You might notice that the plots I post today look a little different from previous days. Here’s a thread about why that is, with a little bit on scientific communication, psychology and politics, and reflections on my own motivation.
I’ve previously posted plots in portrait format, because I figured a large proportion of views would be on phones. But some noted that Twitter lops off the top of graphs when plots are in RTs, making graphs look rather benign if the post wasn’t clicked on.
There’s a clear preference for the landscape version, so I will switch to that. However, a third of respondents expressed a preference for the portrait version, and a few people commented that the landscape version didn’t seem as troubling (and would be preferred by the govt!).
There are a couple of important points here. First, there’s a problem with stretching a short time frame over the longest dimension and squeezing an outcome variable that varies by a factor of 32 or 64 (5 or 6 doublings) into the shorter dimension. Here’s an illustrative example.
On reflection, I think the right way to deal with this problem is to show longer time frames in the plot, as in this one. March seems like a good starting point, given that’s when schools reopened for most children.
But one could start from the beginning of 2021, which allows the viewer to see the previous peak for 10-14 year olds (which we’ve now exceeded). I was going to say something else about the shape of this graph, but maybe I’ll leave that to someone else.
Or one could start from the beginning of the school year, in Sept 2020, which allows the viewer to see the whole second wave (and note that previous variants *did* cause lots of infections in children, and the rate changes correlated with schools opening/closing).
Arguments can be made for all these versions, but my goal is to show what’s happening now, so I don’t want to squash the last month too much. For an overview of how rates have changed across age groups over the last year, the heat map on the dashboard is a very effective plot.
Second, changing the format (portrait/landscape) and the axes has a big impact on how steep the curves are, and hence how “scary” they look, which brings with it a certain moral hazard. You should always ask yourself, “who’s making this graph, and what is their intent?”
If you’re interested in the cognitive psychology underlying perception of graphs, my colleague @STWorg Stephan Lewandowsky has done some interesting work, including experiments investigating the (ab)use of graphs as a tool for misinformation.
As for myself, I’m a scientist, and it’s important to me that the data are portrayed in a way that’s fair, and not misleading. It isn’t my intention to sensationalise or dramatize what’s going on. As in other areas of life, it’s important to #TellTheTruth
At the same time, it’s hard to avoid there being a political element to these graphs, especially at a time when there’s so much polarisation. When MPs argue that we should no longer even publish the data, posting these plots is itself a political act.
Though I’m a scientist, I don’t subscribe to the view that I should therefore be a neutral observer. I’ve been posting plots since the pandemic started, but not the whole time – I start posting plots when exponential growth becomes clear. I do that to make sure people are aware.
And every time, I do that for a while, with little evidence anyone’s paying attention, and life going on as if everything’s fine. Eventually I get so frustrated by that I can’t help ranting/shouting into the void. Here’s a recent example.
In previous waves I’ve lost enthusiasm for posting plots once it’s clear to everyone what’s happening – I’m more interested in warning people about what’s coming than documenting bad things as they happen. (I’ll go back to talking about the climate and ecological emergencies).
So I’ll probably give up on these plots again soon, as we move from “there won’t be another wave” to “surf’s up, who cares?” But I won’t have lost my anger at a government that shows such contempt for its citizens (esp the most vulnerable) and an “opposition” that cheers them on.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Here's the Covid rolling rate for #Bristol (all ages). It's up to just under 400 per 100K now. This is the linear version of the plot, but it's worth looking at the log version (next) ...
The log plot shows that exponential growth is continuing at the same pace. We'll exceed our January peak in a few days (in fact, we almost certainly already have, but have to wait for the 5-day data lag to catch up). There's something else worth noting though ...
I already posted this plot, which shows that the rate hasn't been growing quite as fast in 10-14 year olds in the last couple of days. So why is the overall rate for Bristol still growing at the same pace? The answer is that ...
Here's the latest graph of the COVID rolling rate for 10-14 year olds in England. It's gone up again, though not quite as steeply as a few days ago. That's another couple of thousand children testing positive (just in this age group).
Here's the log plot. The *slightly* shallower slope means the current projection is a rolling rate of ~2100 per 100K by 'Freedom Day'. That means over 100,000 10-14 year olds testing positive between now and then. Current estimates suggest ~8000 of them will experience #LongCovid
And if you're one of those people who likes to pretend that #LongCovid doesn't exist, you might instead consider that of those yet-to-be-infected 100K children, somewhere between 500 and 1000 of them will be hospitalised, and 1/4 of those will go into ICU.
Here's the latest update to the graph showing the Covid rate for 10-14 year olds in England. Still doubling every 7 days.
You can see the consistency of the exponential growth in the log plot. I've taken the liberty of extending the projection to July 19th, so that we can contemplate the Path to Freedom.
BTW, if you're wondering why the first graph looks a bit different, see this thread.
[THREAD] I can't understand why there don't seem to be more parents expressing outrage at the UK government's herd immunity experiment on children that is happening in plain sight.
What is going on???
The government's own figures (from ONS) say 8% of infected kids experience #LongCovid. That could be several hundred thousand children.
You don't want that to happen to your children, right? We don't know what the long-term effects are. For all we know, they could be lifelong.
I understand it's not easy for parents to take preventative action. There's the threat of fines, and how does one combine home schooling with work? But also, I suspect people are going with the flow:"other parents are sending their kids to school, so I'd look weird if I didn't".
Here's how the COVID rate has changed in 10-14 year olds since May. And if you think this is steep, wait until the Govt changes the rules to minimise self-isolation of bubbles. School's throwing a COVID party -- who wants to come? (you have to come) #HerdImmunity@SafeEdForAll_UK
Needless to say, this has real world implications that go beyond charts ...
Seeing as I've posted a few tweets linking Bristol's COVID outbreak to university students, it's only fair to note that the rapid rise in cases has not been restricted to this group. Here's an age group that I've been following (since it includes my children).
Note that the plot in the previous tweet was just Bristol 10-14 year olds. If you're interested in what it looks like for England, here's the graph.
In case, you were wondering -- it's not getting any better. Today's update brings the COVID 7-day rolling rate in Bristol 10-14 year olds up to 157 per 100K.