Employer evidence methodology, source periods and refresh schedule

An employer filing, a labor certification, a visa approval, a job advertisement, an offered job and an actual hire are separate events. DOL determinations and positive LMIAs are government decisions on labor-market processes. The USCIS sponsor directory measures petition decisions. Job Bank and SeasonalJobs observations are advertisements or filing/acceptance events. None is counted as an actual hire. A decision-year trend can also reflect processing speed and backlogs, not only new recruitment.

DOL case numbers are deduplicated across quarterly files; the latest observed determination is retained. Historical LCA annual totals require all four quarterly files, while current cumulative releases are checked for decisions spanning their declared quarters. Complete fiscal years are compared with complete fiscal years. Canadian momentum compares the same calendar quarter one year apart. Certification totals include expired or returned certifications as historical events, and exclude withdrawn records. PERM's two FY2024 form layouts are combined by case number.

U.S. employer groups use reported name and employer state; Canadian LMIA groups use name, province and reported location. Punctuation/case normalization does not roll up affiliates, franchisors or renamed entities. First observations are bounded by stated historical coverage. Job Bank name/province joins remain candidates, and enforcement records remain separate.

USD wage medians use the lower offered amount with at least five comparable observations, equally weighted by case. Full-time hourly LCA offers use 2,080 hours, weekly offers 52 weeks, biweekly offers 26 periods and monthly offers 12 months. PERM uses annual amounts as reported. Seasonal hourly offers stay hourly, and piece rates/unknown units are excluded from hourly medians. Canadian advertised wages stay CAD in their original units.

The mini collects rolling SeasonalJobs feeds, sampled Job Bank postings and Canadian compliance records weekly; checks Canadian LMIA releases monthly; and processes DOL releases quarterly. Source records, hashes and observation versions are retained outside the public repository. A failed source retains its previous successful data date; unavailable sources and archive gaps remain explicit. Raw government releases can lag by months.

All included dataset calculations passed their coverage checks.