Evidence

The measurement and privacy case behind the engine, fully cited. Every page names its numbers and their sources in the first hundred words, and each carries a quotable definition you are welcome to reuse. Where the argument cuts against us, we say so.

Why anonymous results need at least 5 people

What an anonymity threshold is, why five, why severe-distress measures need ten, and why a threshold in the schema is a fact while a threshold in the interface is a promise. Cited: ICO, GSS, SDC handbook.

Confidence intervals for small-team scores

Why a score from 8 people needs a wider interval than one from 80, why the t-distribution is the honest choice below about 30 respondents, and a calculator so you can check ours.

How often should you ask? The evidence on frequency

What survey length does to answer quality, what frequency does to participation, and why a 3-item fortnightly cadence follows from the planned-missing-data literature. Counter-evidence cited on purpose.

Three good questions beat thirty mediocre ones

Single-item validity, what long questionnaires cost in quality, and how a planned missing-data design covers 21 domains with 3 items per person. What the design will and will not claim.

The honesty rules, in full

The ten constraints the engine enforces on every deployment, in the schema and the API rather than in settings. Each rule, why it exists, and how it is enforced across three layers.

The engine is free for any employer at alltoogether.com · the methodology is the open OWHS standard · building it into your product is what this site is for.