The bear correlation

Most arguments about Bigfoot are arguments about single objects. One film, one cast, one photograph, one witness. The bear correlation is different in kind. It is a claim about the whole pile at once, it is quantitative, it passed peer review at a mainstream zoological journal, and it is recent. It is also more modest than the single number that traveled from it, and that number is not the one the author first posted.

The paper

Floe Foxon published "Bigfoot: If it's there, could it be a bear?" in the Journal of Zoology, volume 323, issue 1, pages 1 to 8. It went online January 13, 2024, and appeared in print that May. The Journal of Zoology is the Zoological Society of London's journal, published by Wiley. That the paper exists and passed peer review there is DOCUMENTED

The published article carries the affiliation "Folk Zoology Society, Pittsburgh, PA, USA." The bioRxiv preprints of the same work carry a different one: "Department of Data Management and Statistical Analysis, PinneyAssociates, Pittsburgh, USA."

Foxon's indexed publication record is dominated by nicotine and tobacco epidemiology, in journals including Addiction, Nicotine & Tobacco Research, and Harm Reduction Journal, with separate output in solar physics and a run of cryptozoology papers: the 2023 JMIRx Bio eel paper on Loch Ness, a 2024 Belgian Journal of Zoology paper on moa extinction, and a 2024 monograph, Folklore and Zoology. That is a statistician who treats cryptid reports as a data set. It is also everything this entry will say about the author.

The method states in a sentence. Take the count of sasquatch sightings in each US state and Canadian province, take the black bear population in each, and regress one on the other while holding human population and forest area constant, so that "more people to see things" and "more woods to see them in" cannot do the explaining by themselves.

The inputs, as named in the preprint's own data section:

Two of those deserve a second look, and neither is hidden. The bear numbers come from Table 1 of a survey paper about how agencies respond to human-bear conflict, so they are agency self-reported estimates gathered for a management purpose, not a census. The Canadian forest figures come from a commercial reference website, cited to an archived copy.

Four US states (Rhode Island, Texas, Wisconsin, and Wyoming) and four Canadian jurisdictions (Alberta, Newfoundland and Labrador, Northwest Territories, and Nova Scotia) were dropped for missing bear data. Nine states with no known breeding black bear population (Delaware, Hawaii, Illinois, Indiana, Iowa, Kansas, Nebraska, North Dakota, and South Dakota) were assigned a bear population of zero rather than dropped. Foxon settled on a generalized linear model with a negative binomial distribution and a log link, the standard choice for count data that clumps. DOCUMENTED The paper does not state a final count of jurisdictions in the fitted model, and I could not derive one I would stand behind, so this entry does not give one.

The two numbers

The result reduces to two figures, and it is worth having both, because the second one is the only one most summaries carry.

"On the average, every 1,000 bear increase in the bear population is associated with a 4% (95% CI: 1%-7%) increase in sasquatch sightings." The interval excludes zero, which is what "statistically significant" means here and is the load-bearing claim. DOCUMENTED

"On the average, across all states and provinces in 2006, after controlling for human population and forest area, there were approximately 5,000 bears per sasquatch sighting." DOCUMENTED

Both quotations are from bioRxiv version 3, not from the published article, and that distinction is not pedantry. Wiley's deposited abstract for the published paper has had every numeral stripped out of it by the deposit pipeline. The machine-readable record literally reads "every bear increase in the bear population is associated with a ( CI: -) increase in sasquatch sightings" and "there were approximately bears per sasquatch sighting," with holes where the figures belong. The publisher's full text returns HTTP 403 to unauthenticated requests, so I could not read the printed figures myself. The preprint's wording matches the published abstract word for word around every gap, which is good evidence the numbers carried over unchanged. That is not the same thing as having checked them.

The number moved across drafts

This is the part almost no secondhand account carries.

DOCUMENTED Two things changed, not one. The model family changed between v1 and v2, from a linear mixed-effects model to a generalized linear model. The area covariate changed between v2 and v3, from land area to forest area. Across those revisions the headline ratio moved from 900 bears per sighting to roughly 5,000, a factor of about five, in the single most-quoted quantity in the subject.

