The documentary foundation
The Sources Behind the Range
Original agency productions, preserved portal records and archived PDFs form the dataset. Every case leads back to its recorded sources. Every plotted observation leads to its amount, passage and original document.
The Agency Record and the Preserved Archive
The case explorer contains 83,809 exact LWDA case identifiers and 243,631 source-row associations. Original DIR productions, preserved portal snapshots and later scraper acquisitions remain separately inspectable.
43,622 case-document entries now link to complete originals verified by full anonymous download and SHA-256. 42,468 distinct originals also have searchable source text. Source availability and text-processing status are displayed on each document.
The settlement distribution retains its separately checked extraction cohort. Full-archive text and newly found dollar passages do not by themselves establish a new settlement observation.
What the Distribution Measures
The default distribution contains 1,419 document-stated gross settlement observations. The preserved historical cohort contained 1,105 observations. The accepted full-archive review ledger records 314 additions, 59 replacements and 0 removals; each change retains its review and source evidence.
Each newly accepted or corrected amount is checked against the unchanged original PDF at a cited physical page, with an amount-supporting quotation, original SHA-256, exact case association, claim-scope review and economic-settlement duplicate review. OCR-supported passages require completed review of the rendered original page. Material conditions remain in the observation notes. Candidate amounts, unresolved ranges, caps, formulas and citations to other cases are not automatically included.
Gross amounts can cover class, collective and individual claims together with PAGA. The stage is document-stated; the amount does not establish approval, payment, a PAGA-only allocation or net employee recovery. Source excerpts normalize whitespace and omit standalone pleading line numbers; the PDF controls. The complete archive is broader than this accepted observation cohort, which is not a census of settled matters or a prediction of a particular outcome.
Industry remains unclassified without source support. Source filename dates retain that limited basis. Agency statuses, penalties, balances, fees and aggregate collections are not gross settlement observations. Personal imports remain a separate workspace.
Inspect accepted review provenance and prior-record bindings · Download the source-linked settlement observations
Later Portal Acquisitions
6,091 records refreshed since the preserved July 15 database. Per-record timestamps are included; this is not a claim that the entire portal was refreshed today.
Open retained portal rowsSHA-256 · 078fba98019cf2e16d94bbcf8c476673038ad846721edb5f37aab78ce7c554c7
Original Sources
2026-07-30 — DIR HQ64881 — Part 1 — PAGA Notices Filed (2004-2026).xlsx
Notice records, 2004–July 30, 2026; administrative notices, not lawsuits or outcomes.
2026-07-30 — DIR HQ64881 — Part 1 — RLLVs (2016-2026).xlsx
Report of Labor Law Violation submissions, September 6, 2016–July 29, 2026. Selections are allegations, not findings.
2026-07-30 — DIR HQ64881 — Part 1 — Notice of Investigations.xlsx
Request-specific investigation records; LWDA and BOFE identifiers. Not a statewide outcome cohort.
2026-07-30 — DIR HQ64881 — Part 1 — PAGA Issues and Citations.xlsx
Citation issues, State penalties owed, and balances. Case-level amounts repeat across issue rows; these are not gross settlements.
2026-08-07 — DIR HQ64881 — Part 2 — Fees and Penalties (Jul 2016-Jun 2026).xlsx
120 filing-fee and 126 penalty-collection month rows through June 2026. No case identifiers or gross-settlement definition.
Preserved LWDA public-portal scraper snapshot — July 15, 2026
Sanitized CSV of all 79,503 preserved notice records. Business, case, geography, firm, original scrape/update timestamps, and source-row keys retained; employee/claimant and individual filer fields omitted. Source database checksum separately retained.
Active saved LWDA portal PDFs
30,090 preserved original files, including 30,029 PDF-format originals and 61 non-PDF source attachments. Complete resident bytes have been hashed; public links appear after anonymous byte verification. Archives overlap; these physical-file counts are not distinct-document totals.
