I agree with this, and strongly disagree with the decision by Aaron to moderate this comment (as well as with other people deciding to downvote this). This strikes me as a totally reasonable and well-argued comment.
I’ve thought about this topic a good amount, and if a friend of mine were to tell me they are planning to adopt a child, I would immediately drop everything I am working on and spend at least 10 hours trying to talk them out of it, and generally be willing to spend a lot of my social capital to preven them from making this choice.
For some calibration, risk of drug abuse, which is a reasonable baseline for other types of violent behavior as well, is about 2-3x in adopted children. This is not conditioning on it being a teenager adoption, which I expect would likely increase the ratio to more something like 3-4x, given the additional negative selection effects.
Sibling abuse rates are something like 20% (or 80% depending on your definition). And is the most frequent form of household abuse. This means by adopting a child you are adding something like an additional 60% chance of your other child going through at least some level of abuse (and I would estimate something like a 15% chance of serious abuse). This is a lot.
Like, this feels to me like a likely life-destroying mistake, with very predictably bad outcomes. Given that a large fraction of household abuse is sibling abuse, this is making it more likely than not that your other children will deal with substantial abuse in their childhood. This is not a “small probabilities of large downside” situation. These are large probabilities of large downsides.
So despite the fact that I spent quite a while thinking about adopting vs. having biological children a few years ago and came out in favour of having biological children for now based on similar concerns you (and Dale) raise about more adverse outcomes in adopted children, I find your conclusion to strongly dissuade from adopting very surprising.
You thinking that adopting might likely be a life-destroying mistake does not seem to line up with the adoption satisfaction data Aaron linked. Maybe you meant this specifically for adopting teenagers? It was not clear from your comment.
In many ways, I prefer more awareness about difficulties in adopting over the naive ‘why don’t you just adopt?!’ do-gooders who want to have children often hear. So I am grateful that this topic is brought up, I would just prefer a more clear pointing out of trade-offs, as well as more emotional sensitivity on the topic.
When I looked into this, I looked more at qualitative accounts (and also tried to answer a slightly different question—does demand from potential parents for low-risk adopted children outstrip how many such children there are?) and less at quantitative data. While this seems like a clear oversight in retrospect, apparently this led me to a more negative impression than is warranted now looking at the data you linked. If you had told me that 1,5% of biological children have substantial drug abuse problems as defined in the paper and asked me to guess the percentage for adopted children, I would have guessed way more than 3,5%. I was also surprised by the adoption satisfaction data Aaron linked. So if anything, you are leaving me with a more positive impression of adoption, and are motivating me to look into the topic again.
Thus I am surprised you are so confident in your position you would be willing to spend so much time on dissuading people to adopt. (You are welcome to try to dissuade me that this is even worth my time looking into it!) To me, the outcomes do not seem to be ‘very predictably bad’.
While on average adopted children have worse outcomes than biological children, this really does not need to be true for each individual making this choice. It is also not the only factor which matters. To name some examples which can tilt the decision: infertility, same-sex couples, previous difficult pregnancies or birth trauma, family history of genetic diseases like Huntington’s, more garden-variety heritable risk factors for issues like ADHD, autism and depression, how high risk the potential adoptive children actually are, e.g. based on their age and previous history, etc.
I am unsure how to think about satisfaction data. My general model is that lots of satisfaction data is biased upwards and I can’t really imagine a negative result from that survey, so I really don’t know how much to update on it. I would currently just ignore it, unless someone had a really clever study design where they have some kind of other intervention that is similarly costly and had similar social expectations, but we know is bad for people, that we could use as a control.
And yes, I think concerns like infertility, same-sex couples, and many other things like that can make adoption the best choice for people who really want to have children. But I do think the costs would still be there, you might just not have an alternative.
I also think one can reduce the costs here by a lot by trying to find one of the best kids to adopt, or doing weirder things like trying to find a surrogate mother, which will probably have much less adverse selection effects (though I haven’t thought through this case very much). My concern is much more about the naive way most people seem to handle adoptions, and I think there are ways to reduce the risk to a level where the tradeoffs become much less harsh.
