What Does The Data Say?
Why understanding the bigger picture is essential when we want to see what the data is telling us
We humans have brains evolved to cope with modest challenges such as working out who we can ally ourselves with and who we should avoid so as to maximize our benefit from being part of the group. We can, more or less, work out that putting our hands into fire is not the best possible idea, and that sharp things can cut us.
Beyond that, we’re consistently clueless. We live in a world of undigested simple-minded notions that are almost entirely wrong. We cling to the simple because we can’t grasp the complex. We have zero capacity for understanding causal chains and therefore we are always unable to predict the eventual consequences of our actions.
None of this is our fault, any more than it’s the fault of an orang-outan that it can’t fly, or the fault of a shark that it can’t perform integral calculus. We’re just evolved this way: to do as little thinking as possible in order to conserve blood glucose, which for nearly all of our evolutionary history was scarce and uncertain due to our precarious hunter-gatherer existence.
What this means is that we’re truly dreadful at dealing with the real world. Although a tiny number of people alive today benefit from the intellectual labors of those who have gone before, most people live in a world of total ignorance and magical thinking. Even those we task with increasing our reservoir of knowledge are not exempt. Most science is, sadly, not worth the paper it’s printed on — even when it’s not printed on paper at all. This is because one can complete a PhD and yet receive little or no instruction on how to design experiments so as to make the results meaningful, nor instruction on appropriate ways to analyze data. In consequence nearly all science is in fact garbage, which is why for more than twenty years more thoughtful scientists have been worried about “the reproducibility problem” whereby most results turn out, on closer inspection, to be irreproducible because they weren’t genuine results in the first place but merely accidents of poor design or atrociously inept statistical analysis.
If we can’t even rely on scientists to cope with reality, what chance do ordinary people have?
The answer, obviously, is: none whatsoever.
And we ought to bear that in mind whenever we stumble across some claim, whether it’s made by a supposed expert or in a celebrity blog or a national news source or anywhere else. Decisions made on the basis of misleading claims will never result in good outcomes. Human history is littered with stupid policies resulting from a toxic combination of incorrect reasoning combined with political opportunism. No wonder today ordinary people often feel it’s all too complicated and the best thing to do is sit on the sofa gawping at some mindless entertainment.
There is no shortage of examples of spurious reasoning leading to catastrophically inept policy-making. To take a currently fashionable example, let’s consider the fact many of us have been working from home over the last months due to government restrictions on our freedom of movement. Lots of excitable commentators have been chattering about how this is the way of the future and how we’ll all be working from home from now on. In case anyone’s interested in why this in fact will not happen, I’ve written about it here.
Meanwhile, “experts” are busy telling us about the implications of this new mode of existence. A recent paper by Stanton and Tiwari at Harvard University purports to show that less well-paid workers suffer from being made to work at home because this translates into higher rent or mortgage payments. The argument is that poorer people discover they need more space when they work from home and therefore are “forced” to buy or rent larger accommodation. This shifts the cost of workspace from the company (which used to lease office space) to the individual. Our two brave academics conclude that companies therefore should be encouraged to pay their less affluent employees more in order to compensate them for this increased expense. Or, conversely, that we should all welcome a return to the office because it is “fairer.”
Oh dear. Once we’ve all stopped shaking our heads at the obtuseness of professional academics, we can begin to pick apart the argument to reveal its many flaws.
First of all, there’s the underlying assumption that people will rush out to buy or rent properties that they can’t really afford. Personally I suspect this assumption is valid, as most people seem to have a very imperfect understanding of their finances. According to our Harvard academics, this on average will result in a 7% increase in what’s spent on accommodation by less well-paid families, while wealthier families will see no increase due to the fact they already have a superabundance of rooms they can use as private office spaces.
Next, our academics instantly jump to their conclusion: companies should increase the wages of less-well-paid at-home workers by around 10% in order to compensate for this increased expense or, conversely, welcome all the low-paid workers back into the office so they don’t need to buy or rent larger properties.
Personally I do think that large US corporations under-pay their most vulnerable employees and I’d like to see meaningful pay increases for such people and rather less extravagant compensation for upper-tier executives who, frankly, aren’t worth anywhere near what they’re raking in. But that’s got nothing to do with work-from-home.
To see why, we have to step back and look at the bigger picture — something apparently not considered by Stanton and Tiwari. Wealthy people don’t struggle to pay for their automobiles and the running costs incurred by them, nor do wealthy people worry too much about paying for nannies, child minders, and other home help. But when less well-off people have to commute to the office each day, those costs represent a significant percentage of their net income. Costs which are largely or entirely avoided when working from home.
