Toward a Resolution of the Bigfoot Phenomenon

J. Glickman

Part 1 | Part 2 | Part 3

The Bigfoot phenomenon may be the result of a combination of sociological origin, physical manifestation through willful manufacture, and the by-product of cataloged and uncataloged animals. Observational data related to the Bigfoot phenomenon is presented and analyzed to identify its origin. Human and animal archetypes are used to demonstrate the inclusion or exclusion of these archetypes within the observations. An argument of continuity, the expectation that there may be a continuous record of the existence of an organism, is employed to include or exclude the possibility that the observations originate from an uncataloged animal. The plausibility of an uncataloged animal is examined through ecological analogy.

Monsters, and more specifically myths of Big Hairy Monsters (BHM), are a world-wide anthropological phenomenon. In North America, one such myth, centered principally in the Pacific Northwest, is known as Bigfoot. Many contemporary stories relate individual and group experiences with the Bigfoot phenomenon. Robert Pyle aptly observed, "...the phenomenon of Bigfoot exists." [Pyle 1995]. This single, lucid observation, which differentiates the existence of a Bigfoot from the existence of the phenomenon, forms the basis of this paper. Since we know that the phenomenon exists, what is its source?

The Bigfoot phenomenon may be of sociological origin, it may be physically manifested through elaborate manufacture, or it may be the by-product of an animal, cataloged or uncataloged. Its magnitude and distribution however, are, in the author's opinion, unusual and therefore important to understand. If the phenomenon is of social origin, how did it become so widespread, how does it sustain itself, and why has it been so long-lived? If the phenomenon is of elaborate manufacture, how was geographically and temporally widespread manufacture accomplished and concealed? If the phenomenon is the by-product of a cataloged animal how did human perceptual mechanisms fail? Finally, if the phenomenon is the by-product of an uncataloged animal why is there a dearth of evidence and why are we reluctant to investigate the phenomenon? Whichever of these are eventually proven to be the origin of the Bigfoot phenomenon, humanity will be the beneficiary of its investigation, by gaining new insights into the human animal.

This paper reviews observations of the phenomenon and proposes a methodology for its continued examination. A null hypothesis for this paper is formulated and presented. The observations are cataloged and their sources critiqued, which is followed by the analysis of the observations. From this analysis, new hypotheses are postulated. The conclusion presents the results of this study and provides recommendations for future studies.

Methodology

The methodology that will be used to determine the source of the Bigfoot phenomenon is:

1. Assert that there is a Bigfoot phenomenon.
2. Create a set of hypotheses enumerating the possible sources of the Bigfoot phenomenon. These include, but are not limited to, the social hypothesis, the manufacture hypothesis, the misidentification hypothesis and the uncataloged animal hypothesis.
3. Collect observations. A set of observations have been collected to facilitate the initial analysis of the phenomenon.
4. Analyze the observations to test the hypothesis.
5. Formulate new hypotheses as appropriate.

One argument that is employed to contradict the null hypothesis is the continuity argument. Continuity is an expression of evolution. Relative to the human experience, evolution is a slow process. Species gradually evolve from one to another, and eventually become extinct. There are exceptions, for example cataclysms that create adaptation challenges. Those species able to adapt survive, and those unable to adapt perish.

Some species leave a complex record of their existence, which begins with fossil evidence. Since the advent of man, extant species leave an anecdotal record through man's collective memory.

There are exceptions to both of these. For example, the chimpanzee and gorilla have no fossil record [Jones 1992] and since the beginning of this century seven new species of land mammal have been discovered [NYT 1994]. Therefore, gaps in the record of a species do not constitute unequivocal proof of non-existence.

Nonetheless, these are the exceptions and not the rule. The likelihood of a large North American animal having remained uncataloged and having no fossil record is slim. This is the essence of the continuity test: To make a plausible argument for an uncataloged animal, its continuity may be demonstrated. To demonstrate the possible implausibility of an uncataloged animal, one may illustrate discontinuities in the record.

Hypothesis

The null hypothesis has been carefully chosen because the existence of Bigfoot can not be proven due to the absence of a type specimen, therefore a null hypothesis that requires proof of the existence of Bigfoot is fatally flawed.

Archetypes do exist for proving that observations are manufactured by humans. The null hypothesis must be one that can be successfully contradicted, which may only be done with the human archetype. Thus the null hypothesis must be "The Bigfoot phenomenon originates from an uncataloged animal" because this can be contradicted by proving, for example, that an image captured on movie film is that of a human in a costume. The null hypothesis is:

• The Bigfoot phenomenon originates from an uncataloged animal.