(Arguably nitpicking, in the sense that I suspect this would not change the bottom line, posted because the use of stats here raised my eyebrows)
For some calibration, risk of drug abuse, which is a reasonable baseline for other types of violent behavior as well, is about 2-3x in adopted children. This is not conditioning on it being a teenager adoption, which I expect would likely increase the ratio to more something like 3-4x, given the additional negative selection effects.
Sibling abuse rates are something like 20% (or 80% depending on your definition). And is the most frequent form of household abuse. This means by adopting a child you are adding something like an additional 60% chance of your other child going through at least some level of abuse
For the benefit of those who didn’t click through the link, the rate on their chosen measure is very roughly 3.5% for adoptees versus roughly 1.5% for the general population, which I assume is where the 2-3x came from. I also buy that by adopting a teenager this number is going to be pushed up towards the foster child outcomes (~8%); a guess like 5% (“3-4x”) seems reasonable.
But you can’t directly extrapolate from the ratio on a rare outcome to a typical outcome, e.g. a 20% → 67% (67 = 20 * 5 / 1.5) change in the absolute likelihood of sibling abuse, which I think is basically what you are doing here, though do correct me if I’m wrong since there were some numbers you gave I couldn’t follow. The statistical intuition going into that is rough, but here’s a concrete, if technical, example:
A 1.5% bad tail outcome in a normal distribution means you are 2.17 standard deviations below the mean, a 5% tail outcome means you are 1.64 SDs below the mean, and so you would go 1.5% → 5% just by dropping the mean by 0.53 SDs. But this would only move a 20% likelihood outcome to 38%, well short of 67% or even your 60%. To get a 20% outcome to 60% you need a 1.1 SD move, which would be equivalent to a 1.5% outcome becoming 14%. The choice of normal distribution in the above is arbitrary, but I expect the pattern to hold among reasonable choices for this case.
In less technical language: you don’t have to move a distribution very much to change the probability of tail outcomes by a lot, whereas almost by definition you do have to move a distribution a lot to change the probability of typical outcomes by a lot.
Thanks for this explanation. That part of Habryka’s comment also struck me as very suspicious when I read it, but it wasn’t immediately obvious what’s wrong with it exactly.
Yeah, I think this is a totally fair critique and I updated some after reading it!
I wrote the above after a long Slack conversation with Aaron at like 2AM, just trying to capture the rough shape of the argument without spending too much time on it.
I do think actually chasing this argument all the way through is interesting and possibly worth it. I think it’s pretty plausible it could make a 2-3x difference in the final outcome (and possibly a lot more!), and I hadn’t actually thought through it all the way. And while I had some gut sense it was important to differentiate between median and tail outcomes here, I hadn’t properly thought through the exact relationship between the two and am appreciative of you doing some more of the thinking.
I currently prefer your estimate of “moving it from 20% to 38%” as something like my best guess.
So, one thing I was thinking about was that people frequently use the murder-rate as a proxy for the overall crime rate, and I think I remember people doing that without any adjustment of the type you are thinking about here. Is there something special about the murder rate as a fraction of violent crimes, or should we actually make the same adjustments in that case?
I think similar adjustments should be made if you are extrapolating to crimes with very different prevalence. For example, the US murder rate is 4-5x that of the UK, but I wouldn’t expect the US to have that many more bike thefts.
Proxy seems fine if you’re focused on which country/city/etc. has higher overall crime, rather than estimating magnitude.
(FWIW, attempt at Googling the above suggest ~300k bike thefts per year in UK versus 2m in US, US population 5x bigger so that’s only 1.33x the UK rate. A quick check on bicycle sales in the two countries does not suggest that this is because of very different cycling rates. No links because on phone, but above is very rough anyway. I’m left with somewhat greater confidence that the gap is in fact <<4x, like 1.2x − 2x, though.)
