Why two scores?
An effect can be very well supported in the laboratory and still do little in an online shop. We therefore separate two questions. The effect score (E) answers how well the psychological phenomenon itself is supported. The transfer score (T) answers how well it is supported that it works in digital products, in e-commerce or in a B2B setting. Both run from 1 to 5.
Effect score (E)
| Level | Label | Criterion |
|---|---|---|
| 5 | Established | A meta-analysis or multi-lab replication confirms the effect, the direction is consistent, boundary conditions are known. |
| 4 | Robust | Several independent studies from different laboratories, no notable counter-findings from large replication projects. |
| 3 | Context-dependent | Replications exist but are mixed, or the effect shows only under narrow conditions or is small and heterogeneous. |
| 2 | Weak | Replications mostly failed, or only a partial aspect of the original finding holds. |
| 1 | Preliminary | A single study without replication, or the evidence comes only from popular literature. |
Transfer score (T)
| Level | Label | Criterion |
|---|---|---|
| 5 | Practice-proven | Several field or A/B tests in a digital or B2B setting. |
| 4 | Field-supported | Field studies in a similar context, transfer well justified. |
| 3 | Partially supported | Individual field findings or laboratory studies with realistic stimuli. |
| 2 | Plausible | Only laboratory or scenario studies, transfer justified but untested. |
| 1 | Hypothesis | Transfer speculative, only reasoning by analogy. |
Caps
- Without independent replication: E at most 3.
- A single study without any replication: E at most 2.
- Failed or clearly weaker replications of large projects: E at most 3 until the situation is clarified.
- No field or practice data: T at most 2.
- Books, blogs and agency studies do not count as evidence, only as a pointer to the original source.
- If sources conflict, the lower value applies until the editorial team has resolved the conflict with justification.
How practical tips are labeled
- Supported: backed directly by field or A/B data.
- Plausible: derived from laboratory findings.
- Hypothesis: reasoning by analogy, test with an A/B test before use.
Evidence level and check date
From the scores we derive an evidence level: high (E at least 4 and T at least 3), moderate (E at least 3) and preliminary (E at most 2 or T at most 1). Every page states a check date. We review all scores at least every twelve months and immediately when new meta-analyses or failed replications appear.