What 'Low Confidence' Actually Means on a CLCI Entry (and Why It's Not the Same as 'Probably Wrong')

Each group profile on CLCI Hub carries a Confidence label: High, Medium, or Low. It sits next to the score badge in the header. We have noticed users misreading what it means.

Confidence on this site does not measure whether the patterns described in the entry are true. It measures the density of the public record we are drawing on.

The three levels, formally

• High. Court records, peer-reviewed academic work, multiple corroborating BITE assessments by qualified clinicians, and substantial investigative journalism. The patterns described in the entry are documented in ways that would survive cross-examination. Examples: Scientology, Jehovah's Witnesses, NXIVM, the FLDS.

• Medium. Reputable journalism plus credible ex-member testimony, but limited academic study. The patterns are described by enough independent sources to be defensible, but the formal scholarly literature is sparse. Examples: most Korean megachurches, most named Hindu guru organisations, most Bektashi-style Sufi sub-orders.

• Low. Mostly anecdotal, fragmented documentation. Often a small movement that has produced individual ex-member accounts but limited press coverage and no academic monographs. Examples: Eckankar, Art of Living Foundation, several living-guru circles, a number of online MLM-spiritual hybrid communities.

What Low does not mean

Low confidence does not mean we think the patterns described are dubious. It does not mean we are guessing. It does not mean the entry should be ignored.

It means: if you are about to make a high-stakes decision based on the entry — whether to confront a family member, whether to leave the group, whether to publish on the basis of this material — you should know that the underlying record is thinner than for a High-confidence entry, and you may want to do additional source-checking rather than relying on this profile alone.

For most readers, most of the time, a Low-confidence entry is still useful: it tells you the patterns plausibly described in the public record, and it does so in the same structured BITE-derived shape as the rest of the dataset, so you can compare across entries.

What we changed in the body text

After our pass-5 audit found that several Low-confidence entry bodies were written with the same prosecutorial certainty as High-confidence entries, we went back and added explicit hedging language to the openings of those entries. You will now see phrases like "ex-staff describe", "allegations from individual accounts", and "the entry is rated Low confidence — the score reflects patterns plausibly described in public testimony rather than a settled body of evidence" on entries like Art of Living, Eckankar, and Isha Foundation.

This isn't equivocation. It's calibrated reporting. We want the prose to match the confidence label so the field stops being decorative.

Practical reading rules

• If the badge says High: trust the entry as a defensible reading of the public record.
• If the badge says Medium: trust the entry, but check the cited sources before quoting it externally.
• If the badge says Low: read the entry as a starting point, follow the cited sources, and treat the score as descriptive rather than diagnostic.

In all three cases, the body and the timeline carry information the badge cannot. Read them.

This is educational, not legal or clinical advice.