As a result, the seeming “work from home tax” imputed by our two eager Harvard academics turns out to be entirely illusory even if poorer people really do rush out to pay for larger accommodation — a proposition currently neither supported nor disproved by available data, which is very sparse.
If we imagine simple-minded politicians (and really, there is no other sort) in less far-right States rushing to create public policy to compensate for this seeming injustice, it’s easy to see that there would be two obvious consequences. The first is that less well-off people would be required by their employers to commute into the office as per normal, and therefore that they’d be trapped by geography instead of being able to look further afield as remote workers. The second is that if organizations were compelled to pay their lower ranks a better wage, they’d simply hire fewer people and make them do even more work than is the case at present. So the likely outcomes of any “fair” public policy based on the flawed analysis by our Harvard friends would in fact hurt those it would be intended to help.
Which is, sadly, just what we’ve seen so many times before. Data from Europe seems to indicate that if we really want to ease the lives of the less wealthy then the simplest and most effective solution is to set a mandatory minimum wage per geographic region that permits an adequate standard of living. Every other approach merely creates unintended consequences, nearly all of which are born by the unfortunates the policy is supposed to help.
That caveat per geographic region is, by the way, essential. Nationwide minimum wage standards create terrible distortions in even quite modestly-sized countries and much greater problems in large countries. Here’s why: Image the USA sets a minimum wage of $15 per hour. Assuming a 40-hour work week, that equates to around $31,000 per year, or what we otherwise term a pittance. But in some parts of the USA a two-income household totaling $62,000 per year is easily sufficient to provide shelter and food and leave enough over for other necessities, making that family relatively comfortable in their region. In the San Francisco Bay Area, however, the official poverty line is set at $123,000 per year per person due to the exorbitant cost of living, so a $15 per hour minimum wage would be totally inadequate. Yet setting the national minimum wage at a level suitable for the SF Bay Area would result in non-skilled workers becoming unaffordable for a wide range of employers across the USA and would consequently cause high levels of intractable unemployment because Molly and Joe at the supermarket checkout lines would likely not be able to retrain as computer programmers or project managers in order to find replacement jobs.
As is usually the case, we need to understand the bigger picture before leaping to facile but erroneous conclusions. Data is good, but data always has a context.
Now let’s skip sideways to consider SARS-COV2, our presently most fashionable cause de terreur, that’s provided more financial comfort to media organizations than any number of plane crashes or terrorist outrages.
I’ll start with a confession which is that when this whole thing began I made a very poor assumption, namely that SARS-COV2 would have a similar infection profile to the earlier coronavirus that gifted us the SARS outbreak at the beginning of this century. It turned out, as more data became available, that covid-19 was a great deal more infectious than its predecessor and thus spread far more widely. Fortunately it also was, as is usually the case with viruses, far less deadly. Today, even in the very worst-hit nations, the SARS-COV2 mortality rate is under two-tenths of one percent, and much of that is accounted for by doctors killing their patients by inducing coma and shoving them onto ventilators, which has been shown to kill more than 80% of cases.
As the media revels in presenting context-free statistics that inflate our sense of danger and induces politicians to impose policies designed to minimize loss of votes through creating the appearance of “doing everything possible” we would be well advised to remember the bigger picture.
As of today, 3rd March 2021, the WHO has recorded approximately 2.55 million deaths worldwide attributed to SARS-COV2. This sounds like a huge number until we put it into context. Firstly, it accounts for only around 3% of all deaths over the last 15 months, and most of those who have died have either been very old, very frail, or obese. In all these cases, life expectancy was modest, so had covid-19 not emerged most of these people would have died within the following months anyway. Meanwhile, more than 17 million people have died in the same period from lifestyle-induced causes (primarily obesity and smoking).
In other words, while we’ve all been panicking about coronavirus, nearly six times as many people have dropped dead because they couldn’t stop putting harmful things into their mouths. If we really cared about minimizing mortality we’d be focusing on getting rid of cancer sticks and McSlop so we could save six times as many people every single year than have died from covid-19.
Instead, we’ve all been running around screaming and waving our hands and politicians, as usual, have responded with catastrophically stupid policies. We’re all obliged to wear facemasks and endlessly sanitize everything in sight — except these measures clearly have zero effect because we were then required to go into lockdown. As any acknowledgement that hygiene theater is pointless would make politicians and their advisers look stupid, no one is owning up to the fact. As for lockdown, we’re told it is saving lives. But… what does the data tell us?
Fortunately there has been enough variety of response that we can evaluate the efficacy of lockdown measures. As we all know, Sweden has been very irresponsible indeed and has failed to panic in the approved manner. We’re told that Sweden has paid a very high price for its refusal to believe SARS-COV2 is an existential threat and has thus seen astronomical rates of mortality. But what does the data tell us?
