NJ ESEA waiver accountability profiles

For five years New Jersey printed, for every school and every district, a one-page verdict: whether each student group turned up for the state test, and whether it cleared a bar the state had set for that group in particular. Then the waiver that required it ended, and the whole apparatus stopped. This page is the manual for reading what it left behind — including a glyph that appears on 24,894 rows and means the opposite of what it looks like.

1 · The source

Between 2012 and 2016 New Jersey operated under an ESEA Flexibility waiver — a federal agreement that let states replace No Child Left Behind's adequate-yearly-progress machinery with a plan of their own. New Jersey's plan produced an annual accountability profile: one page per school and one per district, published as a zip of Crystal Reports PDFs on the department's Title I accountability page.

Each page carries three tables. Two academic ones — English language arts and mathematics, ten student groups each — and beneath them a graduation-and-attendance block. Per group and subject the academic tables print how many students were enrolled, what share were not tested, whether participation was met, how many valid scores there were, what share cleared the bar, the bar itself, and the state's determination.

Subgroup            # Enrolled  % Not Tested  Met Part.  Valid Scores  % Proficient  Target  Met Perf.
Schoolwide                 425           0.0        YES           413          97.1      90   MET GOAL
Students with Disab.        54           0.0        YES            49          57.1    64.4       YES*

That second row is the whole problem with this dataset, and section 2 is about it.

We render this on school profiles under How the state has judged this school and on district profiles under How the state has judged this district, above whichever 2024–25 ESSA score the school has. 2,234 school codes and 645 district codes in our slug registry match a verified waiver record on exact NJDOE codes.

2 · YES* means the school missed

Through 2014, an asterisk on a determination marks a group that failed to reach the printed target and was credited anyway.

Each year's own key document says why:

So the state set a target, the group came in below it, the state applied a confidence interval, and the group was credited. Across 2012, 2013 and 2014 there are 24,894 such rows. Of those, 24,891 sit below the target printed beside them; not one sits at or above it; three print no target at all. The arithmetic is unambiguous, and it is why we write that cell as “Missed its target, credited within a 95% confidence interval” rather than as “met”.

And in 2015 the asterisk moves house

The 2015 and 2016 editions publish no performance target and no performance determination at all — so there is no column for a confidence-interval credit to sit in. The asterisk reappears in the participation column, where the printed legend gives it a different meaning entirely:

YES   Met Participation rate
YES*  Met Participation Rate (Participation Averaging applied)
NO    Missed Participation Rate

9,826 rows carry it in those two years. A reader — or a pipeline — that carries the 2014 meaning forward reports those schools as having met a performance bar in years New Jersey set none. We store the meaning per edition, with its column and its authority, and the words on the profile page are generated from that record rather than from a lookup shared across the archive.

3 · Two regimes, not five years

The five editions are not one series. They are two, and the join between them is not repairable by conforming.

The measure changes as well: % Proficient becomes % Meeting Standards, defined on the page as PARCC performance levels 4 and 5. A number from one band and a number from the other are not the same measurement, and the state never published a bridge between them.

Preliminary, final, and unmarked

Three editions are titled Preliminary, one Final, and one is unmarked — but the stage belongs to the document, not the year. In 2013 the school profiles say Final and the district profiles say Preliminary, so a single per-year label states the wrong thing about half the 2013 rows. No final edition of 2012, 2015 or 2016 was ever published. Each year section on a profile states its own stage, taken from that page's own title.

4 · Four kinds of blank

Across the 2012–2014 school rows, only 40.2% carry a performance determination. A typical school has 8 of its 20 cells filled.

That is not a defect. New Jersey set a minimum group size — 40 students for a participation calculation, 30 for a performance one — and below it published nothing at all. But the page prints more than one kind of nothing, and they mean different things:

The -- token is the sharpest of these because it appears in no legend NJDOE ever published. Every profile page prints a key covering MET GOAL, YES, YES*, NO and the single dash. A sixth token turns up in 2,178 performance cells across 2012–2014 and is documented nowhere. Measured across the corpus it always co-occurs with a printed target of - on a group at or above the proficiency minimum of 30. On one 2014 page:

Hispanic       45   0.0  YES   37   48.6    -    --     ← large enough, no target
Black          16   0.0   -    10   60.0    -     -     ← below the minimum of 30

Collapsing those two to one token would make the rows say the same thing about opposite situations. We keep both literals and give them different words and different type.

