What “above average” means here — and what’s still missing

“You’re above average” sounds like an objective statement. The part that’s often left out is the crucial one: above the average of which group? A mean from a US student sample is a different thing from a representative German population norm. The same raw score can read as “unremarkable” against one reference group and “elevated” against another.

Whoever leaves the reference group unstated makes a result look more scientific than it is. We don’t want to do that.

Every result page carries a “comparison basis” line. It names what your result is compared against: a representative German sample, a German convenience sample, a US sample, a student or clinical sample — or whether there is no norm data at all.

We also make clear where the “low / average / high” split comes from: a threshold defined in the validation literature (for example the PHQ-9 cut-offs 5 / 10 / 15 / 20), or an Empiralis orientation range — a sensible but not independently validated split, such as “one standard deviation above the mean”. Where a German-language norming exists, we use it: for the ECR-RD8, the WHO-5, SOMEDIS-A, or the KiGGS values behind the KINDL.

We select established instruments, disclose the original study and licence status for each test, and present the results in plain language. What we cannot show is our own psychometric study of the form: “The Empiralis version of test X was examined in N German respondents and shows reliability Y and validity Z relative to the original instrument.”

That is the difference between science-based — built on documented research — and science-evaluated — the specific implementation empirically checked. Empiralis is, for now, mostly the former. We would rather say so plainly than blur it.

Building our own norm sample or validation study would need one thing: systematically collected data from many people. That is exactly what we deliberately do not do. Empiralis stores no results on a server, there are no accounts, your answers stay in your browser, nothing is sent to us and nothing is sold.

There is a real trade-off between doing our own research and keeping anonymity absolute. For now, we resolve it in favour of anonymity.

Several routes are possible, all strictly opt-in and account-free. First, a clearly separated, fully anonymous data donation: only non-identifying summary values, no hidden re-identification, clearly detached from the normal test flow, and removable at any time without giving a reason. That could yield our own German reference values.

Second, a norming or validation study run together with a university group — as a separate, ethics-reviewed collection. The findings feed back into the interpretation; the raw data does not sit with us.

Until then: always read “above average” together with the comparison basis shown beneath it. A screening is a prompt for reflection, not a diagnosis. If a result stays on your mind, the next step is a conversation with a professional — not a second online test.