The observations will be used to refute the null hypothesis. If the null hypothesis is successfully contradicted, then by implication:

• The Bigfoot phenomenon is of social origin,
• The Bigfoot phenomenon is the result of misidentification,

or

• The Bigfoot phenomenon is manufactured.

The Bigfoot phenomenon may originate from the super-position of observations traceable to multiple hypotheses.

Analysis

Observations of the Bigfoot phenomenon are presented, some of which are circumstantial, and among which there may be coincidence. Since there are no theories yet to model these observations, a danger resides in ascribing meaning to outcomes that are unexpected, for which an as yet absent theoretical model would predict.

Purported observation of the Bigfoot phenomenon include sightings, footprints, sounds, smells, thrown objects, hair, feces and photographs. Several individuals in the Bigfoot research community have attempted to support the phenomenon by trying to correlate the contemporary phenomenon with the European settler's historical record, Native American cultural memory, and the fossil record and are categorized as historical anecdotes.

These will be reviewed in the following sections. Things sensed (seen, heard, smelled, etc.) and subsequently reported without physical record, such as sightings, footprints, sounds, smells and thrown objects are categorized as contemporary anecdotes. In some cases, the individual or group reporting the observation presents a physical record of the event in the form of samples, footprint casts, or photographs. These materials cannot be proven to be authentic, nor do they prove the existence of an uncataloged animal because of the absence of a type specimen. These are categorized as contemporary physical observations.

Categories of observations of the Bigfoot phenomenon are shown in Table 1.

Observations from these classifications are presented in reverse temporal order — from the most recent observations to the oldest observations. Ecological plausibility and BHM as an anthropological phenomenon will be analyzed.

Contemporary Anecdotes

There are many stories, centered principally in the Pacific Northwest, that relate contemporary individual and group experiences to the Bigfoot phenomenon. Many individuals and groups comprise the Bigfoot research community, including Professor Grover Krantz, John Napier, John Green, Ray Crowe, Rene Dahinden, Bob Titmus, Ivan Sanderson and Peter Byrne to name a few. All have made some effort to collect anecdotal observations. In two cases the author is aware of, efforts have been made to formalize the collection of anecdotal observations. One such effort was led by John Green and the other by Peter Byrne.

Table 1: Categories of Observations

Time (inclusive)	Category	Examples
Contemporary (postdate 1958)	Anecdotes	sightings, sounds, footprints, smells, thrown objects
	Physical Record	footprint casts, hair samples, photography (film, video, still)
Historical (predate 1958)	Anecdotes	settler historical record, Native American cultural memory
	Physical Record	fossils

The role of the contemporary anecdotal observations is to support or refute the main hypothesis. Each qualified anecdote is quantified by representing the anecdote as a geo-time coded event, i.e. date, time, latitude, longitude and altitude. This dataset is then analyzed by SPSS 1, a computer-based statistical analysis software package.

Green's Sighting Data

John Green has been involved in the Bigfoot community for approximately thirty years and as of the 1981 printing of his book [Green 1981] claimed to have over 1,500 confirmed sightings. Mr. Green's current data was not formally made available to this study, so the methods employed by him and the manner by which his data are organized cannot be assessed.

As an alternative to using his current data, Green's national sighting data as of November 1977 is summarized in Table 2 [Green 1981]. Green's data is analyzed first because it covers the largest geographic area, and to the best of the author's knowledge, is the only collection of continental data.

Methodology

Green's data will be tested against a simplistic model of expected sighting rates for animals. The probability of receiving a report for a cataloged animal is modeled as:

Pr = Ps. Pa . Ph . Pe (Eq.1)

where,

Pr is the probability function of receiving a report,

Ps is the probability function that an observation results in a report submission,

Pa is the probability function of an animal being at a specific place and time to be observed,

Ph is the probability function of a human being in a specific place and time to make the observation, and

Pe is the probability function of an observer expecting to observe the phenomenon.

The author assumes that the probability that an observation results in a report submission is geographically uniform, so this reduces to a constant. The probability of an animal being in a specific place and time to be observed is directly proportional to the animal's population density. A uniform distribution is assumed. In the event the animal's population density is non-uniform, this becomes superimposed upon the result. The probability that a human in a specific place and time makes an observation is directly proportional to human population density. This is modeled on a per-state basis as the number of square miles divided by the population [Gousha 1995].

