One of the best books I’ve read recently is “How to Read Numbers: A Guide to Statistics in the News (and Knowing When to Trust Them)” by Tom and David Chivers along with it’s accompanying website. Chivers & Chivers have written a short engaging tome that helps us all understand how statistics can be used well and also abused. It’s helpful stuff as we all seek to make sense of numbers being thrown at us every day, especially by journalists (the #1 area of concern of the book). As someone who’s A-Level, undergraduate and first professional training and education included statistics it was a delight to read. I think you might enjoy it too.
That statistics can be misused or mis-reported is a widely-understood truth, made famous by Mark Twain‘s (or was it someone else?) pithy statement:
Some might suggest that in the era of “fake news” there is an increase in statistical abuse. What’s for sure is that there are a lot of what should at least be called “misunderstandings” floating around. As Christians we want to be people who uphold and speak truth so how do we respond when the numbers are being fudged?
You can’t watch the news at the moment without seeing stats about Covid. Infections, growth rates, hospitalisation and so on. But none of these numbers are as contentious as the figures around vaccination. And this is where I want to make my first appeal to Christians – let’s be really careful and conscientious when we’re using numbers in this area.
There may be very good reasons to be hesitant to have the Covid vaccine; we might be concerned about the apparently rushed way in which it has been produced or the effective government compulsion to get vaccinated that some perceive. Those issues are certainly worth discussing and thinking through carefully. But the argument simply can’t be made on the numbers.
A good example of this sort of error is the recently published piece by Rod Lampard on Caldron Pool, “Underreported: Deaths After Vaccine Outnumber COVID Deaths, Australian Stats From Jan-June 2021 Reveal“. Now I know that Caldron Pool has critics aplenty but this isn’t one of those pieces. They’ve published articles in the past from me and I assume they’ll do so again in the future. I contacted them in advance about this one and they welcomed the conversation (which, in my experience, is more than can be said for left-leaning commentators and platforms). So this is an intra-mural conversation, perhaps “wounds from a friend” (Prov. 27:6). I’ve disagreed with some of their material before and agreed with other pieces. This is one that I want to push back a little on.
In his piece Lampard notes that,
As of June 28, 2021, 272 people have died after receiving COVID-19 vaccines in Australia. Only 1 person has died from the virus.
As of June 28, 2021, the COVID-19 fatality roll stands at 910.
Of those 910 COVID deaths, 909 occurred in 2020.
And here’s the chart from the Federal Health Department showing the stats being used (dated 24 June 2021).
Then, he refers to the “fear porn” being disseminated by governments and media around this issue in order to compel people into getting the vaccine.
Before I push back on this use of figures let me be clear. I do think that some of the narrative around COVID has not been helpful. I’m sure that I don’t hold that view to the extent that Rod and others on Caldron Pool do but I am still concerned about how some of this is playing out. I share their worries about any form of coercion through shaming those who are vaccine-hesitant. I think there are far better ways to persuade people. If something is a good thing then it’s goodness is what persuades.
Having said that, I fear that Lampard here has fallen into a classic statistical error which the Chivers brothers call the “Big Number” fallacy (chapter 9) with a bit of “Survivorship Bias” (chapter 20) thrown in for good measure. Here’s my argument, based on the very sources that Lampard helpfully links to in his piece.
The stats simply can’t be viewed in absolute figures. The real way to measure the relative danger of COVID and vaccinations is to see death as a proportion of those affected.
Australia’s Covid death rate, from the figures in the piece, is 910 out of 30408 cases – 2.99% (comparable to what has been seen in other places)
The vaccination death rate should be calculated in the same way. From the Therapeutic Goods Administration (TGA) link provided it’s 272 deaths from just over 5 million vaccine doses. Let’s be really conservative and assume that everyone has had both shots (ie 2.5m people receiving the vaccine – we know it’s higher, but let’s take that figure). So the vaccine death rate would be 272 out of 2.5m = 0.01%.
In other words Covid is some 275 times more deadly than the vaccine and probably (given we assumed double jabbing) even worse. If you were given a straight choice between having the vaccine or contracting COVID, there’s only one sensible option. Of course, it’s a more nuanced choice than that – not everyone will contract COVID. But even if 10% of us end up getting it (and that’s not unlikely in the long run) then we’re still roughly almost 30 times better off getting vaccinated, even if some recipients may even die as a result.
A few years ago my daughter had heart surgery. The surgeon sat down with patient and parents and discussed potential risks. There’s almost always a risk with any medical intervention. But we sat down and did the maths; the risk of surgery while real, was far less than the risk of doing nothing. So we had the surgery. It’s the same thing here with the vaccine when we weigh up the stats and on that basis I’ll be getting vaccinated and would encourage others to do the same.