The v3 abstract is internally inconsistent about which covariate it used. Its methods sentence says "adjusting for human population and land area," while its results sentence says "after controlling for human population and forest area." The published abstract says forest area in both places, so that inconsistency was caught before print. The closing line was rewritten too. Versions 1 and 2 end "If bigfoot is there, it may be many bears." The published paper ends "If bigfoot is there, it could be a bear."

Nothing improper happened here. Swapping land area for forest area is defensible and arguably better, since bears live in forest rather than in desert or cropland, and revising a preprint is what preprints are for. But anyone citing "one sighting per 5,000 bears" as a hard constant should know it is a fitted ratio at the average that moved by a factor of five when the model specification moved.

What the paper says it does not show

The paper's own hedges are more careful than the ratio that escaped it. Foxon wrote that "the present study does not prove that all sasquatch sightings are bear sightings (nor was this study designed for that purpose)." DOCUMENTED

Three weaknesses are stated in the paper rather than by its critics:

• "The RMSE of the model (19) was high relative to the number of bigfoot sightings in 2006 in each state/province (range: 0-21)." The association is statistically significant. The predictive precision is poor.
• The intercept is positive, so the model expects sightings where the bear population is zero. Foxon allows that this "may be interpreted as evidence for the possible existence of an unknown hominid in North America," and offers misidentification of other animals, including humans, as an alternative.
• "For example, people living, hunting, or hiking in forest areas may also be misidentified as sasquatch."

That last one cuts both ways and is rarely noticed. Forest area predicts bears, and it also predicts hunters and hikers, and the model cannot separate the two.

The finding is a strong argument that a large share of reports are misidentified bears. It is not a demonstration that no unknown animal exists, and the paper does not claim to be one.

The prior result

Foxon was not first. Jeffrey Lozier, Peter Aniello, and Michael Hickerson published "Predicting the distribution of Sasquatch in western North America: anything goes with ecological niche modelling" in the Journal of Biogeography, volume 36, issue 9, pages 1623 to 1627, in 2009. They built an ecological niche model from georeferenced sasquatch sightings and footprint reports in western North America, then, in their own words, "compare the distribution of Bigfoot with an ENM for black bear, Ursus americanus, and suggest that many cryptozoid sightings may be cases of mistaken identity." DOCUMENTED

Their actual argument was methodological, and the abstract says so plainly. They set out to "remind researchers of the need for careful evaluation of database records prior to use in modelling," and used Sasquatch to demonstrate "how convincing environmentally predicted distributions of a taxon's potential range can be generated from questionable site-occurrence data." Sasquatch was the reductio. The bear overlap was a side effect that turned out to be the most cited thing in the paper.

Foxon's reference list also cites a 2005 paper by E. A. Blight, "Relationships between the American Black Bear Population and the Bigfoot Phenomenon," hosted as a PDF on a personal site under the name Blight Investigations. That the citation is there is a fact about the reference list, nothing more. It is self-published, not peer reviewed, and I have not read it.

The animal that fits the description

The correlation only means something if a black bear can actually generate the reported experience.

The National Park Service gives the American black bear as "about 3 feet at the shoulder," "5-7 feet when standing upright," and "100 to 600 pounds depending on age, sex, and season." In Great Smoky Mountains National Park the agency puts adult males at "about six feet in length" and "around 250 pounds in summer," with both sexes nearly doubling their weight approaching fall. DOCUMENTED A large boar rearing up sits inside the height range witnesses report and above it in weight.