Analysis

Table 2 is organized on a per-state basis and is ordered in descending normalized frequency. The "Freq." column contains Green's reported observation frequencies [Green 1981]. "Dist." is an ordinal distance reference as measured from the geographic center of the state to the geographic center of Washington. "Sq. Mi." is the number of square miles in the state. "Population" is the 1980 population census figure for the state. "Pop./Sq. Mi." is derived as "Population" divided by "Sq.Mi." "Norm. Freq." is the normalized frequency and is derived as "Freq." divided by "Pop./Sq.Mi."

Therefore:

Eq. 2

"Group" is the assigned cluster group resulting from cluster analysis (presented below). Canadian data is not included, due to incomplete data.

Table 2: Green Sighting Data

Case	State	Dist.	Freq.	Sq.Mi.	Human Population	Pop/Sq. Mi.	Norm Freq.	Cluster Group
1	Alaska	76	20	550,000	400,481	0.73	27.47	A
2	Montana	22	74	147,138	786,690	5.35	13.84	A
3	Oregon	10	176	96,981	2,632,663	27.15	6.48	A
4	Washington	0	281	68,192	4,130,163	60.57	4.64	A
5	N.California(Est.)	25	294	79,347	5,917,141	74.57	3.94	A
6	S.California(Est.)	35	49	79,347	17,751,422	223.72	0.22	B
7	Idaho	15	32	83,557	943,935	11.30	2.83	A
8	Wyoming	31	4	94,914	470,816	4.96	0.81	B
9	South Dakota	44	7	77,047	690,178	8.96	0.78	B
10	Nevada	26	5	110,540	799,184	7.23	0.69	B
11	New Mexico	52	7	121,510	1,299,968	10.70	0.65	B
12	Florida	107	104	58,560	9,739,992	166.33	0.63	B
13	Texas	70	30	267,339	14,228,283	53.22	0.56	B
14	Arkansas	74	19	53,104	2,285,513	43.04	0.44	B
15	Iowa	60	15	56,290	2,913,387	51.76	0.29	B
16	North Dakota	40	2	70,665	652,695	9.24	0.22	B
17	Arizona	45	5	113,575	2,717,866	23.93	0.21	B
18	Kansas	55	6	82,264	2,363,208	28.73	0.21	B
19	Oklahoma	64	9	69,919	3,025,261	43.27	0.21	B
20	Mississippi	83	8	47,716	2,520,638	52.83	0.15	B
21	Nebraska	48	3	77,227	1,570,006	20.33	0.15	B
22	Colorado	42	4	104,247	2,888,834	27.71	0.14	B
23	Missouri	67	10	69,686	4,917,444	70.57	0.14	B
24	Maine	105	4	33,040	1,124,660	34.04	0.12	B
25	Utah	32	2	84,916	1,461,037	17.21	0.12	B
26	Illinois	71	23	56,400	11,418,461	202.45	0.11	B
27	Michigan	75	18	58,216	9,258,344	159.03	0.11	B
28	Georgia	95	10	58,876	5,464,265	92.81	0.11	B
29	Minnesota	53	5	84,068	4,077,148	48.50	0.10	B
30	Indiana	77	15	36,291	5,490,179	151.28	0.10	B
31	Wisconsin	64	8	56,154	4,705,355	83.79	0.10	B
32	Pennsylvania	93	24	45,333	11,866,728	261.77	0.09	B
33	Tennessee	84	9	42,244	4,590,750	108.67	0.08	B
34	Kentucky	84	7	40,395	3,661,433	90.64	0.08	B
35	West Virginia	90	6	24,181	1,949,644	80.63	0.07	B
36	Ohio	84	19	41,222	10,797,419	261.93	0.07	B
37	Alabama	88	5	51,069	3,890,061	76.17	0.07	B
38	South Carolina	98	6	31,055	3,119,208	100.44	0.06	B
39	Louisiana	82	5	48,523	4,203,972	86.64	0.06	B
40	New Hampshire	102	5	9,304	920,610	98.95	0.05	B
41	North Carolina	99	5	52,712	5,874,429	111.44	0.04	B
42	New Jersey	101	36	7,836	7,364,158	939.79	0.04	B
43	Vermont	99	2	9,609	511,456	53.23	0.04	B
44	New York	95	11	49,576	17,557,288	354.15	0.03	B
45	Virginia	96	4	40,815	5,346,279	130.99	0.03	B
46	Maryland	98	12	10,577	4,216,446	398.64	0.03	B
47	Delaware	100	1	2,057	592,225	287.91	0.00	B
48	Connecticut	103	2	5,009	3,107,576	620.40	0.00	B
49	Massachusetts	102	1	8,257	5,737,037	694.81	0.00	B
50	Rhode Island	105	0	1,214	947,154	780.19	0.00	B
	Mean	69.32	28.18	71,362	4,497,982	147.05	1.35
	Median	75.50	7.50	56,345	3,113,392	75.37	0.12
	Std. Dev.	4.18	61.09	11,613	601,667	206.58	4.39
	Std. Err.	29.53	8.64	82,114	4,254,426	29.22	0.62

Table 3 presents bivariate correlation coefficients for Table 2 data between frequency and population, and frequency and population density are computed as a baseline prior to data clustering and is called the baseline correlation.

