Notes on childcare, revisions

[rockhopper72]

A while back I posted some notes on the science of childcare in response to some requests on this sub. People occasionally post links to it here, and I recently saw a (now deleted) post in which people discussed it. A running theme there was that people were skeptical of the research on Quebec that I referenced.

I've revised the article to say more about Quebec, to reference some recent important work and to emphasise some points that got drowned out by the length of the article.

If anyone reads the new version, I would be grateful for concrete feedback on:

• Places where you think I have not represented the state of the science accurately. With citations to back up assertions, please.
• Anywhere where I use technical language you don't follow.
• What can I cut? What is not useful to you? Where am I wordy? I've tried hard to be concise, but the article feels much too long.

Please note that

• A generic comment like "you're oversimplifying the research" is not helpful. "I think X is misleading in light of the fact that paper Y claims Z" is.
• Ad hominem rants about me or the researchers I cite don't help.
• Being doxxed could have career consequences for me, so I have a zero information policy on Reddit. That might frustrate you, but please respect it -- if you don't find the article helpful, you're not obliged to read it or to reply to this post. (It also amazes me when people take personal claims at face value here.)
• I'm not planning to add sections to the article -- it's too long already.

 

[continualseeker]

@rockhopper72 These 2 2014 Cochrane reviews found inconclusive evidence. The Cochrane standards for study inclusion are very high.

High income countries:

https://www.cochrane.org/CD010544/B...an-five-years-of-age-in-high-income-countries

full text: here

Low to mid-income countries:

https://www.cochrane.org/CD010543/B...ars-of-age-in-low-and-middle-income-countries

full text:

https://www.heart-resources.org/wp-content/uploads/2015/06/Brown-TW-2014.pdf

 

[rockhopper72]

@continualseeker Thank you for this. Focusing on the high-income one, I was rather bemused as it only considers one study. They throw away another 13 studies because there were 'unacceptable co-interventions'. The 14 don't include seminal work like the NICHD SECCYD.

It took me a bit to realise that they only consider RCTs and quasi-RCTs.

It's kind of hard to carry out RCTs in this space, because not many parents will sign up for randomly having their children sent to daycare or kept at home! I guess one reaction to that is "all the research is worthless and we know nothing". I'd say that's throwing the baby out with the bathwater. It would end up throwing out nearly all of social science!

--

Wikipedia says Cochrane is a medical setup. There are big differences in the standard of evidence between medicine and social science. In particular, individual studies in social science are shaky foundations for believing, well, anything. If nothing else, many many papers do not even correct for multiple hypothesis testing.

So you can't rely on single studies. Instead you have to rely even more on the mass of evidence -- Once you have multiple studies that reach the same conclusion, you can start to be more confident in it. Here's an example. I don't think there's a single paper on cortisol that I would trust on it's own; the sample sizes are just too small. But there are meta-analyses that bring out very clear patterns that can't possibly be chance. There's a specific paper/graph I really want to show you, but I can't find it!

 

[jmbttown]

@rockhopper72 Isn't the whole point of a Cochrane meta-analysis to gather that mass of evidence? So taking all the small but good quality studies together and seeing whether they're all pointing one way?

Doing more and more of the same fundamentally flawed studies doesn't actually make the evidence any stronger. Especially in social science where they're probably all flawed in the same way for the same reason, because that was the easiest way to collect data.

 

[rockhopper72]

@jmbttown RCT is not quite the same as 'good quality'. RCTs deal with one important problem, the 'correlation is not causation' one. However

• RCTs can still be flawed. E.g. parents are known to be bad at measuring children's behaviour. I would take a non-RCT that used teacher evaluations of behaviour over a RCT that used parent evaluations any day.
• There are other ways of teasing out causality. Quebec, etc., are important because they are natural experiments that show it's not just correlation. (Latest economics Nobel was for natural experiments, so there's lots of recent introductory podcasts on them. E.g. this.)
• Also, we can be more confident about causality when we can see mechanisms. E.g. knowing that a) daycare increases cortisol levels and knowing that b) cortisol levels affect early brain development & can cause long-term mental health issues makes us more confident about the daycare-anxiety link from Quebec.

You may still think that all social science studies are flawed. But I'll bet you believe some of their conclusions! For example:

• child abuse tends to cause long-term psychological problems.
  
• children who grow up in institutions tend to have severe problems later in life.

These are things that were not obvious to people 80 years ago and have been shown by careful research. If you take the line that all non-RCTs are flawed, you'll end up having to reject those and many others. I've never seen anyone do that consistently -- rather objections tend to be levelled selectively at conclusions people find uncomfortable.

