(Based on a link submitted by reader Sean Duncan)
The Independent Investigations Group, which as you loyal, dedicated, and detail-oriented readers know is a Los Angeles, California-based organization that investigates claims of the paranormal and pseudoscience, is affiliated with The Odds Must Be Crazy and provides a lot of our support and backing.
The IIG regularly receives all sorts of communication regarding a wide variety of topics, including requests for advice on how to handle unusual situations related to the IIG’s fields of expertise. In this case, listener Sean Duncan decided to write in and get the IIG’s assistance with a subject he’d been discussing with a friend. Here is that email:
My name is Sean and I live in Shelton, WA. I’m emailing because a skeptic of skepticism asked me about how, sometimes in disasters, thousands of people will die in a particular building yet one will survive for days or weeks because they are in the right place at the right time. I told this person that I would contact the Independent Investigations group because they like to calculate the odds of things. With so many people calling this phenemona a miracle, it might make for a good segment on The Odds Must Be Crazy. If you have the desire to calculate the odds of this Bangladesh woman surviving 17 days, we’d both appreciate it.
Through this communication a long discussion thread was started to address the question and build a complete picture of what would be required to answer it. We found the results really interesting, and have decided to share some excerpts with you below:
Comment by Barbara Drescher:
There is absolutely no way to calculate the odds of such a thing; it would require knowing everything about the building at the time of the collapse as well as defining the context (e.g., the odds of surviving, given that one is in the building when it collapsed, or the odds of it collapsing right when one is standing in that spot?).
How I would respond to such a question would be to ask more questions. If it is a miracle that she survived, then what is it that the other 1100+ people died? How many people do you think survived for several days, but died before they were rescued? Would it be less of a miracle if it was 15 days or more of a miracle if it was 18 days?
This kind of thinking is flawed because it is “post hoc”, or after-the-fact. Given what we know happening, the odds of that happening are 100% (because it already happened). Even if we predicted that a survivor or two would be found this long after, it’s still not remarkable because it happens. People will always be “in the right place at the right time” and “in the wrong place at the wrong time”. When we think about all of the circumstances that must be “lined up” for such a thing to happen, it looks remarkable, but something has to happen. Some set of circumstances is going to be the set that occurs. Someone will eventually win the lottery.
I’m reminded of the research that Hugh Ross did in which he calculated an outrageous probability that the universe would produce human beings. It was so outrageous that he concluded that it must have been an act of God. However that kind of thinking is exactly like asking someone to pick a number between 1 and 600,000,000,000, then being shocked by the number they picked, given that the chances of them choosing that number were 1 in 600,000,000,000.
Comment by IIG Chairman Jim Underdown:
It reminds me of two related issues. The question is like asking what are the odds of surviving a car crash. It depends on the car, the speed, the driver, what it hit – and countless other factors.
The post hoc example I like is the paint bucket that fell off a ladder. What are the odds it would produce exactly that spatter pattern? … 100%!
Comment by Jerry Schwarz:
It may be important to emphasize the difference between the probability that a specific person will survive for that long and the probability that one out of the thousands of people in the building will survive. I suspect that many people don’t understand that difference.
Comment by IIG Steering Member Dave Richards:
The kind of statistics I have a real problem with are ones where there’s a bimodal or multimodal aspect. For example if you plot the ages of death for 10,000 individuals on a histogram, it won’t be a nice bell curve – there’s going to be a big spike in infancy due to childhood diseases, another spike in middle age from heart attack and stroke because that’s when those usually happen, more spikes from various cancers for people that outlive the other stuff, and then finally a spike when the body just finally gives out from old age. To boil such a spiky graph down to a single average age for longevity is pretty much a useless statistic. But this kind of thing is done all the time in news articles.
Jim Underdown responds:
I guess I’m arguing that because each car crash (plane crash, building collapse) is unique, and survivability depends on lots of factors dependent on that particular crash, making general predictions (or assigning odds) about someone surviving any such incident would be beyond the amount of useful information you’d ever have access to in a random event like this. The odds you’d come up with in your car crash statistics might easily be useless unless you added in lots of other controls like speed, car make, alcohol, etc. A sober person who never drives a Mack truck more than 20 miles an hour will be well beyond the insurance company’s risk tables. (Sort of along the lines of shark attack risk for those who never go near water.)
The correspondent is interested in whether we can assign odds to her having survived. There’s quite a difference between calculating the odds that someone would survive, and that this particular woman would survive.
Barbara Drescher rounds up the strategy of developing probability:
Starting with a very specific question is essential and without one, it’s not even possible to guestimate.
And I think that’s the disconnect that people have when they think of these kinds of occurrences as miraculous (I don’t think it’s relevant whether they consider it an act of God or just a really amazing coincidence). Post hoc thinking has the luxury of being vague, but it’s not the vagueness that makes it bad.” And following up: “I just don’t see how that’s relevant. The question isn’t about how statistics are used. It’s about whether an event is extraordinary, probabilistically speaking.
IIG Steering member Spencer Marks adds:
… the way I read the question about the odds didn’t seem (to me) strictly a question about the odds of surviving the collapse of the building, but of the survivor living for 17 days. That question of course is ALSO not a matter of “odds,” but of many different environmental factors such as the ambient temperature, perhaps humidity, his availability to water, his general condition before the collapse … Like Barbara said, this is not a matter of odds but purely biological and physiological science at work, and that should be mentioned!
Bay Area IIG member Leonard Tramiel summarizes:
There is a very good reason that the odds here are different. It’s related to the reason that it is considered a “miracle”.
It happens rarely. We can state the odds of being in a car crash because this happens many times every day. Surviving a building collapse for more than two weeks … not so common.
Given the poor statistics, we are forced to consider computing the statistics and that is hopeless for either building collapse or car crash.
Overall we found this was a rather interesting (you can feel free to disagree with us without hurting our feelings) look into the thought processes that sometimes go into analyzing stories like this.
We now return you to your regularly scheduled stories. There will be further interruptions.