Re:Stanford president resigns over manipulated research, wil...
by godelski on Wednesday July 19, 2023 @09:42PM
(#36795933)
> if both your parents are journalists or [insert whatever profession], it’s as if you’re likely to pick it up no matter how hard you try not to…
How likely is it for you to be {Catholic,Muslim,Hindu,etc} if your parents are? What about the food you like? Interests? Etc
The answer is high, but not 100%. I say this as someone who comes from a _very_ religious family (I'm more areligious), the only person who does anything even remotely STEM related, and has a very different diet than my parents. So triple whammy (but probably unsurprising to get both given one).
It's just likely that you are going to be more similar to your parents than not. Not sure why we should expect otherwise. As an example, growing up in my house there was a lot of discussion of politics and business. It is probably no surprised that almost everyone works in sales. But friends who have parents who are professors grew up listening to and discussing academic ideas. Essentially what is happening is that at an early age we're all being trained and this training is heavily influenced by the teacher models, I mean parents. Sorry, the ML is creeping in ;). I would not be surprised that someone who grew up in a house where both parents are journalists that they wouldn't be explicitly trained to scrutinize authority, gather evidence, and write that down. Lots of learning is done by imitation.
Is this a bad thing? Probably not. Does it mean you don't have free will? Definitely not, but it does mean heavy influence. I think an important aspect is to recognize this, especially if we are going to have conversations about meritocracy and fairness. Understanding this early learning does give credit to claims of quality through generational experience. The question is how much this matters by the time we reach adults and finish our education (wherever that is). I think the longer the education, the less this matters. But it definitely means there's a bias in prestige of that education (and quality if you believe that prestige strongly correlates with quality). It probably isn't a surprise that kids from academics tend to end up at higher ranked universities than kids from non-academics (we see a strong effect when looking at race or class, but these are most likely not causal variables but confounders). It also means that any "meritocratic" system (personal belief is that these are fools errands) will be substantially benefited by generational experience as children will at a young age learn how to optimize for the metrics (which are not necessarily aligned to the goals of those metrics though, which is why the previous note).
Momentum is one hell of a force. Compounding interest isn't just important for money, but nearly any form of resource. There are a lot of benefits to momentum, but the problem is that we often don't account for this in our models and this makes for poor evaluation. I would be very willing to bet that a lot of things people make evaluations on are far more random than they believe. For example I'd put money on that were everything else held equal (making this hard to measure) that high school grades would not significantly affect student outcomes. We can use the studies on students who were admitted through affirmative action to get an indication of this, where we generally see similar results to students admitted normally. There is some noise, but it also isn't unsurprising that if you are doing poorly in high school that this might correlate with problems at home, which can cause a higher likelihood of dropout. But generally, this would suggest that you could shuffle all applications, randomly select, and you wouldn't see a large difference in your students' performance nor outcomes post university (my hypothesis). Some of this is already done, given that many top universities have more applicants that are nearly impossible to distinguish than they have seats for, making selection at least nearly arbitrary for these applicants. But where this may matter in the real world is that prestige definitely correlates with opportunities like internships and work, as companies prioritize recruitment from these locations and prioritize building connections with these professors (who connect students with opportunities). These are probably where we could make adjustments to make a more fair system, but is likely harder than suggested.
TLDR: heavy influence, but you still got free will. Should make you think though...
How likely is it for you to be {Catholic,Muslim,Hindu,etc} if your parents are? What about the food you like? Interests? Etc
The answer is high, but not 100%. I say this as someone who comes from a _very_ religious family (I'm more areligious), the only person who does anything even remotely STEM related, and has a very different diet than my parents. So triple whammy (but probably unsurprising to get both given one).
It's just likely that you are going to be more similar to your parents than not. Not sure why we should expect otherwise. As an example, growing up in my house there was a lot of discussion of politics and business. It is probably no surprised that almost everyone works in sales. But friends who have parents who are professors grew up listening to and discussing academic ideas. Essentially what is happening is that at an early age we're all being trained and this training is heavily influenced by the teacher models, I mean parents. Sorry, the ML is creeping in ;). I would not be surprised that someone who grew up in a house where both parents are journalists that they wouldn't be explicitly trained to scrutinize authority, gather evidence, and write that down. Lots of learning is done by imitation.
Is this a bad thing? Probably not. Does it mean you don't have free will? Definitely not, but it does mean heavy influence. I think an important aspect is to recognize this, especially if we are going to have conversations about meritocracy and fairness. Understanding this early learning does give credit to claims of quality through generational experience. The question is how much this matters by the time we reach adults and finish our education (wherever that is). I think the longer the education, the less this matters. But it definitely means there's a bias in prestige of that education (and quality if you believe that prestige strongly correlates with quality). It probably isn't a surprise that kids from academics tend to end up at higher ranked universities than kids from non-academics (we see a strong effect when looking at race or class, but these are most likely not causal variables but confounders). It also means that any "meritocratic" system (personal belief is that these are fools errands) will be substantially benefited by generational experience as children will at a young age learn how to optimize for the metrics (which are not necessarily aligned to the goals of those metrics though, which is why the previous note).
Momentum is one hell of a force. Compounding interest isn't just important for money, but nearly any form of resource. There are a lot of benefits to momentum, but the problem is that we often don't account for this in our models and this makes for poor evaluation. I would be very willing to bet that a lot of things people make evaluations on are far more random than they believe. For example I'd put money on that were everything else held equal (making this hard to measure) that high school grades would not significantly affect student outcomes. We can use the studies on students who were admitted through affirmative action to get an indication of this, where we generally see similar results to students admitted normally. There is some noise, but it also isn't unsurprising that if you are doing poorly in high school that this might correlate with problems at home, which can cause a higher likelihood of dropout. But generally, this would suggest that you could shuffle all applications, randomly select, and you wouldn't see a large difference in your students' performance nor outcomes post university (my hypothesis). Some of this is already done, given that many top universities have more applicants that are nearly impossible to distinguish than they have seats for, making selection at least nearly arbitrary for these applicants. But where this may matter in the real world is that prestige definitely correlates with opportunities like internships and work, as companies prioritize recruitment from these locations and prioritize building connections with these professors (who connect students with opportunities). These are probably where we could make adjustments to make a more fair system, but is likely harder than suggested.
TLDR: heavy influence, but you still got free will. Should make you think though...