 

[amanda212]

@rockhopper72 I wish more people understood this! This should perhaps be linked/added into how you evaluate evidence

 

[rockhopper72]

@amanda212 Ironically, I had something like this in there in early drafts but cut it for space reasons… I'm still very reluctant to put any more material in.

 

[jmbttown]

@rockhopper72 Thanks for your response. I agree RCTs are not always good quality, and not always possible. Natural experiments involving twins are pretty close, but there aren't enough of them.

However, in the natural sciences you set out to test just one variable, and you should still aim to do this in social science. For example, I'd appreciate a study asking "is the same child happier at 3pm if they're about to leave daycare, or if they're staying for 3 more hours?"

I'm afraid the examples you gave of things I believe "due to social science" are a bit weak and possibly refute your point. Child abuse is not some intuitive behavior that we need social science to warn us against - if anything, it was social scientists who used to claim that you could improve your child's behavior through various cruel interventions. The majority of parents naturally seek to protect their offspring.

And on the second point, school is an institution, so we're going to need a tighter definition of what it means to "grow up" in one. Children in care are, on average, disadvantaged from the moment of conception. Is the problem with institutions about having private space? Having a 'parental' adult in your life who prioritizes you above other children? Or is it about the education level and salary of your carers?

You can see how some of these things could be improved even within a care home setting. So I think it's unhelpful to tell care home workers that their charges will be at risk purely because the government raised them. Likewise I think it's unhelpful to tell parents that daycare is bad because it raises children's cortisol.

By the way, I'm not "defending" daycare particularly. I just think there's not enough good quality evidence about it. There are parents in this thread saying things like "I didn't know daycare was bad", so you are definitely painting a particular picture here, which may inform people's decisions.

 

[rockhopper72]

@jmbttown On institutions, I think we may be talking at cross purposes. Please read https://en.m.wikipedia.org/wiki/Orphanage -- the part about the BEIP is key.

On child abuse, I am not sure what 'cruel interventions' you are referring to. Have a read of chapter 1 of https://www.oxfordclinicalpsych.com...sych/9780198842064.001.0001/med-9780198842064 to get a sense of how it was regarded in 1940.

 

[jmbttown]

@rockhopper72 Ok, from the BEIP study I gather that children who were raised with inadequate love and affection were shown have serious problems later in life. But that was obvious to people 80 years ago. That's why Oliver Twist lived in a workhouse, and not with a loving adopted family.

Anyway I never said that all social science is bad, just that if a particular study is fundamentally flawed then it shouldn't be used at all. Just look at the absolute state of research on human diets.

Regarding cruel recommendations, I was thinking of notions like this:

"American psychologist John B. Watson warned parents of giving children too much love and affection in his book Psychological Care of Infant and Child, released in 1928.

"Never hug and kiss [children], never let them sit in your lap. If you must, kiss them once on the forehead when they say good night."

I take the book chapter you gave to be a counter example; yes we can agree that advice has moved on and attachment parenting is now encouraged. I simply disagree that without social scientists I wouldn't know that neglect is harmful.

Of course there must be some things we've learned through social science, and again, I'm not against the whole discipline. I just think much of the research is not good enough quality.

 

[vanesyu]

@rockhopper72 Orphanage

Historically, an orphanage is a residential institution, or group home, devoted to the care of orphans and other children who were separated from their biological families. Examples of what would cause a child to be placed in orphanages are when the parents were deceased, the biological family was abusive to the child, there was substance abuse or mental illness in the biological home that was detrimental to the child, or the parents had to leave to work elsewhere and were unable or unwilling to take the child. The role of legal responsibility for the support of children whose parent(s) have died or are otherwise unable to provide care differs internationally.

[sup][ [/sup][sup]F.A.Q[/sup][sup] | [/sup][sup]Opt Out[/sup][sup] | [/sup][sup]Opt Out Of Subreddit[/sup][sup] | [/sup][sup]GitHub[/sup][sup] ] Downvote to remove | v1.5[/sup]

 

[continualseeker]

@rockhopper72

I'd say that's throwing the baby out with the bathwater. It would end up throwing out nearly all of social science!

They would probably say that they threw out the bathwater and kept all the babies.

They are pretty skeptical. Not too long ago Cochrane was notorious for saying that there was insufficient evidence for the flu vaccine. I seem to recall that this was due to a dearth of RTCs and to the fact that much of the observational data did not distinguish between actual influenza and influenza-like illness. The evidence got better and they are now assert probable modest benefits for adults and higher certainty benefits for children.

 

[kira23]

@rockhopper72 It’s hard to carry out RCTs, but there are plenty of ways to do quasi-experimental designs (difference-in-difference, regression discontinuity, instrumental variables, etc.) on this type of topic. I don’t think it’s fair to say Cochrane throws the baby out with the bath water for requiring rigorous methodology.

