RSS
The BBC and other media outlets have publicised a disturbing claim in a Parliamentary report, that racism has played “a key role” in the deaths of pregnant mothers.
End racial disparities in maternal deaths – MPs. An MPs’ report is calling for faster progress to tackle “appalling” higher death rates for black women and those from poorer areas in childbirth.
The Women and Equalities Committee report says racism has played a key role in creating health disparities.
But the many complex causes are “still not fully understood” and more funding and maternity staff are also needed.
This is mostly based on a previous official report from: Maternal, Newborn and Infant Clinical Outcome Review Programme November 2022 Saving Lives, Improving Mothers’ Care: Lessons learned to inform maternity care from the UK and Ireland Confidential Enquiries into Maternal Deaths and Morbidity 2018-20. (They have added some interviews with mothers who felt that their concerns were ignored for racist reasons).
The first thing to note is that pregnant mothers are very safe, with a death rate of 6 per 100,000. Even taking a year-long period, and including indirect causes, deaths are extremely rare. Suicide and drugs/alcohol are prominent causes of the few deaths recorded, with cardiac disease and Covid of equal impact. (Hatched bars show direct causes, solid bars indirect causes).
So, for example, Covid was an indirect cause of death, because it happened more widely, and not because of pregnancy.
Half the causes of death are not obstetric. This should be a warning bell about the interpretations later placed on the data.
Later data show that the greatest predictor of later death was pre-existing medical problems including obesity. Obesity is also given as a primary cause. Not receiving advice is also noted as a cause, though not clear if this can be due to not taking advice despite ante-natal classes being offered. Some of the case histories mention this, and it would be relevant to understanding causes, and could be compared to the take up of vaccinations.
Table 2.9 gives us the figures on regarding race differences, the subject of media attention on the Parliamentary statement, but also prominent in the summary of the research report.
The numbers, mercifully, are small. Childbirth is safe in the UK. The White European results are the only one sizeable enough to merit detailed examination, and they show that direct and indirect causes are of almost equal power, slightly favouring indirect. The racial group deaths are few, and some noise is to be expected.
To put this report into context, as regards UK Black Caribbean mothers, we are talking about 3 direct deaths, 6 indirect, so 9 in all. Black Africans (who probably have spent more of their lives in Africa), are 5 direct, 8 indirect, 13 in all. Combine the two, 8 direct, 14 indirect. Tiny numbers, but favouring indirect causes.
Here are the calculations of rates per 100,000
The rates per 100,000 differ considerably, but the base populations also differ considerably, and there are 18 times more whites than blacks. When events are very rare it is probably better to give absolute results, rather than relative risk odds ratios, which could be misleading.
In my view the main finding is that some mothers are unwell prior to childbirth, and often die for reasons not directly caused by childbirth. Some of those causes of ill health are self-imposed.
The Parliamentary group seems to have gone overboard on a racism explanation for which there is no evidence in this study. Indeed, there is virtually no data analysis in this report at all. They give the basic demographics, but few if any two by two tables. They do not show correlations, nor do they construct and test any explanatory models. Can they predict from their findings what puts women at risk? Drugs, drink, being overweight? Little from the authors on these necessary questions. A simple multiple regression on the total sample would have been instructive. The list of participants is impressive, the data analysis rudimentary, and the use of relative risk over-dramatic given the very big differences in sample sizes.
As is usual, they give stern lectures to service providers about the need to further educate themselves. They say less, if anything at all, about mothers keeping themselves healthy, for their own sake, and of their babies.
English
Arabic
Chinese (Simplified)
Dutch
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Persian
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Vietnamese
107 page report and they could not be bothered to do a multiple regression. Fascinating. Not sure how much it would have mattered though since there are so few univariate comparisons in Table 2.10. Consider the value of including the variables in Tables 2.14-16. Any idea how much information like that is available for all pregnancies? I would expect having the NHS would imply fairly comprehensive and uniform data. What do you think?
Given the focus on race it seems odd to me they did not emphasize how much the black RR had improved between 2015-17 (5.27) and 2018-20 (3.68). See Table 2.11. Given the wide CIs the p value isn’t significant, but still…
Figure 2.9 gives yearly data by race. It looks like there was a significant uptick in black maternal mortality in 2015 (2016? the x axis and title differ in the years they cite).
One wrinkle is that they no longer determine fine grained ethnicity directly for all births. Here is the relevant text on page 12.
