Skip to main content
Every indicator in voi carries a citable source. This may be an academic citation, a framework name, or both. These are shown as an attribution badge, together with a link to the original framework or methodology page. This lets you, or anyone reading your report, verify exactly which standard a result maps to. You do not have to take a number on faith.

​
The four source classifications

Every indicator (and every question behind it) is labelled with one of four classifications, showing exactly how much you can trust the wording itself, independent of how good the underlying concept is:
The wording is taken verbatim from a peer-reviewed, scientifically validated instrument. This is the kind of scale used in published research, such as a validated self-esteem, well-being, or food-security scale. Nothing is reworded or paraphrased; the question is exactly what the original instrument asks and no question is skipped to construct the final indicator.
The wording is taken word for word from an established institutional source. Examples are a module from the World Health Organization, an item from OECD or Eurostat, a metric from a recognised impact framework, or the standard item used by a national statistical office or UN agency. This classification also covers the standard proxy question generated for a widely used public indicator, such as one tied to the Sustainable Development Goals or a World Bank statistic, but is not necessarily peer-reviewed and scientifically validated.
The underlying definition comes from a recognised framework, but no ready-made question exists for it. So voi’s AI wrote the actual wording to fit that definition. The concept is externally defined. The phrasing is voi’s. This is also how open-ended, qualitative questions are usually built, since a scientific or institutional source rarely provides one ready-made.
This wording is written for your workspace specifically, either by you or by voi’s AI on your behalf, with no external framework behind it. This is what you get when you add your own question, or when voi fills a gap in your Theory of Change that the library has no good match for. See Editing a question.
At a glance: An indicator built from more than one question is only as strong as its weakest link: it inherits the lowest of its questions’ classifications, never the highest. Mixing one verbatim scientific question with one you wrote yourself yields an indicator classed by the question you wrote. This is deliberate. It means a classification never overstates how rigorously an indicator was actually sourced.

​
Writing or heavily editing a question always classes it as Generated by user

Any question with no library or institutional framework behind its wording, whether voi drafts it to fill a gap your Theory of Change’s library indicators don’t cover, or you write or substantially reword it yourself in Plan, is always classed Generated by user, and this classification is locked. Only voi’s library team can class a question Scientific source or Institutional source, since only they can vouch for a verbatim match to the original source. See How indicators work for how these classifications shape which indicators voi recommends for your study.