As of 3rd March 2021, here’s the ranking of nations by per capita mortality attributed to SARS-COV2 infection:
As we can see, there’s clearly no overwhelming evidence here for the efficacy of lockdowns. Lockdown-free non-mask-wearing Sweden has not experienced anywhere near the mortality rate of Belgium or the UK, both of which have imposed severe restrictions. Interestingly, Belgium is right next to the Netherlands (which has had very light-touch restrictions) with a similar size of population with similar demographics and yet Belgium has experienced more than double the mortality rate of the Netherlands. Clearly lockdowns, social distancing, facemasks, and endless sanitizing have essentially minimal effect on outcomes.
Given their catastrophic negative effects (mass unemployment, massive increase in national debt, steep increase in suicide rate, steep increase in domestic assaults, crippling interruptions of education for school-age pupils, huge increase in mental health problems) we really need to be very sure that panic measures don’t create more harm than good. Sadly, in our modern media-driven world, we panic first and think later (or, more usually, never).
Another revealing accidental experiment has occurred in the USA. As reactions to SARS-COV2 have been driven exclusively by political affiliation, we’ve seen two opposing postures. The Republican response has been to claim that the coronavirus is simultaneously (i) a hoax, (ii) a Chinese bioweapon, and (iii) a plot by Bill Gates & George Soros to inject everyone with microchip tracking devices on behalf of the New World Order or, perhaps, the Lizard People. On the Democrat side we see the relatively harmless coronavirus perceived as the greatest existential threat humanity has ever faced and one that requires a total change in our way of life forever and ever unless we want everyone to die tomorrow by 11.20am.
Both reactions are stupid. They do, however, provide us with two very interesting datasets. Republican Texas has largely refused to implement any measures while Democratic California has implemented the strictest lockdown in the nation. Consequently we can compare results to see what, if any, difference lockdown makes to the mortality rate inflicted by the coronavirus.
As of late February 2021 California had experienced 104 covid-19 related deaths per 100,000 people. For the same period Texas had experienced 127 covid-19 related deaths per 100,000 people. At first glance it seems that strict lockdown measures could be responsible for reducing the mortality rate by around 20%. Not a fantastic result, but not totally insignificant either.
Now let’s look closer.
California and Texas both have large populations (around 40 million and 30 million respectively) but their demographics have significant variations. On average, Texans are fatter, poorer, older, and have far less access to healthcare than their Californian equivalents. As we have plenty of data to show that obesity, income, age, and access to healthcare are all major factors that influence covid-19 related mortality rates, we need to adjust our analysis of the raw data to take these factors into account.
Why is this important? Imagine that we have two places on Earth that are seeing coronavirus deaths. Both these places are doing exactly the same things, but their outcomes are very different: Place A is seeing 200 deaths per 100,000 people while Place B is seeing only 50 deaths per 100,000 people. How can this be, if they’re both following the same public health measures?
When we look deeper we find that Place A is full of old obese poor people with no access to healthcare services while Place B is full of young healthy middle-income people with access to good state-provided healthcare services. The enormous difference in mortality rates is explained not by any differences in their public health measures but entirely by their demographics.
This is why it is essential to adjust for demographics before we can understand what the data is really telling us. For example, in California 24% of the population is obese, whereas in Texas 35% of the population is obese. In California 8% of the population has no health insurance, while in Texas 18.5% of the population has no health insurance. When we take these factors into account we discover that Texas is seeing far fewer covid-19 related deaths than we’d expect. In other words, it’s possible to argue from the data that lockdowns actually kill more people than having no lockdowns at all.
In reality however, the most we can actually say from the available data is that the case for lockdowns is very flimsy indeed. We can definitely say that the huge costs imposed by lockdowns are totally disproportionate to even the most optimistic interpretation of benefits.
In essence, we’ve cut off our legs in order to “save” ourselves from a potential ingrowing toenail. Worse than that, we’ve provided a superb breeding-ground for endless right-wing nutjobs to recruit poor saps for the various fascist causes that have become mainstream Republican dogma. And this, in turn, may ultimately create far more deaths than are imagined by even the most lurid covid-19 sensationalism.
The above are just two examples of the kind of folly that results from us failing to take context into account. Sadly, there are literally thousands of similar examples all around us, with adverse impacts large and small.
Perhaps one day we’ll manage to learn that the big picture really matters. Until then, our endless march of folly will proceed as we stumble and waddle our way blindly towards the total collapse of our ever-more-fragile civilization.