Table 2.12 on page 15 gives the estimates. Note the footnote: “Country of birth not recorded for 20 women who died.” If any of those were in the small groups it could make those RR numbers significantly worse.
Worth noting that this is the ninth MBRRACE-UK annual report of the Confidential Enquiry into Maternal Deaths and Morbidity. I don’t see any nice compilation of past reports, but here is the 2014 report for comparison.
https://web.archive.org/web/20150518123755/https://www.npeu.ox.ac.uk/downloads/files/mbrrace-uk/reports/Saving%20Lives%20Improving%20Mothers%20Care%20report%202014%20Full.pdf
The 2014 report looks more like an academic paper and less glossy to my eyes. Interesting how things change. The 2014 report is longer (120 pages) and denser (two columns with less whitespace). It also appears to have much more about the medical conditions involved.
P.S. The report discusses online supplementary material and the first page asserts: “Compiled report including supplementary material.” Does anyone know where it appears in the PDF? It does not appear to be clearly marked.
The NHS should have the best data in the world, but sadly not so at the moment. I would liked to have seen some old fashioned vulnerability scores (totals of risk variables, for example) or some two by two tables for race and obesity.
Discursive, shallow, but firm in giving recommendations!
Many of the doctors and other health professionals in the UK are Asian, and the low level NHS staff are often black, so once again it is the white people’s equivalent to the cosmologist’s”dark matter” that’s causing the alleged racial disparities. Perhaps the NHS can book some time with the LHC to try to find it.
It’s a pretty good rule of thumb that He Who Plays the Race Card is in the wrong. But how often do they own up to lying about it?
I can think of one example: years ago the CPC admitted that the notorious sign in Shanghai that had said something like “No admittance to dogs or Chinese” was just a bit of agitprop put in place by the Party itself.
I can think of a counterexample: Senator Warren who successfully advanced her academic career by waving a metaphorical tomahawk.
In this case, ‘racism’ means that Europeans are better, and the gap is indeed caused by Europeans being better.
It means a second thing: they actually want the NHS to be more racist. The rate of pregnant black women dying is too low and they want it to be higher. The delta is too small. The same way BLM allegedly wanted blacks to die less often, yet consistently and persistently pursued policies that caused them to die more often.
Some want to create a wider class divide for ego reasons, and others want to be able justify a bigger budget. Some are genuinely motivated by spite and simply wish to cause suffering. These three groups have made common cause, as their interests align.
Bantu in England are indeed mere sheep who can be herded into whatever shape happens to be convenient for their betters. Death rates will go up, unless the line troops are too incompetent to carry out the new death-seeking policies. As an example, the recent rapid policy changes have almost certainly been too zigzaggy for them to keep up. Bureaucratic orders are coming down faster than the bureaucratic OODA loop and they’re functionally DDoSing themselves.
Who cares about dead black mothers?
ACOWW or Afro-Colonization of White Wombs is the real problem.
Get a load of this:
The white woman who was attacked by a black mob has a black lover. Blacks attack whites all over the US, but she has sex with a black man and will have a mulatto child who counts as black.
And even though blacks beat the crap out of her, she thanks a black woman for trying to help her.
Blacks attack whites, but white boys cheer for black athletes, white girls have black babies, and white victims of black violence thank token blacks for trying to intervene.
What a shi* country.
Is it Bottominum Toxin that’s killing them?
Quick, gimme a research grant.
Who knows. Maybe they did look at the connection to obesity, drug taking, and pre-existing conditions – but decided that the results were too inflammatory to be published. Blaming white racism conforms to our modern quasi-religion, and makes further analysis unnecessary. It is a cop-out.
The racism of affirmative action makes more black women never reproducing black master degree diversity hires and childless dying.
Margaret Sanger was a hero. Cuckservatives want more black babies. Ban the clinics in White areas and build big beautiful abortion clinics in black areas.
The United Kingdom would be a much happier place had they not let blacks move there.
If they feel that the UK is so racist, then they can go back to their own ancestral homelands where there are no White people at.
Another thing to keep in mind is that the people of England/Scotland/Wales/Northern Ireland never were allowed to vote yes or no about becoming a “multi racial” society. This was forced on them from the top down by their own government. Prior to WW2, non-Whites numbered less than 10,000 in the entire UK vs 8+ million today.
Could well have been that way.
That is disgusting what their governments have done to them, forcing blacks upon the population. This was all orchestrated by Jews.
