Philadelphia, Brooklyn and Tuscaloosa on a common scale

11 West Philadelphia K-8 schools, 10 schools in and around Park Slope and Prospect Heights, and the 18 schools of Tuscaloosa City Schools, measured on the Stanford Education Data Archive's national scale.

Tuscaloosa City is here because an Alabama family member moved from the district. Brooklyn is here because it is where the Grand Army Plaza family’s child goes to school. In both cases the question was what the other family’s frame of reference looks like measured against Philadelphia’s.

School data 2009-2018, grades 3-8; district data 2019 and 2022. Source: Stanford Education Data Archive (SEDA) 4.1 and SEDA 2022 v2.0.

Reading the numbers

Pennsylvania, New York and Alabama each use their own tests with their own cut scores, so the three states' published “percent proficient” figures are not comparable with one another. SEDA links every state's tests to the national NAEP scale to produce two numbers that are.

Every figure on this page is a number of grade levels, measured against the US average, and can be negative.

Level is how far above or below the national average a school's students score. 0.00 is the national average. −0.60 means students there score about six tenths of a grade level below the national average for their grade — roughly two thirds of a school year behind. +0.88 means about nine tenths of a grade level ahead.
Growth is how much students gain in a year, again against the national average. 0.000 is average progress. +0.075 means gaining about seven hundredths of a grade level more than typical each year, which compounds to roughly a third of a grade level over the five years from grade 3 to grade 8. −0.131 means falling behind by about that much.

The two measure different things. Across the 72,714 US schools in this release, a school's level correlates with the share of its students who are economically disadvantaged at r = −0.81. Its growth correlates with the same measure at r = −0.12 (51,016 schools). Level is therefore close to a measure of which students a school enrols; growth much less so. Both figures are computed from the covariate file published alongside the scores; see SEDA's methods for how the scales are built.

District-level results, 2019 and 2022

In 2019 Tuscaloosa City was 0.17 grade levels above Philadelphia in math. Between 2019 and 2022 Philadelphia's math figure fell by −0.26 and Tuscaloosa City's by −0.10; the median US district fell by −0.15. New York State is absent from this release, so New York City cannot be shown here.

Districtwide, before and after the pandemic. The school-level data above ends in 2018, so this is the current picture: where each district sat in 2019, where it sat in 2022, and how much it lost.

district	subject	2019	2022	change
Philadelphia City SD	math	−0.60	−0.86	−0.26
Philadelphia City SD	reading	−0.52	−0.60	−0.08
Tuscaloosa City	math	−0.43	−0.53	−0.10
Tuscaloosa City	reading	−0.30	−0.34	−0.04
Tuscaloosa County	math	−0.38	−0.42	−0.03
Tuscaloosa County	reading	−0.27	−0.25	+0.03
US median district	math	+0.04	—	−0.15
US median district	reading	+0.04	—	−0.09

Achievement level

Level is the measure all three places have. It is where a school's students score against the national average for their grade, in grade levels.

Achievement level, all three places. Each dot is one school, placed on the national scale in grade levels; the solid rule on each row is that group's median. Zero is the US average. Hover any dot for its national percentile. Brooklyn appears twice on purpose: the ten named schools, and all 400 schools in the borough, which are not the same thing and do not have the same median. The dashed rule on a row is the median of every school in that city; Tuscaloosa City and all-Brooklyn have none, because those rows already are the whole set. The low-income share on each row accounts for most of the distance between them — with the caveat, below, that it is a free-meals measure and Philadelphia's is against its ceiling.

How much of that is who enrols

Level tracks family income closely — r = -0.81 across the 72,714 US schools in this release — so the first thing to ask of any gap between the rows above is how much of it is intake. Fitting level to the low-income share across every US school gives level = +0.70 − 1.33 × (low-income share), and that line predicts each set from its intake alone.

