How evidence grades work

An evidence grade says how much published research supports a task as a measure. It rests on three counts: papers that report the task's test-retest reliability, papers on its reliability or validity, and registered clinical studies that use it as an outcome.

Grades describe the published research on each task, not ConductCognition's version of it. No grade is set by hand. When the counts are taken again, the grades follow them.

Source: ConductCognition test census 0.39.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.

A battery takes the grade most of its tests reach. A minus means some tests have a lower grade, and a battery is never more than one letter above its weakest test. A study you build is graded the same way. Six tests at grade A and three at grade B make A−; two at grade A and one at grade C make B−. A test the census has not counted is listed as not graded and does not count.

A battery with grade A has at least 25 test-retest papers and 25 registered clinical studies behind every one of its tests. 5 of the 26 batteries in the catalog have grade A, and 8 have A−. Each battery's page names the tests that hold it below the next step and the counts they lack.

In the study builder, your protocol shows its grade. See why lists each test with its own grade and marks the tests that set the letter and the tests that add the minus.

The Data pages also give each test a use grade, from A to E, for the number of papers on the task from 2021 to 2025. A task can be in wide use with little published measurement evidence, or the other way round, so the two letters answer different questions.

A grade counts papers and registered studies. It does not judge how good each paper is, and it does not use data from ConductCognition sessions.

Where ConductCognition has measured a test's reliability from its own sessions, the study builder shows that figure beside the grade. The letter stays the published-evidence grade.