Bears rear to see and smell, which is exactly what happens when a person comes into view. Sustained bipedal walking is rarer, and better documented than most people assume. A black bear known as Pedals was first filmed walking upright in Oak Ridge, New Jersey, in 2014, walking on its hind legs because of injuries to its front paws. Officials initially warned the videos might be a hoax. They were not. On October 17, 2016, the New Jersey Department of Environmental Protection announced it believed Pedals had been killed during the state's sanctioned hunt. DOCUMENTED

Mange completes the picture. Incidence of sarcoptic mange in black bears "has increased substantially in Pennsylvania, USA" over the past thirty years and "has been increasingly reported from other states." A 2025 comparative study of 34 scabietic black bears alongside coyotes and red foxes found that "red foxes and black bears displayed more severe outward signs of mange infestation and higher mite burdens compared to coyotes." Its grading scale runs to hair loss over more than two thirds of the body, skin scored as "large areas of thickened, crusted skin," and body condition scored as emaciated. Of the scabietic black bears sampled, 91.18 percent showed acanthosis and 73.53 percent had bacteria present on histology. DOCUMENTED

Take a 300-pound bear, strip the hair from two thirds of it, thicken and darken the exposed skin, drop its weight, and stand it up. That is not a bear as anyone pictures a bear.

Which is the whole dispute over the Jacobs photographs. In September 2007 a deer hunter named Rick Jacobs, in Pennsylvania, captured images on a game camera. Some proponents read the figure as a juvenile sasquatch, and Bushnell and Field & Stream built a promotional contest around the photograph. ALLEGED The Pennsylvania Game Commission said the images showed a bear with mange. CONTESTED Vanessa Woods estimated the subject's arms at roughly 22 inches and its torso at 18.75 inches and concluded the proportions were more comparable to a chimpanzee. ALLEGED The photographs are the cleanest single illustration of why the mange explanation is not a dodge, and they remain unresolved.

The conditions that produce the error

The BFRO's own classification system, written by proponents for proponents, describes the failure mode better than any skeptic has. "The BFRO's report classification system rates the circumstantial potential for misinterpretation, not the credibility of the witness or how interesting the report is." Class A covers "clear sightings in circumstances where misinterpretation or misidentification of other animals can be ruled out with greater confidence." Class B covers "incidents where a possible sasquatch was observed at a great distance or in poor lighting conditions and incidents in any other circumstance that did not afford a clear view of the subject." DOCUMENTED

That is the honest version of the problem, and it comes from inside the field. Distance, bad light, brief duration, and nothing of known size in frame to scale against. Under those four conditions a rearing bear and a large bipedal hominid are not reliably distinguishable, and a witness can be entirely sincere and entirely accurate about the experience while being wrong about the animal.

Where the correlation is weakest

Florida is the paper's named awkward case, and Foxon raises it rather than burying it: "At first glance, there are locations such as Florida with very many sasquatch sightings but relatively few bears (at least compared to some other states/provinces). This appears to be inconsistent with the hypothesis." The answer given is that the comparison "is confounded by the human population and forest area in each state/province," because "Florida has relatively few bears and relatively little forest area, but also a relatively large number of people to make sightings." DOCUMENTED That is a defense of the model rather than an independent confirmation of it.

The other soft spots follow from the data:

• Texas and Wisconsin were dropped for missing bear estimates, and both generate reports. The analysis is silent about them, not supportive.
• Nine states were entered at zero bears by assumption rather than measurement.
• The positive intercept is the general form of the same problem. The model predicts sightings where it says there are no bears.
• Everything rests on a single year, 2006, with bear estimates taken from a 2007 agency survey and Canadian forest figures from a 2017 web article. There is no time dimension to test the association against.

Proponents point at exactly these gaps, and the pointing is legitimate as far as it goes. I found no published, peer-reviewed rebuttal to Foxon in the zoological literature, though that is a statement about what I could locate rather than proof that none exists. Noting that a model has unexplained residuals is not the same as producing a better model of the same data. Until someone does, the bear correlation stands as the strongest quantitative result the subject has, and what it says is that bears explain a great deal and do not explain everything.

