Tag Archives: ons

UK covid stats confusion

There is a lot of confusion, both in the MSM and sites like nakedcapitalism as to what is going on with Covid-19 in the UK. The PHE (daily….ish) stats suggest the number of cases is falling and has been doing so for almost a week now. Hooray?…. But the (lagging) ONS data doesn’t show this. Maybe the ONS data will show the same when it “catches up”. It is about 2 weeks behind in terms of data collection, analysis and interpretation. It is “better” in that it is more of a random sample so bias (in terms of who actually gets tested) is theoretically less, but never under-estimate the interaction between humans (via their psychology) and a supposedly controlled trial. If people don’t want to participate because they suspect the “real result” will annoy them they’ll find ways to mess with you, the statistical designer.

Thus, perhaps the lower rates of positive results in the former are simply because lots of people being “pinged” by their phone app and others who strongly suspect they might have Covid-19 but who aren’t very ill are simply “going to ground” and not getting tested (in fear of being “grounded” for 7-14 days). BBC radio the other day informed people “if you’re pinged you don’t have to self-isolate but if you’re directly contacted you do”. Is that true? And if so, how many heard it and are following it? I dunno.

At this stage there is simply no robust way to know what is going on. People are impatient, however, and insist on speculating so as to “be first to the punch” in getting the interpretation correct, rather than waiting 10 days. Some of these people should know better. When someone blatantly speculating with anecdotes, reported here, only has their official qualifications reported, rather than details of the geographic and other factors that might help us put into context their anecdotal data, then any good statistician should get suspicious. If you live in a North Norfolk Westminster constituency where the MP not only got a plurality but a large majority at the last election (for the ruling Conservative Party) I don’t really think you live somewhere that is remotely representative of the country. When I know your professional (GEOGRAPHICAL) positions, and, via my higher education and friends (living across East Anglia), know practically all public transport routes you might have taken (if not driving) to get to anywhere, and know you are NOT encountering “the average Brit” then I won’t apologise for not paying much attention to your anecdote.

Don’t get me wrong. Anecdotes are SOMETIMES useful. Sometimes an anecdote forms the germ for a theory that ultimately leads to a massive paradigm change. However, if you’re going to quote an anecdote you should be allowing someone like me to put it into proper context. I’ll put more stock into an anecdote from someone who lives life “close” to the knife-edge of what is going on in terms of (for instance) reactions to Covid-19 and this inevitably, from what we know already, involves living close to BAME people, people on both sides of the median income and other such measures that, for better or worse, have come to define “class” in modern Britain. So, here are my personal experiences over the last couple of days, followed by some factors that should enable any competent “data person” to get a “picture” of where I live, how people behave, etc.

What happened today when helping my mother do her weekly shopping:

We went to Sainsbury’s (posher) and then Aldi (less posh – lots more lower income or “class” people). At Sainsbury’s my mum’s first comment was “why is it not freezing in here like usual?” – the temperature was barely below the 17 degrees celcius outdoors. I replied “they have turned the a/c and filters down to minimum to lower costs. Of course this reduces air circulation so don’t linger around people or areas”. (Nice of you to care so much for our welfare Sainsbury’s!) I also noted that the percentage of people wearing masks was no more than about 30%. I’m defining mask usage as “proper mask usage and not using it as a chin diaper as South Park so memorably put it”. Massive staff shortage but not across the board. Staff were “encouraging” (quite forcefully) people to use self-serve check-outs rather than the normal human-operated-tills. We then went to shop 2.

Aldi was freezing and well ventilated (just like before lockdown ended on 17th). Percentage of people using masks (and correctly) was definitely above 75%. Tills responsive to demand, so the queues were much shorter than Sainsbury’s. Us “deplorables”(?) were in the majority but were behaving pretty much in same way as before restrictions were eliminated.

Shopping across Arnold (one of the main suburbs of Gedling – our Parliamentary Constituency which famously was a key brick in the “red wall” that fell to the Tories in 2019 – and one of the main suburban centres in the “donut” surrounding Nottingham City) saw widespread mask usage. Quite a few people pulled them off when exiting shops, but they did that pre-17th too so no change there. There was NO visible change in mask usage compared to the lockdown period. Some people wore masks incorrectly, some pulled them off as soon as they walked down the high street having exited the supermarket, but the tendency had not changed.

Getting a wider context?

