4 ms·

Ah my good friend statistics. It's all about the probabilities, and ratios, and comparisons. But being somewhat light on the details that matter. So sure, inci

Ah my good friend statistics. It's all about the probabilities, and ratios, and comparisons. But being somewhat light on the details that matter. So sure, incidences of cancer are going up, and while I briefly read the article I didn't deep dive into the data. I'll presume they corrected for population growth, the decline in other mortality rates, better testing, more testing, increasing life expectancy etc. Alas youth it's fixated on rate of change, but without (ideally upfront) discussing the are rate. Consider this example; (made up numbers) "Incidences of cancer generally are up 400%". Compared to "incidences of car deaths are down 80%". From those 2 statements alone should you be more worried about cancer or cars? And lots of articles and news report this - rate of change. But rate of change is meaningless without a base number. Consider this (still made up numbers) "Cancer went from killing 1 person per 100 to 4 per hundred" " road deaths went from 95 people per hundred to 19 per hundred". Are you now worried more about cars or cancers? Side note : causes of death is a zero sum game. If one cause decreases the others have to increase. Side note: the most important measure of course is the age if death. If I'm 90 I'm worried about cancer. If I'm 9 cars are a more immediate issue. So yeah, eating more broccoli might reduce my risk of pancreas cancer. But saying it as 'drops the risk 45%' doesn't really help me if the base risk is say 1 in a million. Seat belts though - highly recommended.

Eisenstein 2y ago
Cause of death stats are difficult to interpret because people have to die of 'something'. Is cause of death 'drinking and driving' or was it 'drove into a pole at 60mph' or was it 'blood loss' or was it 'medical term for heart stopped beating'. If you wanted to play with stats you could claim any one of these things.
- strogonoff 2y ago
  Cardiac arrest ← blood loss ← impact with a pole at 60 mph ← drunk driving ← [insert the actual root cause] Sadly, that extra step up the causality ladder tends to meet furious resistance from most natural-scientifically-oriented people. It could be as simple as “lack of will to live due to losing a partner in old age”, for instance—one of a variety of conditions with absent motivation to do things that correlate with not dying, or indeed to not do things that inversely correlate with such, therefore trivially leading to death via whatever first available mechanism (drinking to DUI to accident, or smoking to inflammation to cancer, or lack of sleep to chronic inflammation to cancer, and so on). The particular mechanism itself is not the most interesting part, and if the principle is right then eliminating each pathway is going to be a game of whack-a-mole draining healthcare resources while failing to accomplish the presumed goal (happy continued existence of a person).
  - red-iron-pine 2y ago
    but-for analysis suggests that the drunk driving is the root cause. "but why where they drinking and driving!?" could range from carelessness to chronic mental health issues and addition. too many to pin down, and not direct causes; indirect at best. drunk driving killed them
    - strogonoff 2y ago
      Not really. “But why” can be asked at any point in the chain (“but why did the blood loss occur?”), so it’s not a useful differentiator. The problem is that there’re seemingly no ready clear-cut one-size-fits-all solutions with root causes, while there are some we can use to alleviate the downstream mechanisms of harm (e.g., drunk driving bans, chemotherapy), so we prefer to address those instead. When all you have is a hammer and all that. It doesn’t show that such root causes can’t be helped, only that the topic is perhaps not being investigated as actively (it doesn’t lend itself to the traditional approaches used in natural sciences, and rather than treat it as a difficult challenge it’s easier to label it as too vague, not scientific enough, or beyond an arbitrary physicality line).
klik99 2y ago
But how many people who died in car accidents also had terminal cancer? Seriously though, the trend in America is real, if you're concerned with the impact of pop growth, life expectancy, etc, then see how it compares in US vs countries with the same trends: https://ourworldindata.org/grapher/cancer-incidence?tab=chart&country=OWID_WRL~African+Region+%28WHO%29~European+Region+%28WHO%29~WHO_SEAR~WHO_WPAC~WHO_EMR~USA~KOR~FRA~DEU~BRA~CAN~GBR https://ourworldindata.org/grapher/cancer-incidence?tab=char... Trends matter, and this trend is pretty dramatic

Eisenstein 2y ago

Cause of death stats are difficult to interpret because people have to die of 'something'. Is cause of death 'drinking and driving' or was it 'drove into a pole at 60mph' or was it 'blood loss' or was it 'medical term for heart stopped beating'. If you wanted to play with stats you could claim any one of these things.

strogonoff 2y ago
Cardiac arrest ← blood loss ← impact with a pole at 60 mph ← drunk driving ← [insert the actual root cause] Sadly, that extra step up the causality ladder tends to meet furious resistance from most natural-scientifically-oriented people. It could be as simple as “lack of will to live due to losing a partner in old age”, for instance—one of a variety of conditions with absent motivation to do things that correlate with not dying, or indeed to not do things that inversely correlate with such, therefore trivially leading to death via whatever first available mechanism (drinking to DUI to accident, or smoking to inflammation to cancer, or lack of sleep to chronic inflammation to cancer, and so on). The particular mechanism itself is not the most interesting part, and if the principle is right then eliminating each pathway is going to be a game of whack-a-mole draining healthcare resources while failing to accomplish the presumed goal (happy continued existence of a person).
- red-iron-pine 2y ago
  but-for analysis suggests that the drunk driving is the root cause. "but why where they drinking and driving!?" could range from carelessness to chronic mental health issues and addition. too many to pin down, and not direct causes; indirect at best. drunk driving killed them
  - strogonoff 2y ago
    Not really. “But why” can be asked at any point in the chain (“but why did the blood loss occur?”), so it’s not a useful differentiator. The problem is that there’re seemingly no ready clear-cut one-size-fits-all solutions with root causes, while there are some we can use to alleviate the downstream mechanisms of harm (e.g., drunk driving bans, chemotherapy), so we prefer to address those instead. When all you have is a hammer and all that. It doesn’t show that such root causes can’t be helped, only that the topic is perhaps not being investigated as actively (it doesn’t lend itself to the traditional approaches used in natural sciences, and rather than treat it as a difficult challenge it’s easier to label it as too vague, not scientific enough, or beyond an arbitrary physicality line).

klik99 2y ago

But how many people who died in car accidents also had terminal cancer? Seriously though, the trend in America is real, if you're concerned with the impact of pop growth, life expectancy, etc, then see how it compares in US vs countries with the same trends: https://ourworldindata.org/grapher/cancer-incidence?tab=chart&country=OWID_WRL~African+Region+%28WHO%29~European+Region+%28WHO%29~WHO_SEAR~WHO_WPAC~WHO_EMR~USA~KOR~FRA~DEU~BRA~CAN~GBR https://ourworldindata.org/grapher/cancer-incidence?tab=char... Trends matter, and this trend is pretty dramatic