The frequency is not well correlated to either the population or the population density across the entire dataset. Hierarchical cluster analysis was subsequently performed on the normalized frequency. Clustering was done by case, and a range of solutions from two to five clusters was computed. The result of cluster analysis is presented in Table 4.

The lack of additional cases in cluster group Green5 from cluster group Green4 suggests two things: that the cases in clusters 1 through 4 of cluster group Green5 are differentiated from the rest of the dataset, and that two clusters is the appropriate cluster size since the hierarchical analysis simply rearranged the set of cases in Green4 and Green5.

Cases 1, 2, 3, 4, 5 and 7 are called Group A which consists of Alaska, Montana, Oregon, Washington, Northern California and Idaho. The remainder of the cases are called Group B. The "Cluster Group" column in Table 2 shows the result of clustering.

The same correlations as those computed for the baseline were computed for Group A and B and are summarized in Table 5.

Discussion

The relationship in the clustered data is the correlation between population density and frequency: the Group A correlation of +0.9661 is high relative to the Group B correlation of +0.1244.

A second relationship in the clustered data is the correlation between population and frequency. When Group A is separated from the dataset, its correlation to population rises from +0.1192 to +0.5664.

Group A is differentiated from Group B by its high correlation to population density. This is consistent with the model of receiving a report of a cataloged animal (Eq. 1).

Table 3: Correlation of Green's Data to Population Statistics

	Frequency vs. Population	Frequency vs. Population Density
Baseline Correlation	+0.1192	+0.2673
Significance	0.410	0.061
Cases	50	50

Table 4: Cluster Analysis of Green Sighting Data

Cluster Group Name	Number of Clusters	Cluster1	Cluster2	Cluster3	Cluster4	Cluster5
Green2	2	1	all others	N/A	N/A	N/A
Green3	3	1	2	all others	N/A	N/A
Green4	4	1	2	3,4,5,7	all others	N/A
Green5	5	1	2	3	4,5,7	all others

Let's assume that manufactured reports will be uniformly distributed across the population. If the rate of manufactured reports is constant, then the frequency of reports should correlate to population. To some degree, this is seen in Group B. There may be other unidentified influencing factors such as mean media exposure to Bigfoot, which may influence the density of manufacturing. The author speculates that Group A and Group B represent different phenomenon. Group B may represent manufactured reports because of the correlation to population, whereas Group A may represent a different phenomenon because of its correlation to population density. The author hypothesizes that if Green's data is the superposition of multiple phenomena that this is the expected result.

Sapunov reports a theory of testimonies developed and employed in the USSR in the mid 1980s capable of testing populations of eyewitness reports for authenticity:

The mathematical theory of testimonies was developed mainly on data from traffic incidents (Rossinsky 1984). According to the theory, the distribution of quantitative characters of observed items within a group of witnesses must be normal or Guassian. Subjective biases on the part of witnesses tend to displace the mode of distribution. The qualifications or educational backgrounds of witnesses influence the variance of distribution: the higher the qualifications or education, the less is the variance of distribution. [Sapunov 1988]

Sapunov continues:

According to the theory of testimonies, the extremes of the quantitative traits reported by a group of independent witnesses should be distributed in the tail or tails of a normal or Guassian distribution if the data are authentic (Rossinsky 1984). False reports would be distributed with many peaks, and without tails. The existence of one or two modes suggests a single direction of hoaxing — which is unlikely — or the objective reality of the reports. [Sapunov 1988]

Table 5: Post-Clustering Correlations of Green's Sighting Data to Population Statistics

	Frequency vs. Population	Frequency vs. Population Density
Baseline Correlation	+0.1192	+0.2673
Baseline Significance	0.410	0.061
Baseline Cases	50	50
Group A Correlation	+0.9626	+0.9661
Group A Significance	0.002	0.002
Group A Cases	6	6
Group B Correlation	+0.5664	+0.1244
Group B Significance	0.000	0.421
Group B Cases	44	44

TBRP GIS† 1 Data

(†Geographic Information System)

Peter Byrne has been in the Bigfoot community on a full-time basis for seventeen of the last thirty-five years most recently serving as the Director of The Bigfoot Research Project (TBRP). Whereas Green's data is national with coarse geographic information, TBRP's data is regional with precise geographic information. Based on Byrne's intuition, TBRP focused solely on the Pacific Northwest. In so doing, TBRP was investigating the Group A phenomenon. While this permitted TBRP to study that region in more depth, it is also unfortunate that there is no national data with which to compare to their regional results.

