Methodology & Data Sources

Data Sources

Every article on Areazine is derived from one of the official government feeds listed below. We do not use third-party aggregators or social-media posts as article sources. Each feed has its own polling schedule, and every published article links to its original agency notice.

How We Process the Data

Areazine operates a continuous editorial pipeline that ingests official government feeds and publishes plain-language summaries of each notice. The pipeline runs 24/7 and follows this sequence:

1. Each government feed is polled on its configured interval (30 minutes to 12 hours, depending on the source agency and feed)
2. New records are compared against our database to identify notices not yet published, using unique identifiers from each agency (recall numbers, alert IDs, earthquake event IDs)
3. The source notice is reformatted into a plain-language summary, preserving every factual detail (product names, dates, locations, classifications, magnitudes) without adding speculation or altering numbers
4. Each summary passes through a structured fact-verification step that confirms the key data points from the source notice appear accurately in the published version
5. Summaries that fail verification are rejected and flagged for editorial review rather than published with potential inaccuracies
6. Verified summaries publish to the site as they clear review, and URLs are submitted to search engines via IndexNow for rapid indexing

City Profiles & Computed Percentiles

City profiles combine Census place population and American Community Survey place-level household data with CDC PLACES and CMS Hospital Compare records. A percentile is a corpus-relative placement, not a safety score or recommendation. For population, Areazine divides the number of tracked cities with a strictly smaller positive population by all tracked cities with a positive population. The income calculation uses the same method across profiles with a reported positive median household income. Results are rounded to the nearest whole percentile; tied values therefore share a placement.

“Top” and “bottom” bands describe only where a value sits inside the current Areazine corpus. They do not establish cause, quality of life, danger, or how a city compares with places outside that measured set. The profile states the peer count and source beside each placement so readers can interpret the result in context.

Update Frequency & Data Freshness

Each source is polled on a schedule matched to the agency and feed, ranging from 30 minutes to 12 hours. The per-source list above shows the configured schedule. Publication timing also depends on the upstream agency's availability and our validation pipeline, so a polling interval is not a promise that a notice will appear immediately.

We do not retroactively modify published articles when agencies update their notices. If a recall is expanded or a weather alert is upgraded, a new article is published reflecting the updated information, with a link to the original.

Editorial Standards

• Every article links directly to its original government source
• We report facts from official data, we do not editorialize, speculate, or interpret
• Source agency attribution is displayed on every article
• Severity levels are derived from official agency classifications, not our own assessments
• Articles are time-stamped with the original alert issuance time, not our publication time

Limitations

• Areazine covers only a subset of all government safety publications, we focus on the highest-volume, most publicly relevant feeds
• There may be a delay between an agency publishing a notice and it appearing on Areazine because of source availability, polling cadence, and validation
• Article text is a plain-language transformation of the source notice, not the official notice itself, always follow the source link for authoritative information
• This site is for informational purposes only and is not a substitute for official emergency communications
• Recall counts (FDA, CPSC) reflect reported and classified incidents, not the true incidence of a product or drug defect. Reporting volume is shaped by manufacturer/retailer disclosure practices, media attention, and each agency's own investigation capacity, not solely by how common or severe a defect actually is. A higher recall count for one category or company does not by itself mean it is more dangerous than a category or company with fewer recalls, and a location with more articles is not necessarily less safe than one with fewer, it may simply have denser reporting or a larger population.

Data Accuracy Commitment

Every article links directly to its original government source so readers can verify the information independently. We treat accuracy as non-negotiable: if our validation pipeline cannot confirm that an article faithfully represents the source data, it is not published. We do not prioritize speed over correctness. If you find an error in any article, please contact us immediately and we will investigate and correct it.

State profile CSV

The same state_profiles extract that powers /statistics is published as a free CSV at /data/areazine-state-profiles.csv (CC BY 4.0). Rows are one U.S. state each with city_count, total_population, avg_median_household_income, and avg_median_home_value. The income and home-value columns are unweighted means of city medians, not state medians: a state with many small cities weights them equally with its largest. When reusing the table, name Areazine as the compiler.