Clustering Illusions, Baader-Meinhof, and Death in the Domincan Republic

[Mick West]

A somewhat unexpected property of randomness is that it forms patterns. Many years ago back when I was writing for Game Developer Magazine, I investigated how to automate the creation of virtual landscapes covered with trees. If the positions of the trees were truly random then inevitably there would be a few clusters of trees that were very close together, and empty spaces between the trees. Here's a perfectly random example i generated from random numbers. Notice there are clusters.

This is known as the Clustering Illusion, defined on Wikipedia as:

the tendency to erroneously consider the inevitable "streaks" or "clusters" arising in small samples from random distributions to be non-random.

Something that arises from the clustering illusions is the Texas Sharpshooter Fallacy. This is sometimes deliberate practice of drawing a ring around a random cluster, and then claiming some significance to the fact that there's a lot of points in that cluster. The name comes from a joke about a Texas Cowboy who shoots at the side of a barn. His aim is poor, so there a random scattering.

He goes over to the barn and picks the biggest clusters of shots, and then paints targets centered on those clusters. So now it looks like he's been aiming for that region, and hence there's a reason for this cluster. But all he did pick one random cluster that looked good for him.

Recently there have been new stories about a cluster of deaths of US tourists in the Dominican Republic. Nine Americans have died there over the last 12 months. The causes of these deaths are mostly listed as being heart attacks. However, there have been questions raised about possible poisoning of some sort.

Could this be a clustering illusion? Heart attacks are somewhat randomly scattered through the population, and they also correlate to demographic factors, so there's going to be both real and illusory clusters. Could it be that the Dominican Republic is simple the place that had a random cluster of deaths this year?

Perhaps there's something else at play here - the Baader-Meinhof phenomenon, also called the Frequency illusion. This is a trick your mind plays on you where if you've just recently heard of something, or it's otherwise in your thoughts, then you start to notice it occurring more frequently. The classic example is buying a new car (or just reading about a type of car, like a Tesla), and then you start to see them more often. They are not actually any more common, but you just notice them more.

So perhaps the media, being primed with the idea that there's a suspicious series of deaths in the Dominican Republic, is suddenly noticing things that would have escaped their notice entirely last year. People have died of heart attacks, and even of alcohol poisoning, in the Dominican Republic before, and they will again. Is it actually any more common this year?

Now, we know that 2.7 million US citizens visit the Dominican Republic every year. We know that people die somewhat randomly from heart attacks. Each year around 325,000 adults die of sudden Cardiac Death, which is about 1/1000th of the US population of 327,000,000. Simply extrapolating, that means each year 1/1000th of the 2.7 million people who visit the Dominican Republic will die of sudden heart attacks. That's 2,700 people. If the average trip length is seven days then that means, all things being equal, we'd expect about 52 US tourists per year to unexpectedly drop dead of a heart attack in the Dominican Republic!

It's probably not that high, as people traveling abroad are usually healthier on average than people. But still, a significant number of individually unexpected deaths are to be expected over a year.

 

[sharpnfuzzy]

I read that what made some of the deaths suspicious is that liquid was found in the lungs and that at least one pair was a couple who both died of cardiac arrest. Although it's not impossible I guess for two people to both die of cardiac arrest at the same time, it's probably very improbable.

I know, citations needed, I'll try to find the article again.

Correction:
It was stated as respiratory failure with liquid in the lungs.


https://www-m.cnn.com/2019/06/02/us...ic-death/index.html?r=https://duckduckgo.com/

 

[Mick West]

The statistical likelihood of a number of natural causes deaths does not automatically explain every death. But the number of deaths by natural causes is not itself suspicious.

This type of thing has happened before:
https://www.standard.co.uk/news/wor...-couple-died-for-four-weeks-but-a3919646.html

A British couple found dead in a hotel room on a holiday resort in Egypt both died from heart attacks, Egyptian authorities said today.

John and Susan Cooper, from Burnley, Lancashire, were staying at the Steigenberger Aqua Magic Hotel in the Red Sea resort of Hurghada when they became unwell on Tuesday.