---

Sources

• Floe Foxon, "Bigfoot: If it's there, could it be a bear?", Journal of Zoology 323(1), pp. 1-8, published online January 13, 2024, print May 2024. DOI 10.1111/jzo.13148. Bibliographic record, affiliation, and deposited abstract verified through the Crossref and OpenAlex APIs. The publisher's full text at zslpublications.onlinelibrary.wiley.com returned HTTP 403 and was not read. The deposited abstract has its numerals stripped, as quoted above.
• Floe Foxon, "If it's there, could it be a bear?", bioRxiv preprint. DOI 10.1101/2023.01.14.524058. Version 1 posted January 18, 2023, version 2 January 29, 2023, version 3 May 26, 2023. Version 3 is the source for the coefficient, the confidence interval, the 5,000-bears ratio, the model specification, the data sources, the exclusions and zero-bear assumptions, the RMSE, the Florida passage, and the author's caveats. Version 1, version 2, version 3.
• R. D. Spencer, R. A. Beausoleil, and D. A. Martorello, "How Agencies Respond to Human-black Bear Conflicts: A Survey of Wildlife Agencies in North America," Ursus 18(2), pp. 217-229, 2007. Cited by Foxon as the source of the bear population estimates. Record verified through Crossref and OpenAlex.
• J. D. Lozier, P. Aniello, and M. J. Hickerson, "Predicting the distribution of Sasquatch in western North America: anything goes with ecological niche modelling," Journal of Biogeography 36(9), pp. 1623-1627, 2009. DOI 10.1111/j.1365-2699.2009.02152.x. Abstract quoted from the Crossref deposit and OpenAlex. The full text was not read.
• Floe Foxon, "The Loch Ness Monster: If It's Real, Could It Be an Eel?", JMIRx Bio 1, e49063, July 21, 2023. DOI 10.2196/49063. The author's publication record is characterized only from works indexed under ORCID 0000-0002-4893-9178 in OpenAlex.
• National Park Service, "Black Bears." Source for 3 feet at the shoulder, 5 to 7 feet standing upright, and 100 to 600 pounds. nps.gov
• National Park Service, Great Smoky Mountains National Park, "Black Bears." Source for adult male length and summer weight. nps.gov
• H. S. Tiffin, J. D. Brown, K. Kelly, K. R. Van Why, M. Ternent, A. C. Camire, E. J. A. Schuler, R. T. Marconi, and E. T. Machtinger, "Infestation by Sarcoptes scabiei causes distinct differences in sarcoptic mange disease syndromes among sympatric carnivoran species," International Journal for Parasitology: Parasites and Wildlife 27, 101070, August 2025. DOI 10.1016/j.ijppaw.2025.101070. Full text read at PMC12861646. Source for the sample size, the severity scale, the 91.18 and 73.53 percent figures, the bear-versus-coyote comparison, and the Pennsylvania trend.
• K. D. Niedringhaus, J. D. Brown, M. Ternent, W. Childress, J. R. Gettings, and M. J. Yabsley, "The emergence and expansion of sarcoptic mange in American black bears (Ursus americanus) in the United States," Veterinary Parasitology: Regional Studies and Reports 17, 100303, 2019. DOI 10.1016/j.vprsr.2019.100303. Listed for further reading. The record was verified, but no abstract or full text was accessible, so nothing in this entry rests on it.
• Bigfoot Field Researchers Organization, "The BFRO's Report Classification System." Source for the Class A and Class B definitions and the sentence on what the system rates. bfro.net
• Bigfoot Field Researchers Organization Geographic Database, the sightings source used by Foxon. bfro.net
• "Pedals (bear)," Wikipedia. Source for the 2014 Oak Ridge, New Jersey sighting, the front-paw injuries, the initial hoax warnings, and the New Jersey Department of Environmental Protection statement of October 17, 2016. wikipedia.org)
• "Bigfoot," Wikipedia. Source for the Pennsylvania Game Commission's identification of the 2007 photographs as a bear with mange, and for the Vanessa Woods measurements and conclusion. wikipedia.org
• "Bigfoot in popular culture," Wikipedia. Source for September 2007, deer hunter Rick Jacobs of Pennsylvania, the game camera, and the Bushnell and Field & Stream contest. wikipedia.org