A distateful comment made by one was that the poorer stupid people are the “offenders”. What I saw today was the exact opposite. In fact, I’ve lived in the “poshest” postcode in Australia, which also had the highest rate of pertussis (whooping cough) 2010-2011 when I lived there 2009-2015. (Data now “conveniently aggregated into regions – didn’t wanna show up the sitting PM?????) A friend of my age got it. He, like me, had been vaccinated as a child but unfortunately the vaccination isn’t lifelong and when it was given in the 1970s they assumed it would eradicate this terrible disease and didn’t count on middle class wankers who preferred “to see their alternative therapist who did crap with stones”. I’ve said this before. Physicians in primary care in the UK quite like the “less educated” when it comes to public health campaigns because such people just “follow the rules without question”. You might question whether morally that’s good. But it’s just an observation and one that makes physicians’s lives less intolerable.

Oh, more observations. Half the people entering Sainsbury’s used the hand sanitiser on their hands and shopping trolley (cart). Close to 100% of Aldi customers did so. Aldi have also still got the perspex separators between checkouts. Sainsbury’s has removed them all.

Another observation regarding how “people might be avoiding pubs” – maybe it’s a SUPPLY issue, not a DEMAND issue? Thought about that? ALL THREE PUBS in walking distance of my house, which had served food (under very strict rules, eating outdoors at distanced benches) during lockdown, ironically STOPPED serving food, and REDUCED their intake of drinkers on 17th. Why? No doubt due to to the “pingdemic” and a lack of staff. Only now is food and drink serving capacity beginning to increase again, and very very gradually.

A note on sampling when quoting an anecdote

A very statistical person like me who has also had a lot of exposure to qualitative variability will know how the qual people deal with variability. It is actually quite a clever process but must be used with care. They use “purposive sampling” quite often. The idea is that you must cover “all key groups” but you don’t aim for representativeness of the population. Thus you get 5 straight whites, 5 gay whites, 5 straight blacks, 5 gay blacks…….you get the idea. This way no key group is omitted just because in the “wider population” they comprise <5% (or some other smallish threshold). If you want to quote an anecdote that might hold water nationally in order to provide a potential generalisable explanation for an odd result, you should be purposively sampling to see what “each key group is doing”.

Obviously, if you live in a place that:

  • Was one of the first to house significant BAME populations
  • Has a wide span of incomes
  • Has a good span of sociodemographics
  • Has a good span of any other health/other factors we know to be relevant to COVID-19

Then you probably are in a better position to make statements about “what is going on” and “where it might lead”. Nottingham is one of the best places in the UK satisfying the above criteria. Leicester is another, but Nottingham has a (perhaps by accident) very good electoral “map” in terms of “unitary authority” (autonomous city) surrounded by “donut” of district/borough councils with differing degrees of affluence.

So what do I think?

It’s not about ONE factor. It never is. Multivariable analysis will tease out the effects eventually but in the meantime, having access to various groups we “would sample via purposive sampling” I get insights into what different types of Nottinghamian are doing.

  • BAME groups are profoundly suspicious of the vaccine and COVID itself – this is madness but given historical experience, this is more understandable. My cousin’s wife is BAME and refuses the vaccine, thinking it’s a white plot. FFS.
  • Students are know-it-alls with under-developed senses of their own mortality. My nephew (undergrad student) was just diagnosed with covid. Many have frankly execrable mathematical ability.
  • Richer people can be similar. Think they know better. If you want my view, read Douglas Adams’s books.


Stop looking for single explanations

Class/income is a nice explanation. Unfortunately the data don’t support you. THe wards/areas mentioned are rich/full of students or BAME. Indeed student arrogance has been re-recognised just a week ago.  BAME groups are often the least vaccinated, despite being at the highest risk. Posh rich people are also often anti-vax. So are students – look at the wards in Nottingham with low vaccination rates – those of use who live in Nottm region recognise these as BAME and/or student wards. OF course a lot of students are mathematically innumerate and think they can lecture me on the topic. Hmmmm. Anyway rates are falling across wards no matter “who might gain – Tory or Labour” – hardly implying massively reduced testing.

IF you are going to quote an anecdote to make a major point regarding a CRUCIAL statistic then you must be prepared to provide a LOT of supplemental information about “your circumstances” so anybody with knowledge can check up on it. Otherwise how do we know you aren’t a troll spoofing someone or you are real but getting a grossly biased picture?

I am well known in terms of where I live and my background and circumstances. When I give an anecdote it’s really really easy to establish if I’m “just seeing something local”. I DO NOT CARE IF YOU ARE THE BEST PERSON IN THE WORLD IN A FIELD OF ECONOMICS i RESPECT – IF YOU QUOTE ANECDOTES WITHOUT THE REQUISITE INFO ON YOUR LOCATION AND LOCAL CHARACTERISTICS YOUR VIEW IS NO BETTER THAN ANYONE ELSE’S.