ-9 asks about none of that. This matters clinically: starting an antidepressant alone in undiagnosed bipolar disorder can precipitate a manic episode. It is the single strongest argument for a real evaluation rather than a self-directed conclusion.

3. Self-report drifts

People minimize, particularly around the items that feel shameful — the guilt question and the self-harm question are the two most commonly under-answered. Others answer while in the middle of an acute bad week and produce a score that would look different by Thursday. The instrument assumes an honest, representative two weeks, and it has no way to check.

4. Unvalidated quizzes are just quizzes

Plenty of sites host “depression tests” that are not based on any published instrument. If the page does not name the questionnaire it is built on — PHQ-9, BDI-II, CES-D, something with a citation — the number it hands you means nothing. Check for the name before you take it seriously.

The other instruments, and when they are used

The PHQ-9 dominates because it is short and free. Two others come up often enough to be worth recognizing.

The Beck Depression Inventory (BDI-II) runs 21 items scored 0 to 63, and it weights the cognitive and emotional side of depression more heavily — self-dislike, pessimism, past failure. Therapists use it more than prescribers do, often to track change across a course of therapy.

The Hamilton Rating Scale (HAM-D) is not a self-report at all. A clinician conducts a structured interview and rates 17 to 21 items based on what they observe as well as what you say. It is the standard outcome measure in antidepressant trials, which is why drug efficacy is usually reported in HAM-D points. You will rarely encounter it outside research.

None of these is more “accurate” than the others in the abstract. They were built for different jobs.

Where screening genuinely earns its keep: tracking

The most valuable use of a depression test is the one almost nobody does at home — taking it repeatedly.

A single score is a snapshot with a wide error bar. Four scores taken two weeks apart is a trend line, and a trend line is clinical information. It shows whether a medication is working before you can feel it working, catches a plateau that means the dose needs adjusting, and flags a relapse weeks earlier than memory would. Antidepressants typically need four to six weeks for full effect, and during that stretch a patient’s own sense of progress is unreliable in both directions. The numbers are steadier than the impression.

If you are starting treatment, take the screener the week you start and keep the number. Repeat it every two to four weeks and bring the sequence to your appointments. It changes the conversation from “I think maybe a little better?” to something a prescriber can act on.

Answering honestly is the part that determines accuracy

The instrument is only as good as the input, and there are two predictable places people shade their answers.

The first is the self-harm question — item nine, thoughts that you would be better off dead or of hurting yourself. People skip it or downgrade it, usually out of fear that answering truthfully triggers something. In an outpatient psychiatric setting a positive answer prompts a conversation about safety, not an automatic hospitalization. Under-answering it removes the one item clinicians look at first.

The second is the guilt and worthlessness item, where the accurate answer often feels like self-indulgence. It is a symptom, and it is diagnostically informative precisely because healthy people do not endorse it.

What to do with your result

Under 5, and nothing else is wrong: nothing, for now. Retake it if things shift.

Between 5 and 9: repeat in two to four weeks. Mild scores that stay flat are worth mentioning to a provider; mild scores that climb are worth acting on.

Ten or higher: book an evaluation. Not because the number is a diagnosis, but because at that level the odds of something treatable being present — depression, thyroid disease, an anxiety disorder, a medication effect — are high enough that guessing is a poor strategy.

Any positive answer on the self-harm item, at any total score: same-day contact with a provider or crisis line.

An online psychiatric evaluation takes the screening result and does what the screening cannot — reviews your medical history and medications, asks the bipolar screening questions, orders labs where the history warrants it, and produces an actual diagnosis. If depression is what it turns out to be, treatment can start in the same visit.

Frequently asked questions

Can an online depression test diagnose me?

No. It measures symptom severity. A diagnosis requires ruling out medical causes, screening for bipolar disorder, and assessing duration and functional impact — none of which a questionnaire performs.

Are free online depression tests reliable?

The ones built on the PHQ-9 or BDI-II are, within the limits above. Tests that do not name their underlying instrument are not measuring anything validated, whatever number they display.

How often should I retake one?

Every two to four weeks during active treatment. Every few months if you have a history of depression and want an early warning on relapse. Daily retesting produces noise, not information.

My score was low but I still feel awful. Now what?

Get evaluated anyway. The PHQ-9 measures depression specifically and will miss anxiety disorders, PTSD, ADHD, and medical conditions entirely. A low score rules one thing out, not everything.

Will my provider make me retake the test?

Probably, and that is normal — they need a baseline recorded under known conditions. Bring your own result anyway; a two-week-old score plus today’s score is already a trend.

Turn a score into an answer

Samz Mental Health provides online psychiatric evaluation with a board-certified psychiatric nurse practitioner — medical history, medication review, and diagnosis, not just a number.

Book an Evaluation