Measuring Access

Codebook

What every variable means, how it was built, where it comes from, and the decision behind it

Back to the dashboard

This codebook covers every column in the dashboard's downloads. Each variable lists its name in the file, what it holds in plain words, how it was built, its source, and the numbered decision behind it. The decision numbers (D061 to D074) refer to the MAM census database's decision log.

  1. How the data were built
  2. Districts and IDs
  3. Margins of error
  4. Indicators
  5. Funding inputs
  6. Your results
  7. Map and neighbors
  8. Data files and fingerprints

How the data were built

Every step reads only files whose fingerprint (SHA-256) matches the one recorded when the file was downloaded or made. A step stops if any input does not match, and it never overwrites a file.

  1. Download. 01_download_acs.sh pulls American Community Survey 5-year tables from the Census Data API for Mississippi and Tennessee school districts. The dashboard uses the 2020-2024 estimates. Other downloads: Census TIGER/Line 2024 school district boundaries (04_download_tiger.sh), Tennessee district profiles (05_download_state_files.sh, saved from a browser when the site refused scripted downloads), and, saved from a browser, the Mississippi enrollment file, the Mississippi school meals (CEP) list, the NCES directory, and USDA's SNAP-Authorized Retailer Access Map.
  2. Clean. 10_build_acs_long.R turns each raw ACS file into a long table with one row per district and variable: the estimate, its 90% margin of error, and any Census special code (D067).
  3. Build. Stata 19 do-files make the dashboard's data (D067):
    • 30_dashboard_districts.do: the 277 districts, their IDs, K-12 enrollment, and who the state counts now.
    • 31_dashboard_indicators.do: 17 census indicators with margins of error.
    • 32_dashboard_map.do: the simplified district shapes.
    • 33_dashboard_food.do: the food access indicator.
    • 34_dashboard_neighbors.do: which districts border each other.
  4. Check. Each product was recomputed separately in Python, cell by cell, before it went on the dashboard.
  5. Show. The dashboard's files in data/ are exact copies of those products. The index, re-sharing and dollars are worked out in the browser (D071).

Districts and IDs

The dashboard shows the school systems each state's formula pays (D063, D068).

Name in fileWhat it isHow it was builtSourceDecision
stateState: MS or TNFrom the first two digits of the federal ID (28 is Mississippi, 47 is Tennessee).CensusD063
leaidFederal district ID (NCES LEA ID), 7 digitsThe Census school district code, which is the same as the NCES ID. Kept as text so leading digits are never lost. Always included in downloads, so rows can be matched to other files.Census ACS and TIGER/Line, NCESD063
st_noState district numberst_leaid without the state prefix and leading zeros. MS-0200 is MDE district 200, TN-00751 is TDOE district 751.NCES CCD 2024-25 preliminary directoryD068
st_leaidState LEA ID as NCES lists itAs published.NCES CCD 2024-25 preliminary directoryD068
name_censusDistrict name as the Census writes itAs published, with the state name at the end.Census ACSD063
name_stateDistrict name in the state's fileAs published by TDOE or MDE.TDOE district profile, MDE enrollmentD066
name_ccdDistrict name in the NCES directoryAs published.NCES CCD 2024-25D068
district_typeCensus district type: unified (K-12) or elementary (K-8)From which Census district file the district comes from. Only Tennessee's 16 city systems are elementary.Census ACSD063
districtDistrict name (in your results)The Census name without the state at the end.Census ACSD063

Margins of error

The ACS is a survey, so every estimate has a margin of error. Each _moe column is the 90% margin of error: the true value is likely within that many points of the estimate. Small districts often have wide margins.

Decision: D069.

Indicators

Each indicator has two columns: _est (the value) and _moe (its margin of error). Seventeen are from the ACS 2020-2024 5-year estimates (D069). Food access is from USDA (D073). Higher means more need for every indicator except median household income, which is flipped so lower income counts as more need. Groups follow the dashboard's domains (D072, D073).

Income and work

Name in fileWhat it isHow it was builtSourceDecision
inc185Kids ages 6 to 17 in families below 185% of the poverty line (%). 185% is the reduced-price school meal line.Built from counts. Numerator: kids 6 to 11 and 12 to 17 with an income-to-poverty ratio under 1.85 (the seven groups from "Under .50" to "1.75 to 1.84", for each age group). Denominator: all kids 6 to 11 plus 12 to 17. Times 100.ACS B17024_016 to _022 and _029 to _035, over B17024_015 + B17024_028D069
pov517Kids ages 5 to 17 below the poverty line (%)As published.ACS S1701_C03_004D069
unempUnemployment rate, people 16 and over (%)As published.ACS S2301_C04_001D069
mhhiMedian household income (2024 dollars)As published. On the dashboard it is flipped before scaling, so lower income counts as more need. Downloads show the value before flipping.ACS B19013_001D069
giniGini index of income inequality, 0 (everyone equal) to 1As published. It measures how spread out incomes are in the district, not how many families are poor. Higher counts as more need. Users choose whether to use it.ACS B19083_001D069