Historical saved LWDA portal PDFs
19,197 preserved original files, including 19,130 PDF-format originals and 67 non-PDF source attachments. Complete resident bytes have been hashed; public links appear after anonymous byte verification. Archives overlap; these physical-file counts are not distinct-document totals.
Preserved LWDA document inventory — cases without a notice snapshot
19 document records for cases identified by exact LWDA numbers in archived filenames, with original filename, AP identifier, recorded size/hash and availability. This preserved inventory records the earlier metadata snapshot. The case explorer shows the subsequently recovered originals, their verified hashes and current publication status.
Source Relationships and Coverage
- Case identities join on exact LWDA case-number strings. Original agency rows and historical portal metadata remain separate source snapshots.
- Allegations use the latest RLLV create date per case, with the later worksheet row breaking ties; selections are allegations, not findings.
- Case status is the source’s administrative label, not a verified court-approval or payment-completion determination. Notice dates are not settlement dates.
- Document date is null unless separately verified. Download timestamps and saved filename dates are not promoted to filing or approval dates.
- Original claimant/employee and individual-filer name fields are omitted from public scraper metadata and the searchable index.
- Original agency workbooks are preserved byte for byte. Aggregate fees, penalty collections, citation liabilities and balances are not used as gross settlement observations.
- Investigation rows join by exact LWDA ID. Citation rows join only through the exact BOFE ID recorded in the investigation production; the bridge row is retained. A citation row can relate to multiple LWDA notices. Unmatched citation rows remain in the complete original workbook. State penalties and balances remain separate from settlement observations.
- Later scraper records retain exact case IDs, original database row keys and their actual stored timestamps. Newer nonempty values supply display metadata without removing the original agency rows.
- Original archive files are hashed from complete resident bytes. Hosted links are enabled only after full anonymous download and SHA-256 comparison. Identical bytes within a case share one document entry with all filename and submission aliases retained.
- Full-archive text processing is separate from reviewed settlement observations. Combined processing totals count pages and source text segments, including non-PDF attachments and recovered PDF objects. Native text and OCR retain physical page numbers where the original has readable pages. Text recovered from an incomplete PDF is labeled by source-object segment, without asserted original pagination; the unchanged source controls. New monetary candidates are not automatically included in the distribution.
Agency notice export through July 30, 2026. Historical portal database snapshot July 15, 2026. Later observations retain their acquisition timestamps. The current source manifest identifies the included scraper batch, archive verification and text-processing coverage.
Sample Quantiles
The minimum, maximum, median and quartiles use the documented dollar amounts in ascending order. For the median, an even number of records uses the arithmetic midpoint of the two central amounts. Quartiles use type 7 linear interpolation: for probability p, the zero-based position is (n − 1)p; the result interpolates between the two neighboring amounts. Calculations retain unrounded values; monetary display rounds to cents. The chart places these dollar results on its logarithmic axis afterward.
Kernel Density Plot
Each eligible ridge applies a Gaussian kernel to log10(amount), with the displayed bandwidth measured in log10-dollar units. The default 0.17 is a display choice; it is not a fitted industry parameter. A wider bandwidth smooths more; a narrower one can expose small sample fluctuations. Each curve is divided by its own evaluated maximum for drawing. Ridge height therefore compares shape, while the accompanying count reports sample size. A group with fewer than three records, or only one distinct amount, uses exact-value markers instead of a density curve. This threshold is an interface choice, not a claim that three observations establish a reliable population distribution.
Logarithms and Exponentials
The dollar axis uses base-10 logarithms: moving one unit to the right multiplies the amount by ten. This makes amounts across a wide range readable together. Zero and negative amounts have no place on this axis and are rejected explicitly. Positive amounts remain intact; the view expands to include them. Smoothing takes place in log space, so its shape is not a probability density per dollar. The exact record table remains the reference for the documented amounts.