The Research Team
Institute research is produced and reviewed collectively. We publish no by-lined authorship — the unit of accountability here is the dataset, the method, and the citation, not the individual.
Who we are. The Public Service Institute is a small independent research team working alongside two companion projects: the Public Service Index, which maintains the statistics of record, and pubsec/data, which publishes the compiled datasets. All three draw on one data infrastructure; the Institute's role is interpretation — turning compiled records into research that answers questions about how Canadian public procurement actually works.
How we work. Every publication follows the same sequence. First, compilation: source records are collected from official publication systems — the federal proactive disclosure of contracts, Quebec's tendering publication, the CITT registry, BC award records — and normalized into a single corpus, with vendor-name variants and coverage gaps preserved rather than silently merged. Second, computation: every figure in every report is computed from that corpus at build time; nothing is transcribed by hand, and any number in an Institute report can be regenerated from the published datasets. Third, review: each report is read against its numbers by a second member of the team, with particular attention to causal language — the standard is that data supports every quantitative claim, and interpretation is marked as interpretation.
Editorial standards. Three rules govern the prose. The data leads: reports are structured around what the corpus establishes, not around a thesis the numbers must serve. Inference is labelled: where a finding rests on reading rather than measuring, the report says so. Limits are published: coverage gaps, name-variant problems and unmapped records are documented in each report's methodology note, because a statistic whose boundaries are hidden is a statistic misused.
Corrections. Errors of fact or computation are corrected in place and noted at the foot of the affected report. Reports are dated by compilation; figures shift only when the underlying corpus is rebuilt, and rebuilds are themselves recorded in the methodology notes.
Why no names. Think-tank convention attaches research to individuals, which has merits we respect. Our choice runs the other way: the Institute's work is reproducible by design — the same corpus, the same computation, the same figures — and reproducibility is a property of the method, not the authors. Questions about any publication are welcome and are answered by the team that produced it.