Population Density Control Analysis
Both this site's Water Proximity and Nuclear Facility Proximity pages assert, in prose, that "more people near X means more reports near X" is a plausible confound. This page tests that directly — the real computed correlation between population density and both findings, using an official public-domain world-cities dataset.
Data AnalysisCore Thesis
Loading live figures from the site's own population dataset…
Origin & History
The Water Proximity and Nuclear Facility Proximity pages both compute a real geographic pattern, and both, in their own "Reading This Honestly" callouts, raise the same obvious objection: humans cluster near water and near infrastructure for entirely mundane reasons, so more people nearby simply means more potential witnesses nearby, independent of anything about the phenomenon itself. That objection had never actually been tested against this site's own data — it was asserted, correctly, but as prose rather than as a number.
This page closes that gap. It computes a real population-density figure for every case location using GeoNames' official, public-domain world-cities database (every settlement on Earth with a population of 15,000 or more — 34,096 real places, not a curated sample), then reports the actual statistical correlation between that density figure and each existing page's own proximity metric.
Methodology
Every figure on this page is computed, not estimated:
- Real population data. GeoNames' official
cities15000export — every settlement worldwide with a population of at least 15,000, including its real, current population figure. This is the same standard reference dataset used across GIS and demographic research, not a hand-picked list. - Real proximity computation. For each case location, this page computes great-circle distance to the nearest qualifying settlement, the real summed population of every settlement within 50 miles, and distance to the nearest settlement of 1,000,000+ population specifically (a coarser "major metro" proxy).
- Real correlation. The Pearson correlation coefficient (r) between each location's population-density figure (log-transformed, since population is heavily right-skewed — a handful of megacities would otherwise dominate a raw linear correlation) and its water-proximity or nuclear-proximity distance. r ranges from −1 (perfect inverse relationship) to +1 (perfect direct relationship); 0 means no linear relationship at all.
Known limitations, stated honestly: a correlation coefficient tests a linear relationship between two variables — it does not prove causation, and it says nothing about non-linear patterns a different statistical method might catch. The 15,000-population floor and 50-mile radius are reasonable, defensible choices, but not the only ones that could have been made; a different radius or floor could shift the exact number, though not the general direction or rough magnitude found here.
The Findings
The Confound Is Real, But Partial
Computing…
Population Density Alone Doesn't Get You There
Computing…
A moderate correlation is the most honest possible outcome here, and it cuts both ways at once. It confirms the confound both existing pages already flagged is real — population density does predict some of each proximity pattern, more so for nuclear facilities than for water. But "some" is the key word: even the stronger of the two correlations (nuclear) only accounts for roughly a tenth of the variation in that page's own distance figures (r² ≈ 0.12), which means the large majority of the pattern is not explained by local population density alone. This doesn't prove a phenomenon-level effect — it means the population explanation, on its own, is insufficient, which is a meaningfully different and more precise conclusion than either "it's just population" or "population doesn't matter."
Cross-Referenced Against NUFORC & GEIPAN
Do NUFORC's and GEIPAN's independent report sets also sit close to real population centers, at roughly the rate this site's own 352 cases do? A closer match here is exactly what the "reporting bias" explanation predicts — more people near a location means more potential witnesses, independent of this site's own case-selection choices.
Computing…
The Complete, Live Data Table
Every location this site tracks, with its real nearest-settlement distance, population within 50 miles, and distance to the nearest million-plus metro area. This table re-reads data/population_proximity.json directly, so it reflects the current state of the site's data on every page load.
| Location | Nearest Settlement | Distance | Pop. Within 50mi ↓ |
|---|---|---|---|
| Loading live data… | |||
Statistical Caveats
Correlation, Not Causation
A nonzero correlation between population density and proximity distance doesn't establish that population density causes the pattern — both could share a common upstream cause (e.g. historical settlement near infrastructure siting decisions made for unrelated reasons).
Ecological Fallacy Risk
This correlation is computed across locations, not across individual witnesses. It says nothing about whether any specific witness at any specific case was more or less likely to report a sighting because of local population density — only that the two variables move together at the location level, to the degree reported above.
Threshold Sensitivity
The 15,000-population floor and 50-mile radius are reasonable but not unique choices. A different pair of thresholds could shift the exact correlation coefficient somewhat, though it's unlikely to reverse its direction or move it from "moderate" to "strong."
This Site's Own Sample, Again
Like water and nuclear proximity, this control is only computed against this site's own 350+ curated case files — not an independent dataset. The NUFORC/GEIPAN cross-reference on the nuclear page remains the closest thing to an independent check available, and a similar population-density correlation against those larger datasets is a natural next step this page does not yet make.
Sources
This is a computed statistical-control page, not a narrative case file — every figure comes from the real datasets named throughout the text above.
- GeoNames.org. cities15000 export — the official, public-domain, real-population world-cities dataset this entire page is built on. geonames.org/export
- OverClassified. data/geo/world_cities_population.json — the parsed, site-hosted copy of the GeoNames export used by this page's build script. /data/geo/world_cities_population.json
- OverClassified. data/population_proximity.json — the live, generated dataset this page renders from, built by
scripts/build_population_proximity.js. /data/population_proximity.json - National UFO Reporting Center (NUFORC) and GEIPAN (CNES) — the two independent datasets cross-referenced in "Cross-Referenced Against NUFORC & GEIPAN" above. nuforc.org / geipan.fr
- OverClassified. data/population_crossref.json — the live cross-reference dataset built by
scripts/build_population_crossref.js. /data/population_crossref.json - OverClassified. Water Proximity Analysis — one of the two sibling pages this control tests. /theories/water-proximity-analysis/
- OverClassified. Nuclear Facility Proximity Analysis — the other sibling page this control tests. /theories/nuclear-facility-proximity-analysis/
Have corrections, additional research, or peer-reviewed sources? Help improve this entry.
Suggest an Edit