5 · Graduation and attendance

Beneath the academic tables, four of the five editions print a third block. 2012 prints none, at either level — a printed fact, not a missing table.

Graduation is per student group, and OR is the publisher's operator. A group that cleared the four-year bar or the five-year bar was credited on the graduation-rate indicator. Most rows populate only the four-year column; the five-year exists for groups credited on the longer cohort. All three columns are carried separately and the indicator is never re-derived from the two inputs, because which route the state credited is exactly what a reader wants to know. Columbia High School's 2015 row for Students with Disabilities is the case: four-year NO, five-year YES, indicator YES.

The bar moves every edition, and every cohort is lagged behind the year that judges it:

Attendance is one determination for the whole entity and has no student-group column. The publisher prints it once per page and prints it off the subgroup grid — several points above the nearest printed line. A parser that binds a value to the nearest line would file a school-level determination under White on every page in the corpus. It lives in its own table, at its own grain, and there is no column in which to make that mistake.

One more wrinkle: 2013's district profiles are a different table. They print per-group graduation rates as numbers rather than YES/NO, leave the five-year and attendance columns empty, and print the reason beside them — “5yr Cohort Grad rate and Attendance rate will be included in Final District Profiles.” No final 2013 district profile was ever published. Those rates render as numbers with that note attached; they are never coerced into a determination vocabulary.

6 · The seven-year hole

The waiver ended and the profiles stopped. The archive's 17.zip and 18.zip return 404 — probed on every build of the upstream release, and required to keep returning 404, because a year that started resolving would mean the recorded span is wrong. 11.zip returns 404 the same way at the other end.

Nothing this site holds covers 2017 through 2023. The waiver-era archive stops at 2016 and the ESSA calculation workbooks begin with 2024–25. On a profile page that span renders as its own band, carrying the probe result and the reason, between the ESSA score and the waiver record. No axis, no scale and no mark crosses it.

7 · Why there is no chart

The reader's question is whether the school is going in the right direction. The honest answer is: within a regime, yes; across regimes, no.

Three properties of this data settle the form before any aesthetic question comes up.

The values have no order. YES, YES*, NO, MET GOAL, - and -- are six categories, not six points on a scale. YES* is not a better YES — it is a group that missed. - is a suppression, not a low value. A heat-grid encodes magnitude, and there is no magnitude here to encode.

The grid is mostly empty. 60% of the 2012–2014 performance cells are absences, by the state's own minimum-group-size rule. A colour field that is more than half blank is not a picture of anything.

The span is broken, not sparse. Five annual determinations, a seven-year hole that will never fill, then one scored year on an incommensurable scale. Any axis long enough to hold all of it would be asserting a comparison that does not exist.

So the form is a table per year, in reverse chronological order, grouped into bands with the regime break stated in a sentence — which is the only form that renders those six states as six distinct things instead of six points on a scale that isn't there. The reasoning is recorded in the repository's chart manifest as a no-chart entry so it doesn't get re-litigated.

8 · What we built

The five annual archives arrive as an exact, verified offline pin from our upstream data warehouse; this repository downloads and parses nothing. The build verifies the whole artifact, then fail-closes on the source populations, the determination vocabulary of every edition, the asterisk column per edition, the graduation bars and cohort years, the attendance grain, the identity histogram, and the share of 2012–2014 rows carrying a determination at all — before it writes a byte.

The four small tables are the reason the module can tell one kind of absence from another. Drop them and every blank cell on the page looks like the same blank cell. What we deliberately did not take is the target trajectories — 1,017,348 rows of per-year progress targets — because the target a determination was made against is already on the determination row.

The entities move. 2,143 of 2,325 school codes appear in all five editions; 182 do not, and about 3% of all rows belong to entities that no longer resolve to a current New Jersey organization at all. Those are correct answers, not join failures: we never name-match a historical school into a present-day one to raise the rate. A profile renders the editions its entity appears in and says nothing about the ones it does not.

Source data: New Jersey Department of Education, Office of Title I, ESEA Flexibility waiver accountability profiles 2012–2016 (nj.gov/education/title1/accountability). The annual archives are captured once each at their exact byte counts; the collection is closed.

The text on this page was generated by Claude Opus 5, working as part of a stack of tools created by Lyra Forge.