Food

Name in fileWhat it isHow it was builtSourceDecision
foodkidsKids ages 0 to 17 who live far from a store that takes SNAP (%). Far means more than 1 mile by road in an urban census tract, or more than 10 miles by road in a rural one.For each tract, kids with low access is the 1-mile count if the tract is urban and the 10-mile count if it is rural. Each tract's kids with low access and all kids are split across districts by the share of the tract's people living in each district, then added up. Value = 100 × kids with low access / all kids. This treats kids as spread like everyone else within a tract. USDA's own "kids share" columns divide by a tract's total population, so they are not used. Driving distance was chosen because straight-line distance puts almost every rural tract at zero. The kids, all-residents and low-income versions are nearly identical within tracts (correlation 1.00), so only kids is used. Six districts are at 0.USDA ERS SNAP-Authorized Retailer Access Map, July 2026 release (stores as of June 2025, 2020 Census, 2020 tracts): DD_SRAM_lakids1, DD_SRAM_lakids10, TractKids, Urban. Team tract-to-district crosswalk (share_of_tract_pop).D073
snapHouseholds that received SNAP (food stamps) in the past 12 months (%)As published.ACS S2201_C04_001D061, D069

Health and housing

Name in fileWhat it isHow it was builtSourceDecision
uninsKids under 19 with no health insurance (%)As published.ACS S2701_C05_011D069
crowdHouseholds with more than 1 person per room (%)Built from counts. Numerator: owner and renter households with 1.01 to 1.50, 1.51 to 2.00, or 2.01 or more people per room. Denominator: all occupied housing units. Times 100.ACS B25014_005, _006, _007, _011, _012, _013, over B25014_001D069
plumbHouseholds lacking complete plumbing (%)Built from counts. Numerator: owner and renter households lacking plumbing facilities. Denominator: all occupied housing units. Times 100.ACS B25049_004 + _007, over B25049_001D069
rentburdenRenter households paying 35% or more of income on rent (%)Built from counts. Numerator: renters paying 35.0 to 39.9%, 40.0 to 49.9%, or 50% or more. Denominator: all renter households minus those whose share could not be computed. Times 100.ACS B25070_008 + _009 + _010, over B25070_001 minus B25070_011D069
movedPeople 1 and over who lived in a different house a year ago (%)Built from counts. Numerator: everyone 1 and over minus those in the same house a year ago. Denominator: everyone 1 and over. Times 100.ACS B07003_001 minus B07003_004, over B07003_001D069

Getting around and getting online

Name in fileWhat it isHow it was builtSourceDecision
novehHouseholds with no vehicle available (%)As published.ACS S2504_C02_027D069
nobbHouseholds without a broadband internet subscription of any type (%)100 minus the published share with broadband of any type. Keeps the published margin of error.ACS S2801_C02_014D069
nocompHouseholds without a computer (%)As published.ACS S2801_C02_011D069

Family and language

Name in fileWhat it isHow it was builtSourceDecision
nohsAdults 25 and over without a high school diploma or equivalent (%)100 minus the published share who are high school graduates or higher. Keeps the published margin of error.ACS S1501_C02_014D061, D069
lepPeople 5 and over who speak English less than "very well" (%)As published.ACS S1601_C06_001D061, D069
singleparFamilies with their own children under 18 headed by a single parent (%)Built from counts. Numerator: male and female householders with no spouse present, with their own children under 18. Denominator: those plus married couples with their own children under 18. Times 100.ACS B11003_010 + _016, over B11003_003 + _010 + _016D069

Other

Name in fileWhat it isHow it was builtSourceDecision
kids617Children ages 6 to 17 (a count, not an indicator)Kids 6 to 11 plus kids 12 to 17. Margin of error from the sum formula. Kept for context, not offered as an indicator.ACS B17024_015 + B17024_028D069

Funding inputs

Who each state counts now, and the rules in law (D064, D066). Both states fund a year on the prior year's students, so 2024-25 counts feed 2025-26 (Tennessee) and FY26 (Mississippi) money.

