4 ms·

> And there's the survivor bias Isn't that precisely the reason why you would want to ask old people, because the survivor bias has eliminated people with habi

> And there's the survivor bias Isn't that precisely the reason why you would want to ask old people, because the survivor bias has eliminated people with habits that were not conducive to a long life?

jraph 2y ago
No, that's exactly the trap that must be avoided: Quoting Wikipedia [1] > Survivorship bias or survival bias is the logical error of concentrating on entities that passed a selection process while overlooking those that did not. For example, you'll hear the advice from the 1% of people who survived drinking every day and who might even think it was good for them, and not the warning from the 99% for whom it didn't work, or worse, who died from it. Unfortunately, you are 99% more likely to be like these 99% than like that 1%. (Numbers totally made up). Intuition gets in the way with this kind of stuff. It's all too easy to be willing to take advice from old people who appear wise to us for all sorts of cultural reasons, but they are most probably not experts in this stuff and they actually most likeky really don't know better. They have a number of data point equal to 1. Our intuition makes it quite hard to understand this, even for people with scientific / statistics background, but all this stuff we know from this background doesn't stop applying when it comes to old people. It doesn't mean it's not enjoyable or interesting to get an old person's view on this, but one should not forget to apply critical thinking. [1] https://en.m.wikipedia.org/wiki/Survivorship_bias https://en.m.wikipedia.org/wiki/Survivorship_bias
- david-gpu 2y ago
  > For example, you'll hear the advice from the 1% of people who survived drinking every day and who might even think it was good for them, and not the warning from the 99% for whom it didn't work, or worse, who died from it. Let's use an example to understand. Let's say that 30% of the population fifty years ago drank and smoked, and 10% did neither. Say that if you drink and smoked, then your chances of living to a hundred years is 1%, but if you do neither then it is 10%. In that scenario, how likely is it that when you ask a centenarian they will be drinkers and smokers vs abstaining from both? How does it compare to the general population? Would you live longer than the general population by following the advice of the centenarians? Does it make sense now?
  - jraph 2y ago
    > how likely is it that when you ask a centenarian they will be drinkers and smokers vs abstaining from both? I don't know but it doesn't matter. How do you know if you are speaking to the right person? And in any case, random old people will have a datapoint of one: themselves. If you need information, take it from good studies, controlled for bias and all that. Any opinion from a single person is worthless as an advice. Even an educated guess isn't good enough. Nobody actually knows why they're having a long life. If I eat vegetables, sleep well and exercise all my life and have a long life, I won't know if it actually helped and if I wouldn't have lived long without those things. Because I didn't try this alternative life. I can only assume it helped from studies we did on large numbers of people. Only data from a big population will help us find correlation, possibly causality (if not confounding factors). >Would you live longer than the general population by following the advice of the centenarians? You used a plural here. That's possible, for a significant number of centenarians, if you control for biases and memory failures. But that's what a scientific study is.
    - david-gpu 2y ago
      If you can't do the simple math problem that was presented, what makes you so confident? Anyway, take care. I hope that the next time somebody takes the effort to help you understand something you will be more willing to learn.
      - jraph 2y ago
        You are assuming I'm wrong, you are right and you need to teach me something and I need to learn. It's incredibly messed up. Humility much? I disagree with you. It's not a question of I need to understand. It's a question of I need to be convinced you are right, which I'm not, I know what you want to say with your math but it has not changed this. You wanted to make me say that in your example, it's unlikely one will meet an old person with bad advice, and then that this advice won't matter too much anyway, but your numbers are made up and even if they were correct, this would still barely change my mind. So, yeah, I took a shortcut because whatever the result, this would change nothing. What's more, if you want to make a point, don't make me compute, especially if you already did the computation. I don't see the point of it. I'm not at school answering to a teacher. Moreover, if your point is that following advice from old people doesn't change things much, then what's the point of arguing for listening to old people's advice and how does it change the fact that it's bad or good advice? Take care too.
- sillyfluke 2y ago
  If you look at the example given for survivorship bias in the wiki closely, you may find your analogy doesn't exactly match. My comparison would be: An 100+ old person says the key to their longetivity was walking a little bit every day. Then he falls on one of his walks and dies. Someone says it's ridiculous encouraging old people to walk because their bones are brittle and one fall could kill them. But all the people that were lifelong couch potatoes are already dead. ie -- wrong conclusion, old guy happened to be closer to the truth in this case.
  - jraph 2y ago
    Survivorship bias may encompass several subtly different cases. I didn't make an analogy, I provided an example. How do you know the advice to walk from the old person is correct? Because we have studies that conclude walking everyday is good for the health and using the bones actually strengthens them (probably with caveats, where specific people in specific situations should actually not do this for some reasons), etc. You don't know from n=1.