OT: This is more a comment related to this December post.
https://www.unz.com/jthompson/dunning-kruger-effect-done-with
I just ran across a recent article “debunking” the Dunning Kruger effect. From a math professor.
https://theconversation.com/debunking-the-dunning-kruger-effect-the-least-skilled-people-know-how-much-they-dont-know-but-everyone-thinks-they-are-better-than-average-195527
The author is coauthor of this related (cited in article) 2016 paper.
https://digitalcommons.usf.edu/numeracy/vol9/iss1/art4/
Since comments don’t appear to be enabled there let’s take a quick look here.
First, the headline leads with the oh so popular “debunked,” but in the text we see (emphasis mine).
Surely asserting something is “debunked” requires a bit more than demonstrating it “may be incorrect”?!
He mentions “three reasons Dunning and Kruger’s analysis is misleading.” I’m having trouble extracting the three reasons (is that just me or did he do that poor a job of exposition?), but here is my take.
Version 1. The three reasons are from the immediately following paragraph.
The first is true, but not sure how that argues against DK. The second is a statement of a possible (likely IMHO) mechanism partially underlying DK. Not sure how noting that cognitive error argues against DK. The third directly contradicts the first (do they or don’t they underestimate their performance?) and is also a red herring in that DK deals more with relative assessment of self-performance than absolute.
Version 2 (seems to map to Version 1 except the first reason is assumed true and not demonstrated)
Their evidence has two aspects. The first is a randomized simulation. In that simulation they assume performance and assessment are uncorrelated!!! That is so dumb as to be comical. The whole point of DK is that people misjudge their actual ability in a systematic fashion. Assuming ability and “assessment” are completely unrelated (random!) makes their simulation a useless attempt at an analogy.
The second is this experiment.
The paper claims this:
Not sure about that as justification for using such a coarse measurement compared to DK’s use of percentiles, but the more important problem lies elsewhere. As I noted above DK deals with relative perceived ability (relative to others) while this paper looks at an objective (those three levels) measure of perceived ability.
If I were a math professor I would be embarrassed by publishing that article under my name. What makes this depressing is his bio from the paper linked above.
P.S. Any idea what is up with these recent efforts to undermine the credibility of the Dunning Kruger Effect?
I think the argument I was “passing on” it that once you control properly for regression to the mean, relatively little is left. I agree with you that the above paper seems to be describing the mechanism of the effect, not refuting it.
However, it is a damn pity I have to abandon this explanation.
I think the paper you cited was better than the article and paper I linked. And I agree with your take on the paper you blogged about.
On reflection I think this part of my earlier comment stands up well.
I would add that the overoptimism curve and the regression to the mean seem inextricably linked. Hard to underestimate at 0 or overestimate at 100.
I wonder how close a model of this form could come to reproducing D-K.
1. Everyone’s ability estimate regresses to the mean by an average of 50% (e.g. 0 becomes 25, 100 becomes 75, on average).
2. Everyone’s ability estimate tends to overoptimism by an average of 15% (thresholded at 100 after the regression adjustment is applied).
How do you think something like that would serve as a model of the process underlying D-K?
Put another way: regression to the mean is not just a statistical artifact. It is a useful encapsulation of a real phenomenon. And the Dunning-Kruger effect just builds on that (e.g. because people are not statistical automatons, and most statistics does not deal with hard limits at 0 and 100).
For an analogy, would you consider a statement like “short parents tend to underpredict the height of their children while tall parents tend to overpredict the height of their children” invalidated by an observation that regression to the mean would cause that? I would not (though it might cause me to reconsider whether it was worth naming, depends on the importance of the other effects involved). My focus from there would be trying to understand the other issues involved. The two which come to mind are the Flynn Effect and any other departures from a strict regression to the mean interpretation (e.g. I would not be surprised to see overoptimism play in here as well).
P.S. Sorry for hijacking this thread a bit. I had wanted to delve more into the earlier paper, but was dealing with some things at the time and never got to it.
This article was republished in Fortune and it showed up in my news feed there. Turns out I missed an ever better example of statistical incompetence (take a look at those credentials again!). Emphasis mine.
https://fortune.com/2023/05/08/what-is-dunning-kruger-effect-smart-intelligence-competence-john-cleese/
What a joke. Though I imagine he would claim he meant median rather than mean if pressed. In case anyone needs an example, consider a test scored 0 to 10 with 9 people scoring 10 and 1 person scoring 0. Average 9, percentage of people scoring above average 90%. So much for “mathematically impossible.”
P.S. I think I see why Eric Gaze does not like the Dunning-Kruger Effect.