Covid, Ghostbusters, and Fraudulent Data
First, something from the UK:
I will give you a little backstory on this, because Prof. Morris is reacting to the following graph that has been circulating on twitter:
I have no clue why somebody put the data labels on the graph, as they do not help in any way. But let me make it simple: the orange line are the vaccinated! And they have HIGHER MORTALITY!
THAN THE UNVACCINATED!!!!
The extremely simple explanation for this state of affairs is that this bucket covers an age range of about 50 years, and that the most vaccinated folks tend to be the oldest, and the unvaccinated folks are the youngest.
There is a huge risk differential for COVID mortality by age, and basically, extremely few kids die of COVID.
Also, we’re talking 1 vs. 3 per 100,000 people per week. These are already very low death rates.
Professor Morris’s point is that those who have watched only Ghostbusters 2021 are most likely very young, and those who have watched only Ghostbusters 1984 will be over age 50.
Here are more serious responses:
Covid-19 Actuaries Response Group: Simpson’s Paradox and Vaccines
UK Office for National Statistics: Coronavirus Deaths: Understanding ONS data on mortality and vaccination status
The Ever-Returning Simpson’s Paradox
I’ve written about this earlier:
In all the cases, you have the same dynamics:
Old people have much higher vaccination rates than younger people
Old people have much higher mortality risk than younger people (whether from COVID or other causes)
Even when COVID risk is reduced for old people, if you aggregate all the vaccinated together, you’ll get a lot of vaccinated old folks dying
And, sorry, 59 years old is not young. It’s the tail end of the Boomers. You guys are old, and there are serious mortality risks. Even after vaccination, your risk levels are higher than a bunch of unvaccinated 10-year-olds, is this really so difficult to understand.
Be Aware of Fraudulent Data
This one doesn’t actually have to do with COVID, but something that came up some months ago.
I wrote a piece on it, Distrust and Verify:
I learned a lesson about reliance on data from an article I published in February 2013, “Everybody Cheats, at Least Just a Little Bit.” It was a review of Dan Ariely’s book, The (Honest) Truth About Dishonesty.
It turns out that an insurance example I used to illustrate the theme of the book was based on fraudulent research.
Maybe I should have taken a hint from the title of the book.
It turns out a key example in the 2013 review was based on fraudulent data, and how that was detected and the nature of the fraud is explained in this Data Colada post:  Evidence of Fraud in an Influential Field Experiment About Dishonesty
If you’re the type who prefers videos (or just audio), here’s a video of me discussing the issues:
It annoys me that I was so credulous as to think that not only did moving the attestation had an appreciable effect, but that an insurance company would ignore an appreciable effect in a business with razor-sharp margins.
I don’t want to speak about the day job, but sometimes I laugh at the fatuous statements of the various “We’ll show those fuddy-duddies in the insurance industry!” newcomers who find out exactly why those fuddy-duddies have certain “weird” barriers in place. In some cases, it’s a matter of both insurers and regulators learning from centuries of insurer, consumer, and capital market behavior of how these things fail. Just a little hint. In some cases, the issue isn’t insurance per se.
And I will shut up about insurance, and say Happy Thanksgiving!
For those who enjoy Turkey Day, my traditions are: Macy’s parade, sumo tournament, and the ever-present MST3K Marathon!