Similar comments could be made about extrapolating from the large number of US billionaires (way more per capita than any other country IIRC) to the relative rates of people earning more than $200k/$50k/etc. That case might be more intuitive.
A less important motivation/mechanism is probabilities/ratios (instead of odds) are bounded above by one. For rare events ‘doubling the probability’ versus ‘doubling the odds’ get basically the same answer, but not so for more common events. Loosely, flipping a coin three times ‘trebles’ my risk of observing it landing tails, but the probability isn’t 1.5. (cf).
E.g.
Sibling abuse rates are something like 20% (or 80% depending on your definition). And is the most frequent form of household abuse. This means by adopting a child you are adding something like an additional 60% chance of your other child going through at least some level of abuse (and I would estimate something like a 15% chance of serious abuse). [my emphasis]
If you used the 80% definition instead of 20%, then the ‘4x’ risk factor implied by 60% additional chance (with 20% base rate) would give instead an additional 240% chance.
[(Of interest, 20% to 38% absolute likelihood would correspond to an odds ratio of ~2.5, in the ballpark of 3-4x risk factors discussed before. So maybe extrapolating extreme event ratios to less-extreme event ratios can do okay if you keep them in odds form. The underlying story might have something to do with logistic distributions closely resemble normal distributions (save at the tails), so thinking about shifting a normal distribution across the x axis so (non-linearly) more or less of it lies over a threshold loosely resembles adding increments to log-odds (equivalent to multiplying odds by a constant multiple) giving (non-linear) changes when traversing a logistic CDF.
But it still breaks down when extrapolating very large ORs from very rare events. Perhaps the underlying story here may have something to do with higher kurtosis : ‘>2SD events’ are only (I think) ~5X more likely than >3SD events for logistic distributions, versus ~20X in normal distribution land. So large shifts in likelihood of rare(r) events would imply large logistic-land shifts (which dramatically change the whole distribution, e.g. an OR of 10 makes evens --> >90%) much more modest in normal-land (e.g. moving up an SD gives OR>10 for previously 3SD events, but ~2 for previously ‘above average’ ones)]
Yep, I should have definitely kept the probabilities in log-form, just to be less confusing. It wouldn’t have made a huge difference to the outcome, but it seems better practice than the thing that I did.
Strongly upvoted. This comment provides solid evidence in support of its argument, and led me to substantially raise my estimate of the physical risks involved in adoption. (Edit: This holds even in light of AGB’s reply; as he points out, the numbers on risk are still quite high even if you make an adjustment of the type he recommends.)
I also appreciate Habryka’s willingness to speak out in favor of a comment that was heavily downvoted and moderated, just as I appreciated Dale’s good intentions in his initial comment.
I found Habryka’s supporting evidence to be more relevant than Dale’s, and his argument clearer — such that it meets the standards of rigor I was hoping for given the topic at hand.
*****
My biggest update here is on the rate of sibling abuse, which I hadn’t realized was nearly as high as it seems to be. From page 25 of this report:
The report’s definition of “severe assault” (ignore the bit about parents, it seems like sibling violence was measured in the same way):
I’ll note that this still leaves room for interpretation. Both of my siblings hit me with objects when I was growing up, leaving marks or bruises in some cases, but I’m still glad they were born, and we get along well as adults. (We also got along well as children, most of the time.)
I’d be really interested to see data like this broken down by the type/severity of violence. I’d guess that a substantial portion of that “37 percent” number comes from things most people would perceive as normal (e.g. a 7-year-old punches her 8-year-old brother in the arm, a 10-year-old hits his twin brother with a Wiffle bat). But I could be wrong, and even much smaller numbers for “truly severe” incidents could mean a lot more risk than I’d expected.
*****
Despite this update, “never adopt” still feels extreme to me, given the chance of a very positive outcome (a child has a loving family who helps to support them, rather than no family) and the high satisfaction rates of adoptive parents (keeping in mind that those numbers are likely to be skewed by a bias towards reporting satisfaction).