One other thing. Yes, the death rates in 2021 are very interesting, but also quite possibly indicative of the effect of the vaccine and also how much we’ve learned since this whole thing started. There could be any number of reasons for a lower rate; lower incidence of COVID, less strained health resources, earlier intervention from experienced health practitioners etc.
Granted, Lampard’s article is really about the way that the media have covered this issue and I’m in agreement with some of his sentiments there. But on what the numbers tell us I’m afraid he’s mistaken, even if he’s right to tell us that others aren’t reporting these things well either.
Which brings me to another case study…
The ABC’s reporting of the Anglican Church’s research into Family Violence is another good example of selective use of statistics and some poor retelling of what is really going on. Again, as with Lampard’s piece on COVID vaccines, the effect is to undermine what are otherwise important things that need to be said. Nobody denied that Family Violence was happening in the Anglican Church nor that some responses in the past had been very poor. This is a topic that we will need to keep returning to and with urgency.
Nevertheless, the article is disappointing in its emphases and omissions. Consider the first line:
A landmark report has found the incidence of domestic abuse is greater amongst Anglicans in Australia than the general population.
It’s a broad unqualified statement: domestic abuse occurs in greater incidence amongst Anglicans than in the general population of Australia. And yet is that what the report actually claims and in such stark terms?
Not quite. This is what the headline report on the findings states:
The study was an online survey of over 2,000 males and females, aged 18+, conducted in December 2019. The Online Research Unit hosted the survey and provided the respondents. Results for a sample of the general public (n=1146) were compared with Australians who identified as Anglican (n=825). A larger sample of Anglicans (n=1382) was used to compare those who attended church regularly with those who didn’t. These were non-probability samples from online panels so representativeness to the wider population cannot be claimed.
National Anglican Family Violence Research Report
Top Line Results
Note that last sentence carefully.
- The samples were non-probability. They weren’t randomly selected albeit there was some structure to the panels that they were drawn from. (Chapter 4 of the Chivers book: Biased Samples)
- As a result “representativeness to the wider population cannot be claimed.
This is a technical statement. In the field of statistics there are “populations” (the group you’re trying to find something about) and the “sample” (the smaller group you select and examine in order to draw conclusions about the population). There are ways of selecting your sample that allow you to state with confidence that they are representative of the population. But in this case the study’s report wants us to be clear on something – while the research can show the existence of an issue (and it certainly does) it cannot be said to “prove” or “show”. This sample couldn’t do it because it was biased (another technical term meaning that it was skewed). In this case the sample was skewed towards those who chose to take part. People chose to take part in such research for a number of reasons, mostly because they have an active interest in the topic. So it’s possible that the research actually over-reports domestic violence. Or it’s possible that it under-reports it (and there’s reasons why it could go either way). We just don’t know. And the report makes that point too.
Another thing worth considering (chapter 5 of the Chivers book) is whether the reported difference between the general Australian incidence rates and the “Anglican” rates (if they could be relied upon) are statistically significant (do they fall within the “margin of error”?)? Again, we’re not given this information. They might be, they might not be. But how would we know? It requires some sensitivity in reporting.
Section 3.3 of the report sets out these difficulties in some detail (noting “because it utilises non-probability samples, it is not possible to claim that the whole population or all Anglicans are represented”) along with some steps taken in partial mitigation.
But the journalist presents the report as having demonstrated these things unequivocally. And that’s a shame because what it means is that when that mis-reporting is recognised for what it is it detracts from the possibility of the report being taken seriously on it’s own terms. And the report should be taken seriously; it’s well worth reading through for it’s detailed description of the nature of much of the abuse that is occurring and various responses.
Seeking and Telling the Truth
As always, here at davidould.net we’re equal opportunity offenders and I’m sure the above will have annoyed people from all across the political and social spectrum. There’s more than enough there for everyone to pick something to get upset about.
Alternatively there’s more than enough for the fair-minded reader to see that I’m actually a proponent of better and more responsible communication by governments and media over COVID and vaccines and I’m deeply committed to eradicating the scourge of Domestic Violence from our churches (and, of course, our nation) even though some will want to unfairly portray me otherwise.
How you read this article therefore comes down, in part, to whether you want to be fair with the information provided or not. But that’s my point about the whole thing. As Christians we’re lovers of truth; we don’t slander and misrepresent and we ought to not affirm when others do it either (however unintentional it may or may not be). The truth is never something to be afraid of, even if it doesn’t fit our agendas.
Grappling properly with this means we can be worried about government and media approaches to vaccine rates and see that vaccines are clearly safer than contracting COVID itself.
Grappling properly with this means we can be deeply committed to the eradication of domestic violence and be worried about how the topic is reported.
Grappling properly with this means seeking more truth, blessed truth and statistics.