Cross-cutting 路 The canonical glossary
Data Governance & Definitions
One canonical definition for every metric this office publishes 鈥� census dates, cohort-grain rules, suppression, and source models. Every dashboard and report links here rather than restating a definition, so two numbers can’t quietly disagree.
NWCCU 1.BFERPASingle source of truth
Definitions
How each headline metric is defined
These definitions are load-bearing. The warehouse has already been bitten by grain confusion (retention where “# Retained” could exceed “# Students”) and by schema drift (renamed columns in a downstream model) 鈥� this page exists to prevent a repeat.
| Metric | Definition | Grain | Source model |
|---|---|---|---|
| Enrollment headcount | Distinct students active as of the official census date | Student | gold.fact_enrollment |
| Retention | Cohort members still enrolled one year later 梅 cohort members | Cohort (enrollment-grain via tracking_id) | gold.enrollment_retention |
| Graduation (150%) | SUM(is_included_in_cohort 路 grad_150) / SUM(is_included_in_cohort) | Cohort | gold.enrollment_outcomes |
| Alumni count | Distinct students with has_graduated = 1 (not GRAD enrollment rows 鈥� a multi-credential graduate counts once) | Student | gold.student_graduation |
| Licensure pass rate | First-time / ultimate pass on the program’s national exam | Program 路 cohort | seed: program_licensure_outcomesPlanned |
| Employment in field | Graduates employed in the field of study | Program 路 cohort | seed: program_licensure_outcomesPlanned |
Cohort grain vs. enrollment grain
Per-cohort retention must be computed at enrollment grain (via the fact tracking id), not student grain 鈥� otherwise “# Retained” can exceed “# Students” for a cohort. Graduation is a cohort-grain ratio keyed on is_included_in_cohort (the physical warehouse column). Align to the physical schema, not renamed downstream copies.
Small-cell suppression (FERPA)
Disaggregating per-program, per-cohort across several dimensions at a small graduate school produces cells with fewer than ten students. All public disaggregated dashboards suppress cells where n < 10, citing FERPA.
Census & "as-of" dates
The enrollment funnel refreshes four times a day; cohort and census models are point-in-time. Every published metric states its official snapshot date, so Fast Facts on the home page cannot silently disagree with the Fact Book.