Some factors which seem like they could, in tandem, sharply reduce the risk Habryka identifies:
Adopting a child younger than your current children
Adopting a girl rather than a boy (the above report found that boys were more likely than girls to commit violent acts, but I don’t know how big the difference is)
Doing everything you can to learn whether a child has a history of violent behavior (if a 15-year-old has a totally clean record, that seems like useful evidence)
But I’m not confident about the impact of any of these.
I agree with this, and strongly disagree with the decision by Aaron to moderate this comment (as well as with other people deciding to downvote this). This strikes me as a totally reasonable and well-argued comment.
I’ve thought about this topic a good amount, and if a friend of mine were to tell me they are planning to adopt a child, I would immediately drop everything I am working on and spend at least 10 hours trying to talk them out of it, and generally be willing to spend a lot of my social capital to preven them from making this choice.
For some calibration, risk of drug abuse, which is a reasonable baseline for other types of violent behavior as well, is about 2-3x in adopted children. This is not conditioning on it being a teenager adoption, which I expect would likely increase the ratio to more something like 3-4x, given the additional negative selection effects.
Sibling abuse rates are something like 20% (or 80% depending on your definition). And is the most frequent form of household abuse. This means by adopting a child you are adding something like an additional 60% chance of your other child going through at least some level of abuse (and I would estimate something like a 15% chance of serious abuse). This is a lot.
Like, this feels to me like a likely life-destroying mistake, with very predictably bad outcomes. Given that a large fraction of household abuse is sibling abuse, this is making it more likely than not that your other children will deal with substantial abuse in their childhood. This is not a “small probabilities of large downside” situation. These are large probabilities of large downsides.
So despite the fact that I spent quite a while thinking about adopting vs. having biological children a few years ago and came out in favour of having biological children for now based on similar concerns you (and Dale) raise about more adverse outcomes in adopted children, I find your conclusion to strongly dissuade from adopting very surprising.
You thinking that adopting might likely be a life-destroying mistake does not seem to line up with the adoption satisfaction data Aaron linked. Maybe you meant this specifically for adopting teenagers? It was not clear from your comment.
In many ways, I prefer more awareness about difficulties in adopting over the naive ‘why don’t you just adopt?!’ do-gooders who want to have children often hear. So I am grateful that this topic is brought up, I would just prefer a more clear pointing out of trade-offs, as well as more emotional sensitivity on the topic.
When I looked into this, I looked more at qualitative accounts (and also tried to answer a slightly different question—does demand from potential parents for low-risk adopted children outstrip how many such children there are?) and less at quantitative data. While this seems like a clear oversight in retrospect, apparently this led me to a more negative impression than is warranted now looking at the data you linked. If you had told me that 1,5% of biological children have substantial drug abuse problems as defined in the paper and asked me to guess the percentage for adopted children, I would have guessed way more than 3,5%. I was also surprised by the adoption satisfaction data Aaron linked. So if anything, you are leaving me with a more positive impression of adoption, and are motivating me to look into the topic again.
Thus I am surprised you are so confident in your position you would be willing to spend so much time on dissuading people to adopt. (You are welcome to try to dissuade me that this is even worth my time looking into it!) To me, the outcomes do not seem to be ‘very predictably bad’.
While on average adopted children have worse outcomes than biological children, this really does not need to be true for each individual making this choice. It is also not the only factor which matters. To name some examples which can tilt the decision: infertility, same-sex couples, previous difficult pregnancies or birth trauma, family history of genetic diseases like Huntington’s, more garden-variety heritable risk factors for issues like ADHD, autism and depression, how high risk the potential adoptive children actually are, e.g. based on their age and previous history, etc.
I am unsure how to think about satisfaction data. My general model is that lots of satisfaction data is biased upwards and I can’t really imagine a negative result from that survey, so I really don’t know how much to update on it. I would currently just ignore it, unless someone had a really clever study design where they have some kind of other intervention that is similarly costly and had similar social expectations, but we know is bad for people, that we could use as a control.