Mr Cooper, 69, died in his room. Mrs Ormerod said her 63-year-old mother, a Thomas Cook employee, then died in an ambulance en route to hospital.

Mr Cooper died of "sudden failure in the heart muscle and respiratory failure," according to a Hurghada mortuary official, the BBC reported.

It came as Thomas Cook [a UK vacation company] banned holidaymakers from staying at the resort for at least the next month.

There were all sorts of speculation about carbon monoxide poisoning or being poisoned by the insecticide used. But eventually, it was revealed to be heart attacks triggered by food poisoning.
https://web.archive.org/web/2018091...25a700-b69c-11e8-ae4f-2c1439c96d79_story.html

CAIRO — Tests showed that e.coli bacteria were behind the death of two British tourists in a hotel in Egypt's Red Sea resort of Hurghada, the country's chief prosecutor said on Wednesday.

The statement by Prosecutor Nabil Sadek came a week after travel company Thomas Cook said that there was a "high level of e.coli and staphylococcus bacteria" at the Steigenberger Aqua Magic Hotel where John and Susan Cooper died Aug. 21 after falling ill in their room in the five-star hotel.

Forensic tests showed that John Cooper, 69, suffered acute intestinal dysentery caused by e.coli, and Susan Cooper, 64, suffered Hemolytic-uremic syndrome (HUS), likely because of e.coli, Sadek said.

He said that tests also showed no links between the couples' death and the spraying of their neighboring room with lambda-cyhalothrin 5 per cent. The insecticide is safe to use, according to the statement.

The couple's bodies showed "no criminal violence" and other tests showed no toxic or harmful gas emissions or leaks in their room and tests on air and water at the hotel found nothing unusual, the statement said.

 

[Graham2001]

A similar phenomenon happened in the US in 2001, there was an apparent string of shark attacks, which got a lot of media attention, but ironically that was in a year with lower attacks than average.

Wikipedia has what seems to be a well put together article on the event:

The Summer of the Shark refers to the coverage of shark attacks by American news media in the summer of 2001. The sensationalist coverage of shark attacks began in early July following the Fourth of July weekend shark attack on 8-year-old Jessie Arbogast, and continued almost unabated—despite no evidence for an actual increase in attacks—until the September 11 terrorist attacks shifted the media's attention away from beaches. The Summer of the Shark has since been remembered as an example of tabloid television perpetuating a story with no real merit beyond its ability to draw ratings

https://en.wikipedia.org/wiki/Summer_of_the_Shark

 

[Agent K]

Reminiscent of the "sonic attacks" or "health attacks" at the U.S. embassy in Cuba, but here we're talking about deaths and not just vague symptoms.

Police are looking into whether "bootleg booze" containing toxic chemicals could have contributed to the string of deaths that have occurred in the Dominican Republic between June 2018 and now, the New York Post reports...
Of the seven people who died, three people reportedly drank from the hotel room's minibar beforehand. Four of the deaths occurred at the same luxury resort complex, called Bahia Príncipe resorts, which is currently being investigated by authorities...
In an interview with the New York Times, Tom Inglesby, director of the Johns Hopkins Center for Health Security, said that although it's too soon to say for sure, the symptoms are "consistent with poisoning." So far, we know people have died from heart attacks, as well as internal bleeding and fluid in the lungs and brain. "It's rare for travelers to die of unknown causes like this, and to have a high number of them in a relatively short period of time is alarming, shocking, sad," he told the Times...
On Friday, Lawrence Kobilinsky, a forensic science professor at John Jay College of Criminal Justice in Manhattan, told the New York Post that the symptoms resemble methanol or pesticide poisoning.
https://www.refinery29.com/en-us/20...-poisoning-symptoms-dominican-republic-deaths

At the 2016 Olympics in Rio, a number of athletes had stomach problems.

