Close Encounters of the Data Kind
For long have people been fascinated with unknown and hidden subjects: UFOs, aliens, ghosts, and other paranormal subjects have captured the imagination of many, over many years, often in waves. In the absence of official or mainstream scientific credence, popular mythologies of suppression and conspiracy have been embraced by many. Rumors, hearsay, and urban myths can become viral, especially in the age of social media. And the motto may well be, as the poster on Fox Mulder’s wall in the popular series The X-files declares, “I want to believe.”
With such thoughts in mind, I have undertaken, as a beginning student of data science, an exploration of a dataset compiling slightly over a century of UFO sightings as collected by the National UFO Reporting Center (NUFORC). This dataset is readily available on a number of platforms, such as Kaggle: https://www.kaggle.com/NUFORC/ufo-sightings#complete.csv. I have restricted also to sightings within the United States, which by far constitute the bulk of the cases.
We will here look at some of the basic questions concerning how common, and distributed in both time and geography, have such sightings been? What may the data tell us? Let’s begin with time.
Distribution in Time
UFO sightings, it seems, rise gradually through the modern period in waves. An interesting question, of course, can be why? Are there broader social causes, for example, at play, underlying the waves? Data, of course, may suggest correlations, though causation is something we can perhaps only speculate on.
The data covers a period of time from 1910 at earliest, through May 8, 2014. Prior to 1940 there are a mere total of 14 reports. The beginning of wider modern interest is often cited as in two pivotal incidents of 1947: Kenneth Arnold’s sighting of “flying saucers” (less due to the objects shape than to their apparent aerodynamic behavior as he followed them in his plane near Mt. Rainier in Washington State), and the incident at Roswell, NM, in which an alien craft is said to have been crashed and been recovered (only for the accounts to be retracted the following day).
Thus the waves of sightings began, it appears, actually in 1945 (nearly double from 1944, such as the numbers were). 1945 , of course, also saw the beginnings of the nuclear age: the Trinity test at Sorroco, NM on July 16th, followed soon by the bombings of Hiroshima and Nagasaki (August 6th and 9th respectively). That seems likely more than merely coincidence. Indeed, it is often noticed in UFO literature.
Following a dip in 1948 (returning to 1946 levels), there was a steady increase in sightings in increasingly higher waves, through the Cold War, and into a even greater growth of interest during the 1960s — 70s, with no doubt it’s periods of social upheaval, change, and experimentation.
And then the rate of sightings settles, apparently, through the 1980s and early 1990s, to around 200 per year, only to climb steadily from 1994 through the new millennium, with numbers growing rapidly from mere hundreds to thousands. One can of course imagine the circumstances for this growth: anxieties concerning the upcoming millennium, the Y2K scare, and later the popular interests in and wary regarding the arrival of 2012, heralded by many to be the end of the world (at least as we have known it to be) and perhaps the beginning of a New Age. One in which many believed extraterrestrials may take an important part.
Distribution in Space (or Geography)
California shows to be, by far, the state with the most sightings, more than twice that of Washington, which is second. Does this distribution appear to change, with time periods? Apparently not. The majority of sightings has seemed to be there. What, if anything, would make California unique in that regard? Do aliens tend to prefer Californians?
As can happen with graphs, of course, that may be deceptive. A scatter plot of sightings across the U.S. map actually shows a far greater density of sightings in the Eastern US than the West as a whole. But, of course, eastern states tend on a whole to cover smaller geographical areas. So while sightings are denser along the eastern seacoast, the sightings are distributed among many more states, each receiving smaller shares. And aliens (if indeed UFOs are piloted by such) probably don’t care much about state boundaries nor state’s political dispositions.
But do UFOs express other geographical preferences of where they may appear? It is common in the UFO world that sightings may be more prevalent in the vicinity of military, and especially nuclear weapon facilities. However, closer examination of these maps (which may be done at https://karencfisher.github.io/ufosightings/sightings.html) does not show any such clear correlation. (Maps show air force and nuclear facilities along with UFO sightings by decade, as shown.)
More precisely, given the hypothesis that UFO sightings tend to cluster around such sites, it is prudent to evaluate the null hypothesis: that there is no such tendency (and the distribution of sightings is more or less random in that regard). The map then fails to reject the null hypothesis. It does not support the popular view.
If anything, sightings as well as military installations appear (as one may expect) to cluster around areas of greater population. Simply, perhaps, there are more people to see them. Sightings are less common in more remote (and darker) areas of the country, though one may think them more ideal locations to see things in the sky.
UFOs continue to be a mystery. While we may explore their distributions in time or geography, this provides little in way of real explanation. Data can show interesting correlations (or lack thereof), but it cannot necessarily answer the deeper questions: what are they? And if they are indeed crafts piloted by intelligent beings, who are they? Where have they come from? And why are they here? Lights in the sky and various objects, and their meaning, may well continue to elude our final conclusion. Perhaps, in the end, that is why they are just that: “unidentified.”