And yes, I think concerns like infertility, same-sex couples, and many other things like that can make adoption the best choice for people who really want to have children. But I do think the costs would still be there, you might just not have an alternative.
I also think one can reduce the costs here by a lot by trying to find one of the best kids to adopt, or doing weirder things like trying to find a surrogate mother, which will probably have much less adverse selection effects (though I haven’t thought through this case very much). My concern is much more about the naive way most people seem to handle adoptions, and I think there are ways to reduce the risk to a level where the tradeoffs become much less harsh.
(Arguably nitpicking, in the sense that I suspect this would not change the bottom line, posted because the use of stats here raised my eyebrows)
For the benefit of those who didn’t click through the link, the rate on their chosen measure is very roughly 3.5% for adoptees versus roughly 1.5% for the general population, which I assume is where the 2-3x came from. I also buy that by adopting a teenager this number is going to be pushed up towards the foster child outcomes (~8%); a guess like 5% (“3-4x”) seems reasonable.
But you can’t directly extrapolate from the ratio on a rare outcome to a typical outcome, e.g. a 20% → 67% (67 = 20 * 5 / 1.5) change in the absolute likelihood of sibling abuse, which I think is basically what you are doing here, though do correct me if I’m wrong since there were some numbers you gave I couldn’t follow. The statistical intuition going into that is rough, but here’s a concrete, if technical, example:
A 1.5% bad tail outcome in a normal distribution means you are 2.17 standard deviations below the mean, a 5% tail outcome means you are 1.64 SDs below the mean, and so you would go 1.5% → 5% just by dropping the mean by 0.53 SDs. But this would only move a 20% likelihood outcome to 38%, well short of 67% or even your 60%. To get a 20% outcome to 60% you need a 1.1 SD move, which would be equivalent to a 1.5% outcome becoming 14%. The choice of normal distribution in the above is arbitrary, but I expect the pattern to hold among reasonable choices for this case.
In less technical language: you don’t have to move a distribution very much to change the probability of tail outcomes by a lot, whereas almost by definition you do have to move a distribution a lot to change the probability of typical outcomes by a lot.
Thanks for this explanation. That part of Habryka’s comment also struck me as very suspicious when I read it, but it wasn’t immediately obvious what’s wrong with it exactly.
Yeah, I think this is a totally fair critique and I updated some after reading it!
I wrote the above after a long Slack conversation with Aaron at like 2AM, just trying to capture the rough shape of the argument without spending too much time on it.
I do think actually chasing this argument all the way through is interesting and possibly worth it. I think it’s pretty plausible it could make a 2-3x difference in the final outcome (and possibly a lot more!), and I hadn’t actually thought through it all the way. And while I had some gut sense it was important to differentiate between median and tail outcomes here, I hadn’t properly thought through the exact relationship between the two and am appreciative of you doing some more of the thinking.
I currently prefer your estimate of “moving it from 20% to 38%” as something like my best guess.
So, one thing I was thinking about was that people frequently use the murder-rate as a proxy for the overall crime rate, and I think I remember people doing that without any adjustment of the type you are thinking about here. Is there something special about the murder rate as a fraction of violent crimes, or should we actually make the same adjustments in that case?
I think similar adjustments should be made if you are extrapolating to crimes with very different prevalence. For example, the US murder rate is 4-5x that of the UK, but I wouldn’t expect the US to have that many more bike thefts.
Proxy seems fine if you’re focused on which country/city/etc. has higher overall crime, rather than estimating magnitude.
(FWIW, attempt at Googling the above suggest ~300k bike thefts per year in UK versus 2m in US, US population 5x bigger so that’s only 1.33x the UK rate. A quick check on bicycle sales in the two countries does not suggest that this is because of very different cycling rates. No links because on phone, but above is very rough anyway. I’m left with somewhat greater confidence that the gap is in fact <<4x, like 1.2x − 2x, though.)
Similar comments could be made about extrapolating from the large number of US billionaires (way more per capita than any other country IIRC) to the relative rates of people earning more than $200k/$50k/etc. That case might be more intuitive.