A number of competitors at the Rio Olympics have reported stomach problems. Team GB officials have denied that athletes have fallen victim to food poisoning at the Olympic athletes' village in Rio, despite a number complaining of upset stomachs. A gastro-intestinal expert suggests that the culprit may, in some cases, be 'leaky gut' syndrome...
Professor Playford commented: "It comes as little surprise that some athletes at the Olympics are experiencing stomach complaints. The conditions are ideal for 'leaky gut' -- hot temperatures and high levels of exercise are a heady combination for the condition. The good news is that our research has shown that there is a natural and readily-available solution to the problem."
https://www.sciencedaily.com/releases/2016/08/160818085806.htm

 

[Tedsson]

The clustering illusion can be a nightmare for epidemiologists and public health workers trying to tease some cause/effect out of a dataset.

A good example of this is the cluster of childhood cancers around the Windscale/Sellafield nuclear power station in N.W England.

Big cluster of childhood cancers in one small area a few miles away. Sellafield was the obvious culprit and was blamed immediately. Obvious conclusion really. Or was it?

Whilst immediately "obvious" there was no repetition of the cluster over time nor was there any similar effect in nearby areas. The level of radiation from the power station was dwarfed by the background radiation (the bedrock in the area is plutonic and emits high levels of radon).

Further information here: https://scienceblog.cancerresearchu...-light-on-cancer-clusters-near-nuclear-sites/

So all very inconclusive. Sometimes clustering appears, sometimes it doesn't. Possibly the influence of increased monitoring/reporting? Difficult to tell. It is conceptually difficult to overcome the post hoc ergo propter hoc fallacy but it looks as if it was a random cluster (albeit one with a clearly identifiable villain. Possibly).

As an aside: as a big fan of get rich quick schemes I noticed there is usually quite a lot of clustering in lottery numbers. It probably won't help you win the jackpot but it might improve your "luck" if you generate a random number and then two or three more within +/-5 for two of your six numbers. Alternatively you can calculate the likelihood of winning and regard a lottery as a regressive tax on the gullible.

 

[Mick West]

The clustering illusions, or rather the more general human tendency to find patterns, trends, or correlations in random data, is why you see "Winning Number" displays like this next to roulette tables. It looks like the wins are clustering on black. But it's just random, a small sample, and someone taking a photo when the sign just happened to look unusual.

Source: https://wizardofodds.com/games/roulette/hot-numbers/

 

[deirdre]

Not really the same thing as deaths or cancer, since ASD is a rather subjective diagnosis with tots, but there was a big cluster scare of autism in Brick Township, NJ back in the 90s too. Turned out to be aggressive monitoring and reporting and just not having accurate data of rates from other locales.

As discussed below, intense case finding activities may have contributed to the high rate in Brick Township. Another important point to consider when interpreting the rate found in Brick Township is the lack of U.S. data on the prevalence of autism, although there is no reason to believe that the rate in U.S. populations should differ appreciably from other population groups

https://www.cdc.gov/ncbddd/developmentaldisabilities/documents/brick-report.pdf


I cant find data on Disneyland (guess they are keeping it hush hush) but hard to imagine there are less than 9 deaths a year, since people aren't used to the heat and on vacation you exert yourself more than you realize.

 

[Tedsson]

As an erstwhile medical geographer it would be remiss of me not to highlight the antithesis to random clustering by mentioning the famous cholera map produced by John Snow in Soho, London in 1854. Effectively the birth place of medical geography.

There was a cholera epidemic in 1854, one of many in the first part of 19thC London. It caused over 600 deaths (in what was, and still is, a relatively small area). John Snow was a local doctor and plotted the incidences on a streetmap. He got this:

The cases of cholera (in red) clustered around the water pump in Broad Street (centre, in blue). Snow realised the water was contaminated and had the council remove the handle from the pump. This stopped the cases of cholera so it looked pretty much as if he had hit the nail on the head. He had. The well feeding the pump (the source of all local water) was a few feet away from a contaminated cesspit. There was no clustering around the other local pumps (in blue).

Almost immediately this did away with the miasma theory of disease and replaced it with germ theory. It also prompted a big change in the attitude of the authorities and the last cholera outbreak in London was in 1866.

Further detail on the Broad Street outbreak here: https://en.wikipedia.org/wiki/1854_Broad_Street_cholera_outbreak

My takeaway from this which has served me well for 45 years is - there was a brewery in Broad Street. None of the brewers got cholera because they only drank beer (made from boiled water). Beer saves your life - QED.

