Typology connector

The corpus has been asking for this longer than it has been able to answer.

education-agency.ont.yml declares a typology property — “Department typology grouping (rural, small town, suburban, urban) used for peer comparison” — and an edge, peer-of, whose description reads “The agency is a comparison peer, by typology or by wealth position”. Neither could be filled from anything committed. Five agency nodes carry a prose characterization and then record the same absence in their own unfilled: block:

- field: the department's typology code
  why: The characterization is entered; the department's own code is not.

Cleveland Municipal, Toledo City, Upper Arlington City, Perrysburg Exempted Village and Northern Local each say “Exact department typology code not yet entered.”

Two more places record the cost rather than the absence. scenario-delta was written to cross funding incidence against typology as well as wealth, and states that it cannot: “Typology is not an available axis… no committed fixture carries a typology code”, [open]. And Fordham Institute — "Ohio House puts the brakes on Cupp-Patterson" warns that a typology figure in that commentary cannot be checked here and should not be reported as though it could, because “assessed valuation per pupil is the nearest proxy the corpus holds and it is not the department’s typology”.

So four recorded open items, all naming one missing file, while the department published it openly the whole time.

Declare dew-typology as a connector in its own right, Wired with no remainder, feeding education-agency, with the 2013 School District Typology workbook as its single source, and commit its Typology By District sheet as a rebuilt fixture beside the CUPP profile in dispersion.

The four classifying measures are carried, and named with their vintage. Enrollment, median income, student poverty and percent minority go into the extract as enrollment_2013, median_income_2013, student_poverty_2013 and pct_minority_2013. That is the argument dew-edchoice-designated-list already makes for carrying the EdChoice criteria beside the designation: a classification that cannot be interrogated is a label, and a label whose derivation nothing records is exactly what this corpus objects to elsewhere.

The suffix is doing real work. The assignment is in force and the measures behind it are eleven years old, and those two facts have to travel together or the second is lost.

Kept apart from dew-foundation deliberately, on the reasoning facts-and-figures-connector established for the same choice. dew-foundation is per-agency state aid and every fixture it produces answers “how much does this district get”. This one answers “what kind of district is this”, which is a property rather than a payment, is not revised on a budget cycle, and is published on a different page under a different regime.

Four open items close and one axis opens. The five agency nodes can carry the department’s code; scenario-delta’s [open] on typology becomes answerable; and the Fordham commentary’s warning can name the file that settles it rather than a proxy that does not.

The vintage becomes a standing caveat. The scheme was set in 1996, revised in 2007 and again in 2013, and the workbook’s own docProps/core.xml gives dcterms:modified as 6 January 2015 — untouched for eleven years. A district coded High Student Poverty was one in 2013. Every use of this data has to carry that, which is why the measures are committed alongside: the drift is measurable against the report card rather than merely suspected.

The codes must not be read as an ordered poverty band. The department orders them by a location composite, least urban to most, with poverty as the secondary term inside each locale. Median student poverty runs 46.4% at code 1, 36.7% at 2, 29.0% at 3 and then back up to 51.5% at code 4 — above codes 1, 2, 3 and 5 alike. Anything that bands 1→8 on an ordered ramp is asserting a monotone that four of the eight steps break.

Code zero is a refusal, not a ninth kind. Five districts are removed from the analysis and are the only rows with a suppressed measure; for the two Bass Island districts every measure is suppressed. Three of the five are in the FY2027 panel. A reader must not take 0 as “least urban”.

It joins the panel exactly. All 609 districts in the FY2027 panel carry a code and none is missing; the five surplus rows are districts closed or merged since 2013. That is the expected shape of an eleven-year-old roster against a current one, and the reverse of the failure that would matter.

Derive a typology from poverty and enrollment, which the panel already carries. Rejected, and it is the alternative worth naming most loudly. The two axes are on the panel and a plausible eight-way split is a few lines of arithmetic — and the result would be an invented classification wearing the department’s category names. fordham-house-bridge-commentary had already written the objection down before the question was asked: the nearest proxy this corpus holds is not the department’s typology. Publishing a derived grouping under those labels is precisely the defect that entry exists to prevent.

Use the NCES urban-centric locale code instead. Rejected. ULOCAL is in the cached CCD members for 1994-95 through 2013-14 and ccd.rs knows where the byte sits; the modern 029 files dropped it to a separate EDGE geocode file that is not cached. So it is both stale and a different measure — a federal locale classification, not Ohio’s composite — and neither derives the other.

Take the CUPP Similar District Data sheet, which is already cached. Rejected because it does not contain this. The sheet reports a district’s similar-district averages and never names the group; its sharedStrings carries no typology label at all. The two files are complements rather than substitutes.

Wait for a revision before committing an eleven-year-old assignment. Rejected. It is the assignment in force — there is nothing newer to wait for, and three revisions in thirty years means waiting is indefinite. The vintage is recorded everywhere it is used instead.

Extract the exemplar sheet too. Rejected. Five illustrative districts per code, all of them already on the assignment sheet. A fixture built for a question nobody has is a fixture nobody checks.