A less important motivation/mechanism is probabilities/ratios (instead of odds) are bounded above by one. For rare events ‘doubling the probability’ versus ‘doubling the odds’ get basically the same answer, but not so for more common events. Loosely, flipping a coin three times ‘trebles’ my risk of observing it landing tails, but the probability isn’t 1.5. (cf).
E.g.
If you used the 80% definition instead of 20%, then the ‘4x’ risk factor implied by 60% additional chance (with 20% base rate) would give instead an additional 240% chance.
[(Of interest, 20% to 38% absolute likelihood would correspond to an odds ratio of ~2.5, in the ballpark of 3-4x risk factors discussed before. So maybe extrapolating extreme event ratios to less-extreme event ratios can do okay if you keep them in odds form. The underlying story might have something to do with logistic distributions closely resemble normal distributions (save at the tails), so thinking about shifting a normal distribution across the x axis so (non-linearly) more or less of it lies over a threshold loosely resembles adding increments to log-odds (equivalent to multiplying odds by a constant multiple) giving (non-linear) changes when traversing a logistic CDF.
But it still breaks down when extrapolating very large ORs from very rare events. Perhaps the underlying story here may have something to do with higher kurtosis : ‘>2SD events’ are only (I think) ~5X more likely than >3SD events for logistic distributions, versus ~20X in normal distribution land. So large shifts in likelihood of rare(r) events would imply large logistic-land shifts (which dramatically change the whole distribution, e.g. an OR of 10 makes evens --> >90%) much more modest in normal-land (e.g. moving up an SD gives OR>10 for previously 3SD events, but ~2 for previously ‘above average’ ones)]
Yep, I should have definitely kept the probabilities in log-form, just to be less confusing. It wouldn’t have made a huge difference to the outcome, but it seems better practice than the thing that I did.
Strongly upvoted. This comment provides solid evidence in support of its argument, and led me to substantially raise my estimate of the physical risks involved in adoption. (Edit: This holds even in light of AGB’s reply; as he points out, the numbers on risk are still quite high even if you make an adjustment of the type he recommends.)
I also appreciate Habryka’s willingness to speak out in favor of a comment that was heavily downvoted and moderated, just as I appreciated Dale’s good intentions in his initial comment.
I found Habryka’s supporting evidence to be more relevant than Dale’s, and his argument clearer — such that it meets the standards of rigor I was hoping for given the topic at hand.
*****
My biggest update here is on the rate of sibling abuse, which I hadn’t realized was nearly as high as it seems to be. From page 25 of this report:
The report’s definition of “severe assault” (ignore the bit about parents, it seems like sibling violence was measured in the same way):
I’ll note that this still leaves room for interpretation. Both of my siblings hit me with objects when I was growing up, leaving marks or bruises in some cases, but I’m still glad they were born, and we get along well as adults. (We also got along well as children, most of the time.)
I’d be really interested to see data like this broken down by the type/severity of violence. I’d guess that a substantial portion of that “37 percent” number comes from things most people would perceive as normal (e.g. a 7-year-old punches her 8-year-old brother in the arm, a 10-year-old hits his twin brother with a Wiffle bat). But I could be wrong, and even much smaller numbers for “truly severe” incidents could mean a lot more risk than I’d expected.
*****
Despite this update, “never adopt” still feels extreme to me, given the chance of a very positive outcome (a child has a loving family who helps to support them, rather than no family) and the high satisfaction rates of adoptive parents (keeping in mind that those numbers are likely to be skewed by a bias towards reporting satisfaction).
Some factors which seem like they could, in tandem, sharply reduce the risk Habryka identifies:
Adopting a child younger than your current children
Adopting a girl rather than a boy (the above report found that boys were more likely than girls to commit violent acts, but I don’t know how big the difference is)
Doing everything you can to learn whether a child has a history of violent behavior (if a 15-year-old has a totally clean record, that seems like useful evidence)
But I’m not confident about the impact of any of these.