So sometimes it pays to look very closely at clustering.

 

[Mick West]

So sometimes it pays to look very closely at clustering.

Absolutely! That's a pretty clear-cut case, but even when it's less clear it's nor something to automatically ignore - especially if there's nothing else to go by.

 

[Agent K]

Beer saves your life - QED.

Unless it's toxic, as may turn out to be the case in the Dominican Republic. Well, maybe not beer but booze.

 

[Mick West]

https://www.cnn.com/2019/06/23/world/hard-rock-dominican-republic-tourists/index.html

The Hard Rock Hotel and Casino is removing liquor dispensers from guest room minibars at its resort in Punta Cana, the Dominican Republic, the general manager of the resort told CNN.

Hard Rock Hotel and Casino decided last week to remove the liquor dispensers and hopes to "provide more tranquility for guests," GM Erica Lopez said. The decision to remove the dispensers was made independently and not as a result of the two deaths that happened at the Hard Rock Hotel and Casino at Punta Cana, Lopez added.
The decision followsa series of American tourist deaths in the Dominican Republic, some of which may have involved liquor.

Media scare spoils vacation for alchoholics! I never knew there was such a thing as a hotel room "liquor dispenser"

Possible a good idea though, regardless of if it's being refilled with hooch.

 

[deirdre]

• June 23, 2019

SANTO DOMINGO, Dominican Republic — The tourism minister grimaced as he read aloud the causes of death in autopsy reports of the growing number of Americans who have died recently while vacationing in the Dominican Republic: Heart attack. Septic shock. Pneumonia.

…..

Hitting back at the alarms raised by relatives of the dead Americans, Dominican officials assert that the number of deaths in recent months is no greater than would be expected statistically in a country visited by more than 2 million Americans each year.

The United States government, which holds remarkable sway in the Dominican Republic compared with some other parts of the Caribbean and Latin America, appears to support the Dominican government's contention that the alarm over the deaths may be exaggerated.

https://www.nytimes.com/2019/06/23/us/dominican-republic-tourist-deaths.html

 

[Tedsson]

https://www.cnn.com/2019/06/23/world/hard-rock-dominican-republic-tourists/index.html

The Hard Rock Hotel and Casino is removing liquor dispensers from guest room minibars at its resort in Punta Cana, the Dominican Republic, the general manager of the resort told CNN.

Hard Rock Hotel and Casino decided last week to remove the liquor dispensers and hopes to "provide more tranquility for guests," GM Erica Lopez said. The decision to remove the dispensers was made independently and not as a result of the two deaths that happened at the Hard Rock Hotel and Casino at Punta Cana, Lopez added.
The decision followsa series of American tourist deaths in the Dominican Republic, some of which may have involved liquor.

Media scare spoils vacation for alchoholics! I never knew there was such a thing as a hotel room "liquor dispenser"

Possible a good idea though, regardless of if it's being refilled with hooch.

Ah come on Mick. That looks almost exactly like the back wall of any bar in England. Even down to the Beefeater, Black Label and Smirnoff.

I guess that by "liquor dispenser" they mean what Brits would call the "optic" i.e the bit you can see above the three pronged press dispenser.

Heavens knows why they bother locking the bottles up as there is nothing to stop you drinking the lot of them (unless there is something very clever behind that white bar).

Well, I suppose there is common sense.

 

[mrfintoil]

Baader-Meinhof phenomenon, good to finally have a name for this. I remember back in the days when discussing "chemtrails", believers often stated that they had not seen persisting contrails in the past, therefore persisting contrails is a new thing that should be met with suspicion.

While persisting contrails were likely harder to spot in the past due to lesser air traffic, lower bypass engines, ect, it was certainly not impossible to spot them decades ago.

I remember when I got a new, rather unique haircut many years ago. I really had not seen it on anyone before, until I got the haircut and started noticing that quite a few actually wore it besides me. It was quite tempting to believe I started a fashion trend...

 

[Dawnrazor]

We stayed in Punta Cana last year just down the road from the Hard Rock Hotel. The optics in our room were exactly like in that pic! Strange how my wife and I barely used them when it was free like that.