The low-income measure has a ceiling, and it is not the same distance away for every row. SEDA's perecd is built from free-and-reduced-lunch eligibility, and a district serving universal free meals books nearly every school at the maximum whatever its families earn. Philadelphia has done that district-wide since 2014. So the at ceiling column is part of reading every other number in this table: 8% of US schools are recorded at 95% or above, against 64% of the West Philadelphia eleven.

set	schools	low-income	at ceiling	predicted level	actual	difference
The 11 West Philadelphia schools	11	97%	64%	−0.60	−0.53	+0.07, flips at 92%
All Philadelphia schools	279	97%	55%	−0.60	−0.63	−0.03, flips at 100%
The 10 Brooklyn schools	10	30%	0%	+0.29	+0.64	+0.34
All Brooklyn schools (the whole borough)	400	88%	15%	−0.47	−0.20	+0.27
Tuscaloosa City Schools (the whole district)	18	80%	5%	−0.36	−0.61	−0.24

Flips at is the low-income share at which that row's difference would change sign. Both Philadelphia rows flip within a few points of where they are recorded, and more than half of their schools are already at the ceiling, so neither difference can be read off this line: it is a statement about meal policy as much as about schools. The Tuscaloosa and Brooklyn rows would have to move a long way to flip, and almost none of their schools are at the ceiling.

That leaves one comparison the ceiling does not touch. The two Brooklyn rows are one district on one measure, and neither is pinned: the ten named schools are 30% low-income and the borough is 88%, with medians of +0.64 and −0.20. The line predicts most of that gap from intake alone, and what is left is nearly the same for both — +0.34 for the ten and +0.27 for the borough. Whatever separates the ten from Brooklyn is mostly which children walk through the door.

Level versus growth

Across these schools level spans about two grade levels and growth spans about three tenths of one. The two do not rank schools in the same order.

Achievement level against learning rate. Every dot is one school, placed on the national scale in grade levels. Horizontal is achievement level; vertical is growth per year. The crossed lines are the US average on each. Hover any dot for its two percentiles. Only one of the 10 named Brooklyn schools appears here: SEDA publishes a growth estimate for far fewer New York schools than elsewhere, which is set out below. The faint green cloud behind it is every Brooklyn school that does have one — 132 of the borough's 400 — so the named school can be read against its borough rather than on its own.

Philadelphia (the 11 K-8 schools) Tuscaloosa City Schools Brooklyn, the ten (1 of 10 has a growth estimate) all Brooklyn (132 of 400)

Schools where level and growth diverge

Several schools rank very differently on the two measures. Lea is at the 4th percentile nationally on level and the 68th on growth. Meredith and Greenfield are the reverse: 93rd and 69th on level, 29th and 33rd on growth. In Tuscaloosa, Arcadia is 23rd on level and 99th on growth; Alberta School of Performing Arts and Northington are 8th and 7th on level, 95th and 90th on growth.

Meredith and Tuscaloosa Magnet Elementary occupy nearly the same position: 93rd percentile on level for both, 29th and 28th on growth. They overlap on the chart.

The 11 West Philadelphia K-8 schools. Sorted by achievement level, highest first. The two % Black columns are a decade apart: the first is SEDA's own covariate file, measured over the same 2009–2018 window as the scores beside it; the second is the federal count for 2023-24.

school	grades	SEDA 2009–2018				% Black and enrolment
school	grades	level	%ile	growth	%ile	2009–18	2023-24	roll 2023-24
Penn Alexander	Elementary	+0.88	98th	+0.075	88th	33%	13%	583
Meredith	Elementary	+0.59	93rd	−0.016	29th	19%	9%	515
Greenfield	Elementary	+0.17	69th	−0.010	33rd	41%	12%	730
Powel	Elementary	−0.14	40th	—	—	80%	60%	306
Nebinger	Elementary	−0.33	25th	−0.016	29th	54%	36%	404
Bache-Martin	Elementary	−0.53	14th	−0.131	1st	77%	43%	567
Marian Anderson	Elementary	−0.55	13th	+0.003	42nd	86%	41%	309
Harrington	Elementary	−0.65	9th	−0.053	11th	97%	91%	289
Lea	Elementary	−0.82	4th	+0.037	68th	87%	59%	484
Comegys	Elementary	−0.96	2nd	−0.030	21st	97%	93%	267
Alain Locke	Elementary	−1.03	1st	−0.040	16th	92%	77%	261

Tuscaloosa City Schools. Every school in the district with a SEDA estimate. The two % Black columns are a decade apart: the first is SEDA's own covariate file, measured over the same 2009–2018 window as the scores beside it; the second is the federal count for 2023-24.