TBRP collected ancedotal observations by soliciting reports via a toll-free telephone number through newspaper advertisements. During the month of May 1996, TBRP received two-thousand-two-hundred-sixty telephone calls, most of which were categorized as nuisance calls from children. Since 1992, TBRP has collected approximately one-thousand regional anecdotal observations, three-hundred and seventy-four of which have been deemed credible by TBRP, though the methodology by which this determination was made is subjective.

Methodology

When TBRP received a non-nuisance telephone call it identified what type of anecdotal observation was being reported and filled out a survey form specific to this type. There was one survey used for sightings (15 pages), one for footprints (11 pages), and a combined survey for sounds, smells and thrown objects (12 pages). The surveys were authored by TBRP and were not examined by a survey professional for bias or leading questions.

A subset of these anecdotes were geocoded and entered into a computer database (This dataset is referred to as TBRP1 and is shown in Figure 1). TBRP staff employed an informal model of what constituted a credible report which they developed intuitively. The credibility of an anecdote was assessed by the subjective application of this informal model. If the anecdote matched their informal model closely enough, it was deemed credible. This method filtered the anecdotes according to TBRP staff expectations and skewed the computer database toward the staff's informal model. Anecdotes were further categorized with a credibility rating of "A" through "C" based upon the personal judgment of TBRP staff.

A limited amount of information was entered into a computer database, which included a case number, date of occurrence, location description, latitude, longitude, altitude, one or more anecdote classifications consisting of sighting, footprint, sound, smell, or thrown object, and the credibility rating. As of June 17th, 1996, three-hundred and seventy-four anecdotes were cataloged by TBRP as credible, all of them in the Pacific Northwest. One-hundred and sixty-seven of these have complete information including date, altitude, and geocoding. These one-hundred and sixty-seven reports, which are referred to as Group I, are the dataset for the analysis below.

Definitions of Anecdotal Classifications

There are five anecdotal classifications recognized by TBRP. These are sightings, footprints, sounds, smells and thrown objects. Anecdotes are cataloged as a:

• Sighting when the observer reports seeing a Bigfoot. If a photograph is presented the case is still given a sighting classification.
• Footprint when the observer reports seeing a large footprint. If a plaster cast of one or more of the footprints is presented the case is still given a footprint classification.
• Sound when the observer reports loud whistling, screaming, or roaring.
• Smell when the observer reports an overpowering, noxious smell.
• Thrown Object when the observer reports objects thrown.

Whenever more than one classification is applicable, multiple classifications are associated with the case.

Analysis

TBRP's geocoded data was analyzed for patterns. Correlation coefficients were computed for all pairs of latitude, longitude, altitude, month and year in the dataset that had complete information. No significant correlations were found.

A new dataset was created, containing twelve cases, one for each month (This dataset is referred to as TBRP2). Frequency data by month, mean monthly latitude, mean monthly longitude and mean monthly altitude were aggregated from dataset TBRP1 and entered into dataset TBRP2. Mean monthly temperature and mean monthly precipitation for Portland, Oregon were manually added to dataset TBRP2. Correlation coefficients were computed for all pairs in dataset TBRP2. The only significant correlations found were between mean latitude, mean longitude and mean altitude, suggesting that there is a geographic pattern to the location of the reports filed with TBRP. This geographic pattern could be correlated with where the population lives, where people misidentify animals, where people are seeing an uncataloged animal, etc.

Figure 2 shows a high density of reports in and near Hood River County, Oregon. While the hot spot toward the center appears to be reporting the bias, the diagonal band from the upper right to the lower left is of interest. This area corresponds to the maximum altitude portion of the Cascade range to the south and west of Cascade Locks, Oregon, and to the north and east of Stevenson, Washington and Carson, Washington. These areas are very rugged and inaccessible. It is interesting to note that this high density area of reports originates from a low-population density area.

Part 1 | Part 2 | Part 3