Research methodology

Certuvo has not yet published original research. No surveys, studies or statistics of our own appear anywhere on this site, and any exam statistics we cite come from the official publications named on the citing page. The /research section of this site remains unpublished and noindexed until authentic data exists that meets the standards below. We publish the methodology first so that when research does appear, it can be judged against a bar we committed to in advance — not one written after the fact to fit the results.

Research involving people — surveys, interviews, or analysis of platform usage — requires consent that is informed (participants are told what is collected, why, and how findings will be published), explicit (opt-in, never assumed from product use or buried in general terms), and revocable (participants can withdraw, and withdrawn data is excluded from future analyses and publications). No consent, no data — there are no exceptions for data we find convenient to have.

Research datasets are stored anonymised: no names, email addresses or other personally identifying data are kept in the analysis dataset, and identifiers used to manage participation are held separately from responses and never published. Published findings are aggregated so that no individual participant can be identified from them, including through small-cell combinations of attributes.

Every published study must disclose enough about its sample for a reader to judge what the findings are worth: the sample size (n), the collection window, how participants were recruited, and the response rate where applicable. A finding from 40 self-selected respondents is a different object from one drawn from thousands, and our reporting must make that difference visible — in the write-up itself, not in a footnote a summary can drop.

Every published study must carry a methodology section describing how the data was collected and analysed, and a limitations section stating plainly what the study cannot support — sampling bias, self-report effects, and the line between correlation and causation. Before publication, each study receives a statistical review as part of the standard review workflow, checking that the analyses are appropriate and that stated conclusions do not outrun the data. Headline claims a reader is likely to quote must be defensible as quoted.

Where participant consent allows, the anonymised dataset behind a published study will be downloadable alongside it, so third parties can check our analysis rather than trusting it. And one rule sits above all the others, restated from the AI content policy and the editorial standards: we do not fabricate research data, in whole or in part, by any means, human or automated. A study we cannot run honestly is a study we do not publish.