school	grades	SEDA 2009–2018				% Black and enrolment
school	grades	level	%ile	growth	%ile	2009–18	2023-24	roll 2023-24
Tuscaloosa Magnet, Middle	Middle	+0.63	94th	+0.029	62nd	49%	49%	167
Tuscaloosa Magnet, Elementary	Elementary	+0.58	93rd	−0.018	28th	46%	44%	303
Rock Quarry	Elementary	+0.51	90th	+0.054	78th	12%	11%	655
Verner	Elementary	+0.47	89th	−0.024	24th	27%	21%	621
Northridge	Middle	+0.33	82nd	+0.027	61st	24%	36%	727
Arcadia	Elementary	−0.36	23rd	+0.145	99th	56%	72%	420
Woodland Forrest	Elementary	−0.39	21st	−0.022	25th	79%	82%	506
University Place (elem)	Elementary	−0.50	15th	+0.026	61st	78%	79%	612
Oakdale	Elementary	−0.59	11th	−0.083	4th	99%	92%	289
Southview	Elementary	−0.62	10th	+0.054	78th	94%	94%	645
Eastwood	Middle	−0.67	8th	+0.006	45th	83%	85%	722
Central	Elementary	−0.69	8th	−0.043	15th	98%	80%	301
The Alberta School of Performing Arts	Elementary	−0.69	8th	+0.099	95th	90%	84%	705
Northington	Elementary	−0.70	7th	+0.081	90th	81%	—	—
University Place (midd)	Middle	−0.70	7th	−0.076	5th	85%	—	—
Skyland	Elementary	−0.73	6th	+0.025	59th	93%	82%	456
Martin L King Jr	Elementary	−0.73	6th	−0.003	38th	99%	90%	475
Westlawn	Middle	−0.85	4th	−0.015	30th	99%	88%	539

The 10 Brooklyn schools. Park Slope, Prospect Heights, Windsor Terrace and Gowanus. Sorted by achievement level, highest first. Not all of them are elementary schools — the grades column says which. The two % Black columns are a decade apart: the first is SEDA's own covariate file, measured over the same 2009–2018 window as the scores beside it; the second is the federal count for 2023-24.

school	grades	SEDA 2009–2018				% Black and enrolment
school	grades	level	%ile	growth	%ile	2009–18	2023-24	roll 2023-24
PS 321 William Penn	Elementary	+1.05	100th	—	—	8%	3%	1234
MS 51 William Alexander	Middle	+0.96	99th	—	—	12%	8%	963
PS 107 John W. Kimball	Elementary	+0.93	99th	—	—	6%	2%	517
PS 39 Henry Bristow	Elementary	+0.78	97th	—	—	10%	5%	360
PS 154 Windsor Terrace	Elementary	+0.67	95th	—	—	5%	4%	426
PS 10 Magnet	Elementary	+0.60	93rd	—	—	11%	7%	813
PS 11 Purvis J. Behan	Elementary	+0.46	88th	—	—	75%	52%	852
PS 9 Teunis G. Bergen	Elementary	+0.23	74th	—	—	55%	32%	771
PS 282 Park Slope	Elementary	+0.10	63rd	−0.013	31st	64%	31%	539
PS 133 William A. Butler	Elementary	−0.21	34th	—	—	33%	27%	617

Why Brooklyn is thin on the growth chart

SEDA publishes a growth estimate for far fewer New York schools than it does elsewhere. Across the release:

district	schools	with a level estimate	with a growth estimate
Philadelphia	301	279	241 (86%)
Tuscaloosa City	18	18	18 (100%)
New York City	1282	1214	432 (36%)

Of the 10 Brooklyn schools here, 1 has a growth estimate. That is a fact about New York's reporting in this release, not about these schools, which is why the level figure above carries all three places and the growth chart does not.

PS 118 Maurice Sendak is absent from the tables altogether: it opened in 2013, and the school-level release pools 2009-2018, grades 3-8.

