Nearest Grade Average Method

State as of April 2026: proposed, tested against current AGPs, endorsed by the committee for further pursuit, blocked from full adoption by ESSA Plan Constraints.

OEAA’s own framing was “dusting off” the method — it is not new. It was used in 2018-19 for the test transition to PSAT 8/9 (grade 8) and SAT with Essay.

How it works

  1. Create a previous score history for the nearest grade — the closest grade for which a full model is estimable.
  2. NTILE(50) the prior-year-1 scores of that nearest grade.
  3. Set the norm group: assessment, content area, grade, and prior-year-1 achievement NTILE(50).
  4. Calculate the average AGP for each norm group.
  5. Validate, then apply.

Why NTILE(50)

Fifty buckets is a precision/stability trade made explicitly. Fifty is fine enough to give “many small buckets” of output AGPs — more precision than the logistic regression method’s “handful of large buckets” — and coarse enough that each bucket averages 1,000 or more R-package AGPs, which is what buys the stability.

Precision, relative

Method Precision unit
Quantile regression (R SGP package) Individual student score history
Nearest Grade Average Group score history — many small buckets
Logistic regression Group score history — a handful of large buckets

Slightly less precise than the R SGP package; more precise than logistic regression.

What it buys

Every high school grade can project three years out, the same horizon as elementary and middle school. That directly addresses the grade 11 one-year-of-runway problem described in Adequate Growth Targets.

Evidence presented

Four Growth Index scatterplots (deck slides 21–24, images only):

  • Content areas, model alignment — precision between R package and NGA AGPs is “pretty close”
  • Content areas, model application — meaningful impact on overall rating
  • English Learner, model alignment
  • English Learner, model application — same magnitude of impact on the EL Index using WIDA

Summary as given: fairly tight alignment, more reasonable targets for both content areas and English Learners, and meaningful impacts on aggregate Growth and EL Progress index values.

OEAA has not determined whether that impact is desirable and said it would need time to research it. Tracked as Is the impact of NGA on index values desirable?.

The EL timing detail

The deck asserts EL impact will increase as Michigan reaches five years post-COVID. The mechanism is only in slide 20’s speaker notes:

By 2026-27 WIDA NGA AGPs would be applied to grades 8-12. However, we are choosing to limit to data post covid (i.e., 2021-22 and forward). Meaning as of 2024-25, the most we can project out is 3 years.

The projection horizon is currently constrained by data availability, not by policy. As the post-COVID window lengthens toward five years, the EL horizon extends and the impact grows.

Committee position

Endorsed — Pursue the Nearest Grade Average approach — with the accompanying view that the adequacy threshold itself should rise (Set adequate growth above the 65th percentile).