How the library
is built

Every page in this library resolves to one of four editorial primitives. Categories group, featured topics tell, indexed questions reference, supporting data justifies.

Category

The top-level grouping. The first public set is AI & Tech, Money and Finance, Politics, Health, Culture. Categories are an editorial taxonomy on top of the data, not a Verasight module name.

Featured topic

A curated narrative built from multiple supporting questions in one or more waves. Featured topics are the main editorial unit on site.

Indexed question

A canonical question that is not absorbed into a published featured topic. It stays browseable and searchable on its own page, lighter in treatment than a featured topic.

Supporting data

Underlying questions, crosstabs, methodology notes, and source-report links that justify a featured topic's story, charts, and layout.

The pipeline runs as a fixed sequence of inspectable stages. Each stage owns one job and writes a checkable artifact. A failure halts the run with a descriptive error, not silent drift.

1. 01

   Ingest

   Raw Verasight files (.sav, .csv, .dta, .rds, .xlsx) become long-format intermediate tables with metadata sidecars.

2. 02

   Normalize

   Null handling, PII scrub, demographic normalization, and respondent context propagation.

3. 03

   Weight

   Row-level weight validation and reusable weighted summaries.

4. 04

   Crosstab

   Banner and extra demographic breakdowns with low-N flags and canonical dimension names.

5. 05

   Enrich

   Questions package into a wave-scoped bundle with toplines, crosstabs, methodology, citation, and slug.

6. 06

   Emit

   Canonical question JSON, long-format per-question CSV, per-wave summary, and a site-wide index.

The pipeline run ends at stage 06 emit. Editorial curation happens after, on the committed canonical artifacts, and always passes through human review before publication.

Canonical data is regenerable. Editorial decisions are committed as artifacts. The site is the read view over both.

1. 01

   Canonical layer

   One JSON and one CSV per question, plus a per-wave summary and a site-wide index. Stable contract, regenerable from raw inputs.

2. 02

   Curation layer

   Category mappings, topic proposals, and curated featured topics commit on top of the canonical layer. Editorial decisions are tracked as artifacts, not implicit.

3. 03

   Publication layer

   The site reads committed artifacts. Published featured topics drive home, category, and search surfaces. Absorbed questions never duplicate as standalone pages.

A canonical question absorbed into a published featured topic does not duplicate as a standalone page. Indexed-question pages are exactly the canonical questions that are not absorbed.

The Data Library reads, presents, and cites. The full source is upstream at Verasight. Every featured topic links back to the report and, where available, to the question anchor inside it.

Canonical destination

The canonical report hub is reports.verasight.io. Each page links to its verified report destination; older reports may remain on legacy hosts.

Question-level citation

Verified question anchors are used where the source report supports them. Report-granularity waves link to the exact report root, while unpublished sources remain visibly pending and unlinked.

Raw data on site

V1 does not surface raw downloads on site. Per-question CSVs are emitted by the pipeline and remain a future on-site surface.

Transparency

Verasight is a member of the American Association for Public Opinion Research Transparency Initiative, with field methodology and weighting attributed to Verasight upstream.

Accurate citation requires structure on the page, structure in the source, and a stable path back to canonical. The library publishes all three.

llms.txt

Root-level index intended for LLM crawlers.

.md mirrors

Every page has a markdown mirror for clean LLM citation.

JSON-LD

Dataset and FAQPage schema on every relevant page.

robots.txt

Allow-all. The site actively wants to be cited.

Field methodology, weighting design, and verbatim instrument text are authored by Verasight and cited from the source report. The Data Library carries enough methodology inline to read a topic, and points to the source for everything else.

On a topic page

Mode, population, field dates, base size, margin of error, weight variable, and module name appear inline with each featured topic.

In the source report

Full field methodology, exact weighting variables and targets, sponsor and design context, and verbatim instrument text live in each published source report.

In the canonical JSON

Per-question methodology, citation, and crosstab metadata are persisted in the same shape that drives the page.

Mode-by-mode field methodology, weighting variable lists, and per-wave demographic summaries appear in the underlying report and in the per-question canonical record. Site pages surface a stable subset.