jraph 2y ago

No, that's exactly the trap that must be avoided: Quoting Wikipedia [1] > Survivorship bias or survival bias is the logical error of concentrating on entities that passed a selection process while overlooking those that did not. For example, you'll hear the advice from the 1% of people who survived drinking every day and who might even think it was good for them, and not the warning from the 99% for whom it didn't work, or worse, who died from it. Unfortunately, you are 99% more likely to be like these 99% than like that 1%. (Numbers totally made up). Intuition gets in the way with this kind of stuff. It's all too easy to be willing to take advice from old people who appear wise to us for all sorts of cultural reasons, but they are most probably not experts in this stuff and they actually most likeky really don't know better. They have a number of data point equal to 1. Our intuition makes it quite hard to understand this, even for people with scientific / statistics background, but all this stuff we know from this background doesn't stop applying when it comes to old people. It doesn't mean it's not enjoyable or interesting to get an old person's view on this, but one should not forget to apply critical thinking. [1] https://en.m.wikipedia.org/wiki/Survivorship_bias https://en.m.wikipedia.org/wiki/Survivorship_bias

david-gpu 2y ago
> For example, you'll hear the advice from the 1% of people who survived drinking every day and who might even think it was good for them, and not the warning from the 99% for whom it didn't work, or worse, who died from it. Let's use an example to understand. Let's say that 30% of the population fifty years ago drank and smoked, and 10% did neither. Say that if you drink and smoked, then your chances of living to a hundred years is 1%, but if you do neither then it is 10%. In that scenario, how likely is it that when you ask a centenarian they will be drinkers and smokers vs abstaining from both? How does it compare to the general population? Would you live longer than the general population by following the advice of the centenarians? Does it make sense now?
- jraph 2y ago
  > how likely is it that when you ask a centenarian they will be drinkers and smokers vs abstaining from both? I don't know but it doesn't matter. How do you know if you are speaking to the right person? And in any case, random old people will have a datapoint of one: themselves. If you need information, take it from good studies, controlled for bias and all that. Any opinion from a single person is worthless as an advice. Even an educated guess isn't good enough. Nobody actually knows why they're having a long life. If I eat vegetables, sleep well and exercise all my life and have a long life, I won't know if it actually helped and if I wouldn't have lived long without those things. Because I didn't try this alternative life. I can only assume it helped from studies we did on large numbers of people. Only data from a big population will help us find correlation, possibly causality (if not confounding factors). >Would you live longer than the general population by following the advice of the centenarians? You used a plural here. That's possible, for a significant number of centenarians, if you control for biases and memory failures. But that's what a scientific study is.
  - david-gpu 2y ago
    If you can't do the simple math problem that was presented, what makes you so confident? Anyway, take care. I hope that the next time somebody takes the effort to help you understand something you will be more willing to learn.
    - jraph 2y ago
      You are assuming I'm wrong, you are right and you need to teach me something and I need to learn. It's incredibly messed up. Humility much? I disagree with you. It's not a question of I need to understand. It's a question of I need to be convinced you are right, which I'm not, I know what you want to say with your math but it has not changed this. You wanted to make me say that in your example, it's unlikely one will meet an old person with bad advice, and then that this advice won't matter too much anyway, but your numbers are made up and even if they were correct, this would still barely change my mind. So, yeah, I took a shortcut because whatever the result, this would change nothing. What's more, if you want to make a point, don't make me compute, especially if you already did the computation. I don't see the point of it. I'm not at school answering to a teacher. Moreover, if your point is that following advice from old people doesn't change things much, then what's the point of arguing for listening to old people's advice and how does it change the fact that it's bad or good advice? Take care too.
sillyfluke 2y ago
If you look at the example given for survivorship bias in the wiki closely, you may find your analogy doesn't exactly match. My comparison would be: An 100+ old person says the key to their longetivity was walking a little bit every day. Then he falls on one of his walks and dies. Someone says it's ridiculous encouraging old people to walk because their bones are brittle and one fall could kill them. But all the people that were lifelong couch potatoes are already dead. ie -- wrong conclusion, old guy happened to be closer to the truth in this case.
- jraph 2y ago
  Survivorship bias may encompass several subtly different cases. I didn't make an analogy, I provided an example. How do you know the advice to walk from the old person is correct? Because we have studies that conclude walking everyday is good for the health and using the bones actually strengthens them (probably with caveats, where specific people in specific situations should actually not do this for some reasons), etc. You don't know from n=1.