Summary

Median level is −0.53 for the eleven Philadelphia schools, +0.64 for the 10 named Brooklyn schools and −0.61 for Tuscaloosa City. The Philadelphia–Tuscaloosa difference is 0.08 grade levels; the ten Brooklyn schools sit 1.16 above the Philadelphia median.
That +0.64 is a median across ten schools that were named individually, not a figure for Brooklyn. Across all 400 schools in the borough the median level is −0.20 — 0.83 grade levels below the ten, and 0.33 above the Philadelphia median. The ten are 30% low-income against 88% for the borough, which is most of the difference.
Within Philadelphia the range is +0.88 (Penn Alexander) to −1.03 (Alain Locke), about 1.9 grade levels — larger than the gap between any two of those medians.
All three sets have a few schools well above the national average and a larger number below it, and in each the ordering tracks the racial and economic composition reported in the tables.
Penn Alexander and Rock Quarry are the highest-demand school in each district. Their levels are +0.88 and +0.51. Over SEDA's window their enrolments were 33% and 12% Black; in 2023-24 they are 13% and 11%.

These schools are not who they were

The scores on this page are pooled over 2009-2018, grades 3-8, and so are the demographics SEDA ships with them. Intakes have moved a long way since. Measured the same way in 2023-24 by the federal count:

school	% Black, 2009–18	% Black, 2023-24	change
Marian Anderson	86%	41%	-45
Bache-Martin	77%	43%	-34
PS 282 Park Slope	64%	31%	-34
Greenfield	41%	12%	-29
Lea	87%	59%	-28
PS 9 Teunis G. Bergen	55%	32%	-24
PS 11 Purvis J. Behan	75%	52%	-23
Penn Alexander	33%	13%	-21

The eight largest falls, in percentage points. Every full pair is in the tables above. This is why the two columns are shown side by side rather than the older one alone: a reader taking the SEDA-era composition for the current one would be reading a school that no longer exists.

Caveats

Scores and demographics are from different decades. The level and growth estimates can only be as current as SEDA's release, which pools 2009-2018, grades 3-8. The 2023-24 composition is shown beside them because it is available and it has moved, but it cannot be used to reinterpret scores that were measured before it. They are two facts about the same building, not one measure.
Not every school here is an elementary school. The Philadelphia set is K-8 buildings; the Brooklyn set is one family’s school and its neighbours, and one of those is a middle school. The grades column in each table says which, and a middle school's scores are not comparable to a K-5's.
New York is not in the district table. SEDA's 2022 district release contains 0 rows for any New York district — the state is absent from that release — so the pandemic-drop comparison is Philadelphia and Tuscaloosa only.
New York's schools are filed under one citywide id. SEDA rolls all 32 geographic districts into a single LEA, where the federal collections keep them separate. The school number is the same in both, so the ids crosswalk; the join is checked by confirming every school lands in that citywide LEA.
The school-level data ends in 2018. SEDA's only public school-level release covers 2009-2018, grades 3-8; the current release is behind a registration wall. The district table is the up-to-date part.
Level is not quality. It correlates with student economic disadvantage at r = −0.81 across US schools. Growth is closer to a school effect (r = −0.12 with the same measure), but student mobility and selection still affect it.
Philadelphia's district figures exclude charters, which are separate districts in Pennsylvania and enroll roughly a third of the city's public school students. If charters draw stronger students, the district number understates Philadelphia.
The low-income measure is not comparable across districts. SEDA's perecd comes from free-and-reduced-lunch eligibility, and a district serving universal free meals books nearly every school at the maximum whatever its families earn. Philadelphia has done that district-wide since 2014, which is why 64% of the West Philadelphia eleven are recorded at 95% or above against 8% of US schools, and Tuscaloosa City has 1 school at that ceiling. Nothing in these files avoids it: there is no income measure in the school-level release that is not built from meal eligibility. The at ceiling column above is there so a reader can see which rows have run out of room, and percent Black is reported beside it in the tables because all three states collect it the same way.
Growth estimates are noisier than levels, especially at small schools. Powel has no growth estimate at all and is omitted from the chart.

Built from the Stanford Education Data Archive (SEDA 4.1 school file and SEDA 2022 v2.0 district file), archived on Stanford's digital repository. Analysis code in advice/sdp-retention. Teacher retention · Chronic absenteeism.