Methods and how this was made

Made by AI agents, with a person directing and checking the work.

Question

Since Lawrence's new Land Development Code took effect (April 1, 2025), are permits being issued for the newly allowed housing types: ADUs, duplexes, three- and four-plexes, attached homes and small-lot homes?

Data

• City permit records. We used the full public "Permits" and "Plans" downloads from the City of Lawrence Citizen Connect portal (EnerGov system), retrieved October 25, 2026. The export holds 18,584 permits and 7,015 planning records.
• Census Building Permits Survey (BPS) annual place files for 2015-2025 and the January-August 2026 file, for Lawrence city (FIPS place 38900).

Steps

1. Keep new-home permits (work classes New Single Family Dwelling, New Two Family Dwelling and New Multi-Family) and residential ADU permits (see assumption A2).
2. Count a permit when it has an issue date and is not void, withdrawn, denied or expired.
3. Assign a housing type from the work class and description keywords (A2).
4. Compare April 2023-March 2025 with April 2025-September 2026 and annualize both periods.
5. Read BPS units and the "months reported" field. A value of 0 means the Census Bureau imputed the year.
6. Present scenarios using local fee data and one cited outside example.

Checks (run before every build: checks/run_checks.py)

1. No silent blanks: required fields are filled and no permit number appears twice.
2. Control totals: for 2021-2024, residential permits issued per year equal the Planning Director's published totals (an external reality check). The BPS file totals are recomputed from the raw files.
3. Provenance: every data file is listed in sources.md, and every output has a sidecar file noting its source.
4. Flags: imputed BPS years are flagged; the before/after counts add up to the permit-level file.
5. History freshness: history/timeline.md is newer than every output.
6. Raw data read-only and no empty files.
7. Privacy: published CSVs contain no addresses, names or descriptions.

Questionable results we flag

• Census BPS for 2024-2026 is imputed and disagrees with city records.
• The 2023 townhome count rests mostly on one project.
• Pending applications include a large 2024 group still "In Review".

How this was made

This site was researched, analyzed and written by AI agents, with a person directing and checking the work. The agents downloaded the public data above, wrote Python scripts (pandas, matplotlib, geopandas) to clean and count it, and drafted the text. Every number on the site is produced by those scripts from the saved raw files. A guardrail script must pass before the site is built, and a security audit runs before anything is published. The process follows the one used for an earlier Lawrence transit-access study. Corrections are logged on the Amendments page.

Assumptions

Dates: work done 2026-10-25 (America/Chicago). Code effective date 2025-04-01. Before window 2023-04-01 to 2025-03-31 (24 months); after window 2025-04-01 to 2026-09-30 (18 months). Data cutoff for counts: 2026-09-30 (last complete month).

• A1 Counted permit = issued date present and status not Void, Void - Withdrawn, Denied, Expired, Plan Approval Expired. Counts are permits (buildings), not units.
• A2 Housing type from the city's work class plus description keywords: ADU = Residential Accessory Structure/Addition/Remodel whose description says "ADU" or "accessory dwelling" (excluding trade work inside an existing ADU); townhome/attached = New Single Family whose description says townhome, attached, zero lot or C3; three/four-plex = description says triplex, 3-plex, 4-plex, fourplex or 3/4-unit (not 13-plex, 14-plex or "C3 units", a builder model name); everything else by work class.
• A3 City records before 2021 are treated as incomplete (legacy migration) and not used for trends. Reality check: 2021-2024 residential permit totals equal the Director's published counts exactly.
• A4 Census BPS: "months reported" = 0 means the whole year is imputed by the Census Bureau. We flag, not correct, these.
• A5 Zoning: the permit "mainzone" field is mostly blank, so we do not attribute permits to specific districts.
• A6 Pipeline: unissued applications filed more than 12 months before 2026-10-25 are labeled possibly stalled.
• A7 Fee scenario uses the median "amount paid" on permits issued after the code change. Fees paid may include more than the building-permit fee.
• A8 South Bend design savings (~$10,000/project) are the city of South Bend's estimate, not a Lawrence figure. The city's own design cost is not estimated (no quote).
• A9 Margin of error: no survey-sample data are used; Census BPS place figures are administrative counts or imputations, not sample estimates, so no MOE is published. Imputed years are flagged.
• A10 Privacy: published permit-level CSV omits addresses, owner/applicant/staff names and free-text descriptions.
• A11 Site name, neutral and anonymous: "Lawrence new housing tracker". Planned address placeholder lawrence-kansas-civic-reports.pages.dev/housing/ (combined hub) (not deployed).