METHODS
How we read and discuss research
Start with a specific question, read the original evidence, and explain the study’s design, findings and limits. This page sets out our research-reading principles and proposed ways to evaluate experiences.
Start with a need and context
Specify who needs what kind of support and when. Expression, self-understanding, relationships, and sustained psychological change call for distinct research questions.
Distinguish findings from interpretation
Separate the paper’s results, the authors’ explanations, and our design proposals. Observational associations and randomized comparisons support different inferences.
Match outcomes to the aim
Consider immediate feelings, autonomy, later actions, and sustained change separately. Report no change, unsuitable experiences, and withdrawal. Usage alone does not establish support.
Preserve sources and participant choice
Describe samples, methods, versions, and limitations. Link original sources and disclose AI involvement and interests. Treat personal disclosures and chat histories with care.
Reading evidence labels
Evidence labels are editorial reading aids. Design, sample, measurement, bias, and context all affect interpretation. A moderate or preliminary label does not establish product efficacy. Design insights are proposals that still require testing.
Clinical treatment effects require studies in the relevant populations and outcomes. This site discusses everyday emotional support and self-understanding; each article states its own evidential scope.