Name in fileWhat it isHow it was builtSourceDecision
enroll_k12K-12 enrollment, 2024-25Tennessee: the K-12 total in the district profile. Mississippi: total enrollment minus Infant/Toddler, Pre-K, Sped Pre-K, Grade 13, Adult Education and Secondary GED. A starred cell among those (fewer than 10 students, hidden by MDE) is counted as 0, so a Mississippi count can be high by up to 9 per starred cell.TDOE district profile 2024-25, MDE enrollment downloadD066
base_pctShare of students the state counts now (%)Tennessee: the economically disadvantaged percent, which TDOE publishes as a whole number, so a district can be off by up to half a percentage point. Mississippi: the district-wide identified student percentage from the school meals (CEP) list, data as of April 14, 2025. This is the school meals measure, close to but not confirmed identical to the low-income count MSFF uses.TDOE district profile 2024-25, MDE Office of Child Nutrition CEP listD066
base_countStudents the state counts nowenroll_k12 × base_pct / 100.BuiltD066
base_measureWhich measure base_pct isText describing the measure and file.BuiltD066
enroll_sourceWhich file enroll_k12 is fromText describing the file and what was subtracted.BuiltD066
ms_starred_cellsMississippi: hidden enrollment cells counted as 0How many of the subtracted groups were starred for the district. Empty for Tennessee.MDE enrollment downloadD066
cep_id, cep_nameMississippi: the district's ID and name in the school meals listMatched on the MDE district number. West Tallahatchie is listed under meal-program ID 00149 and was matched by name. Empty for Tennessee.MDE CEP list (PDF, read with pdfplumber)D066
formula, formula_yearThe state formula and yearTISA 2025-26 (Tennessee), MSFF FY26 (Mississippi).State lawD064
base_amountBase amount per student, in law (dollars)Tennessee $7,295. Mississippi $6,842.61, from George County School District's FY26 budget hearing (July 29, 2025), not yet confirmed in an MDE document.TISA Overview (April 2026), George County SD FY26 budget hearingD064
poverty_weightExtra share of the base for each counted studentTennessee 0.25 (economically disadvantaged). Mississippi 0.30 (low-income).State lawD064
concentration_weight, concentration_lineMississippi's concentrated poverty ruleMississippi: 0.10 of the base for each low-income student above 0.35 (35%) of enrollment. Tennessee: 0 in this file. Tennessee's 5% concentrated poverty weight is set school by school, so it is not modeled and stays at its current amount.State lawD064, D065

Your results

Computed in the browser from your choices (D064, D065, D071). Each state is ranked on its own. With both states in one file, each state's rows are still ranked within that state.

Name in fileWhat it isHow it was builtDecision
val_indicatorThe district's value on each indicator you choseThe _est value from the indicators file, before flipping.D069
sc_indicatorThat value after scalingMedian household income is flipped first (multiplied by −1). Then, across the state's districts:
  • Ranks: the share of the other districts with a lower value, times 100.
  • 0 to 100: (value − lowest) / (highest − lowest) × 100.
  • z-scores: (value − state mean) / state standard deviation (population formula).
The same rules as the Arizona Student Poverty Index Lab.
D069, D071
w_indicatorYour weight on that indicatorPoints out of 100.D064
scalingThe scaling method you choserank, minmax or z.D071
indexYour index of needThe sum of weight × scaled value, divided by 100, rounded to 9 decimals so districts with equal scores tie exactly.D071
rank_nowRank by the share the state counts now1 is the highest base_pct in the state. Exact ties go to the district listed first alphabetically.D065
rank_yoursRank by your index1 is the most need. Exact ties go to the district listed first alphabetically.D071
rank_movePlaces moved uprank_now − rank_yours. Positive means the district ranks as higher need on your index than on the state's measure.D065
count_nowStudents the state counts nowSame as base_count.D066
count_yoursStudents counted after re-sharing by your index
  1. Districts are ordered by your index, highest need first.
  2. The state's actual district rates (base_pct) are sorted from highest to lowest and handed out in that order.
  3. Each rate is multiplied by the district's K-12 enrollment.
  4. The counts are scaled so the state total equals the actual total. A district that would go above its enrollment is held at its enrollment and the rest are scaled again.
Your index decides the order. The real spread of rates stays the same.
D065
dollars_nowPoverty-weight dollars nowcount_now × poverty weight × base amount. Mississippi adds max(0, count_now − 0.35 × enrollment) × 0.10 × base amount.D064
dollars_yoursPoverty-weight dollars under your indexThe same formula with count_yours. In Mississippi the state total can shift a little because the 35% rule is recomputed. In Tennessee the total stays the same.D064, D065
changeChange in poverty-weight dollarsdollars_yours − dollars_now.D064
change_per_studentChange per K-12 student (dollars)change / enroll_k12.D064

Map and neighbors

Data files and fingerprints

The files in data/ are exact copies of the database products. A file's fingerprint (SHA-256) changes if even one character changes.

FileMade bySHA-256
districts.csv30_dashboard_districts.doea7759ca0e7df88ff6144a75b78101f5ac6504cd1da000c3875a6c53f9db0f8c
indicators.csv31_dashboard_indicators.do8c870d39bc5a687368738076dbd63eee73402c34c9a0ec67f26b6c155395010b
map.geojson32_dashboard_map.do1101a3c55047a759cff6ddda4b04fe54520e19779fc23b4d85456ebddb90cc02
food.csv33_dashboard_food.do09e699a888f049d0f6bae32380924b46135f9444dec8a0bb8c8c63355933f6ef
neighbors.csv34_dashboard_neighbors.do06b550f90b61e61bfc12ec76f108aabd5f7dbe9e7cc511350d69a2a27ad19574

Full sources, with links, are in the README.