Hacker News Reader

Ed Yong at the Atlantic looked into that:

"These differences persisted even after the team accounted for factors like the doctors’ years of experience and the patients’ age, ethnicity, other diseases, educational level, or the hospitals to which they were admitted."

https://www.theatlantic.com/science/archive/2018/08/women-mo...

Years of experience isn't the only way to measure skill. There are software engineers with 20 years experience that aren't as good as one with 2 years experience. I'd like to see the skill levels of the doctors normalized by something else, like whether they can somehow be identified as part of a group that specializes in high risk cases.

Isn't this Ageism? There are tons of 20-year experience engineers who are better than 2-year experience engineers. Yes, there are outliers, as with any normalized population, but typically years of experience in a skilled field correlates with ability to do the job well.

It's just as ageist to assume that a young person doesn't know anything

Not really, since they had less time to learn and experience things.

Many prejudices are justifiable as statistically likely. The problem is that even though you may be right most of the time, you will unfairly disadvantage the people you were wrong about.

How is that ageism? Being experienced simply does not equate to being skilled. It is correlated but some people, irrespective of age or experience, just aren't as good at $task than other people.

The person above wrote that there exists in a subset of engineers that have 20 years of experience, a group of engineers that are not better than engineers in a subset that have 2 years of experience

It was offering perspective, not a blanket statement about the entire set

That isn't ageism.

Again, there are outliers in a normalized population. Assuming that it's an entire subset without evidence is bias, i.e. ageism.

Simply being a perspective doesn't shield it from being biased.

What form of evidence would you like?

I don't see the difference with the exact opposite of the statement (ie. that some engineers with more work experience are in fact better than others with less work experience), aside from the potential novelty. How would you prove that statement, so that someone else could prove the opposite. Are you saying that in outliers shouldn't be considered at all, when they are in fact a subset of a population? If so then you missed the point that someone above was talking about outliers and that was the only observation made.

Alternatively, is your stance that there is no difference to claim something is ageism if it is directed at an older population or a younger population? that noticing that older experienced engineers would also fit the criteria for ageism against a younger population making this whole conversation moot?

I spoke to a doctor about publishing statistics and the major concern with him and his peers was exactly what you mentioned. Riskier cases will make your statistics drop, but the more experienced doctors would almost always get the most difficult cases.

That's not true ... after all, that would mean putting MBA type managers in charge of hospitals (like is being done all across the US) is going to lead to disastrously bad decisions as they will institute individual incentives. Giving those experienced doctors little choice : either refuse cases that might die ... or get fired.

Right, which is why some experienced doctors can be overly conservative. The ones that take up the hard cases are therefore penalized even more.

You see the same thing with lawyers. The ones with a 100% trial success rate are very choosy about which cases they accept or are more inclined to pressure clients to settle.

The phenomenon is real. Whether it exists at the heart of this study or is simply being used to excuse underperforming doctors....

That's sarcasm, right? Because that's what has happened, and been happening while healthcare in the US is treated as a for-profit enterprise.

I wonder what the doctor thinks of publishing risk adjusted statistics. That is, publishing the difference between expected and actual outcome.

I’d also posit that female ER doctors could as a group, simply be better than the average male.

Not inherent to their gender, but because they are a more motivated group to overcome the systemic bias in medicine to get there.

If you get to 50/50 this would mean the stats converge, but the same findings would need to be repeated across other conditions as well.

Then we would've seen the same trend with female doctors treating male patients.

Not necessarily. It could be that only good doctors treat female patients well, while all doctors treat male patients well. For example, if women patients are in some way harder cases while male patients are not, then treatment of women patients may separate good doctors from the rest.

Alternatively, women patients speak differently to male doctors in ways that make it more difficult to make accurate and reliable diagnoses.

Educated professionals are certainly capable of exhibiting bias. But professionals are also more aware of the issue, and are even taught (formally and informally) compensating strategies. People off the street? Not so much....

OR just that female doctors paid more attention to the fact that heart attack symptoms are different for women than for men. This may be as simple as it affecting them. If the male takes longer to recognise heart attack symptoms of females, it could easily produce worse results.

https://www.healthline.com/health/heart-disease/heart-attack...

Maybe we have?

The article says "the other differences were not significant", but that does not mean there was no difference. It means "there was not enough evidence to reject the assumption that there were no other differences". That's one problem with this kind of statistics: absence of evidence is not evidence of absence, but people (including head line writers) act as if it was.

I have noticed the same thing in tech. Only the healthiest orgs I been at have had women clustered around the median, most are at least a std deviation above the median.

Here is a link to far more specific data from the main study in the article. I suggest people look over the data themselves.

http://www.pnas.org/content/pnas/suppl/2018/07/31/1800097115...

The data appears to not be in the friendliest of formats (tables with an axis labeled 'X'? No way my professors would've let that slide).

There's also a mis-labled column in table S2. Which one is Female Doctors Male Patients?

The third column (according to the top rows of the table).

In the table it also specifies which is male and which is female.

Thanks for the link. With that paper in hand, I can say confidently that the Scientific American article is a flat out lie, starting from the headline.

Chart S1 plots the distribution of the predicted probability for survival in four conditions. The distributions for "Male Doc, Female Patient" and "Female Doc, Female Patient" are visually indistinguishable and peak at around 0.96. "Female Doc, Male Patient" peaks at around 0.97, "Male Doc, Male Patient" at 0.98 or so. The difference in the position of the maximum is real, but tiny; the distributions for female patients are much wider (which visually exaggerates the difference). The math seems to agree with my reading, as it says: "when physician gender (male / female) is regressed upon (ŷ), conditional upon controls, there is no significant correlation between ex ante probability of survival and the gender of the physician."

So, dear "Scientific American": It is completely untrue that "Women Die More from Heart Attacks Than Men—Unless the ER Doc Is Female". The data says that "Women Die More from Heart Attacks Than Men—No Matter Who Treats Them" and that "Men Die More from Heart Attacks Than Other Men—When the ER Doc Is Female"

No thanks for turning a study that found tiny differences into a politically charged statement.

But then why would Brad Greenwood, the author of the study, state that "All of those are statistically indistinguishable except for male doctor–female patient", when it's obvious to the naked eye that there are other difference? I guess one difference made it over the arbitrary threshold of significance, and others didn't. That's just the nature of SHIT (Statistical Hypothesis Inference Testing).

I'd also like to see if the female doctors simply follow the checklist better than male doctors.

ER outcomes have been shown over and over to improve when there is a checklist available and followed.

That would only apply if we saw the same trend with female doctors treating male patients.

> The researchers divided 500,000-plus cases into four categories: male doctors treating men; male doctors treating women; female doctors treating men; and female doctors treating women. “All of those are statistically indistinguishable except for male doctor–female patient,” says Brad Greenwood

FYI, just because effect A is statistically distinguishable from zero and effect B is not statistically distinguishable from zero, that does not imply effect A is statistically distinguishable from effect B.

It could still be the case that female doctors are better overall, even if it was only distinguishable from zero in a single branch.

(I looked at paper myself but am not confident I can interpret it correctly.)

The article mentions that women are more likely to present with "atypical" symptoms (which actually may be typical for women). Maybe almost all men present with "typical" symptoms or female doctors are more diligent about checking women for "atypical" symptoms.