 

[rockhopper72]

@kira23 I don’t think they admit DID, etc? Their definition of quasi-RCT seemed quite restrictive. But I’m only going by the definition - feel free to correct me. (https://handbook-5-1.cochrane.org/c...e_definitions_and_criteria_for_randomized.htm )

 

[hettie]

@continualseeker This research seems quite old (2014) and having a look I found intersting papers. For examples from this year found this: https://www.tandfonline.com/doi/abs/10.1080/10409289.2021.1995687

The blog has extensive citations, just a review from 2014 won't be enough to discredit it.

 

[hettie]

@rockhopper72 Whoa, this amazing research! And now I am worried as my daughter goes ~40h of childcare a week.. I'll read the citations, and give you feedback. It will take time but I will for sure

 

[soulwinner92]

@rockhopper72 I find your findings for Quebec very weird and will look into this study later. Everything I've saw was going in the opposite direction.

 

[rockhopper72]

@soulwinner92 I would like to stress that they are not my findings! I quoted the NBER summary precisely so that no one thought I was trying to spin the research.

When you say everything you’ve seen, do you mean research or news articles or personal experience?

 

[soulwinner92]

@rockhopper72 Both. I live in Quebec and work with school-age children (in a low SES neighborhood). I see them coming to school with or without daycare prior to school and can say that attending a daycare helps a lot of expected social behaviors at 5 yo. Of course I know you were getting findings from research.

I'll try to dig some info. Also, it is to note that not all kids attend daycare in Quebec (around 57% in 2014). From all kids in educational daycare, not all kid attend governmental daycare (32% in 2017)(22% in private daycare, 16% in private subsidized daycare, 31% in family-based daycare (a lady who keep kids in her home but that follow certain guidelines)) There are also around 16% of kids in unrecognized daycare. We also have pre-kindergarten (4 yo) in low-income neighborhoods, because a lot of the kids in those neighborhood don't go to daycare. Pre-K is (almost) free whereas daycare range from 9$ per day (government daycare) to 50$ a day (private daycare).

There has been studies about the enormous gap in the quality of care. Government daycare is the best care (https://link.springer.com/article/10.1007/s10566-009-9088-4 ), but less low-income children are in them, maybe due to the difficulty of access, the cost of going to daycare and the lack of flexibility (doesn't allow to have a night job for example).

Chilcare in Quebec seems to be a great equalizer for readiness to school:

https://osf.io/vmh9b/

Quality of daycare impacts the outcomes of the children. Quality of daycare depends on multiple factors: https://www.tandfonline.com/doi/full/10.1080/03004430.2016.1202948

You need these quality factors to be able to predict LOWER rates of temperamentally difficult children: https://psycnet.apa.org/record/2014-13067-001

https://www.tandfonline.com/doi/full/10.1080/02568543.2015.1009201

A big big survey of Child Development in Kindergarten in Quebec was done in 2012 and 2017 and fueled a lot of research : https://statistique.quebec.ca/en/en...ey-of-child-development-in-kindergarten-qscdk

*****

So to take Quebec as a whole and say that crime has risen, when only a very small subset of children had access to good quality daycare is a bit misleading at least, isn't it?

Also, question : would you consider better mother well-being and lower number of household under the poverty line to be good consequences of having affordable daycare so that mothers can go back to work?

 

[rockhopper72]

@soulwinner92 Lots in here -- I'll take a couple of points to start with.

> Childcare in Quebec seems to be a great equalizer for readiness to school:

https://osf.io/vmh9b/

That's consistent with what I wrote. I said that low-income children benefit from starting daycare earlier, and high-income children from starting later. Another way of putting that is that at a fixed age (e.g. 2) low-income children do better in daycare and high-income children do worse. So if all children are put in daycare it does indeed narrow the gap in outcomes.

That said, school readiness is a fuzzy concept and is often used to include things which are known to be bad for children in the long run. E.g.

There was considerable pressure for children to attain early independence, both physical and emotional, under the apparent misapprehension that this can be equated with ‘good' development. In fact, other research suggests that it is more likely to produce a kind of ‘pseudo self-reliance' described by Eva Holmes (1977), which prevents children from seeking appropriate support from adults and inhibits their learning processes.

Jackson, 1996. People Under Three: Young Children in Day Care.

There's a much more thorough discussion in Pianta (1999). Enhancing relationships between children and teachers.

-------------------------

> So to take Quebec as a whole and say that crime has risen, when only a very small subset of children had access to good quality daycare is a bit misleading at least, isn't it?

Can you say more about how it is misleading? (I don't mean that to be confrontational -- I really want feedback here.) I wrote

Overall quality was actually pretty bad, but still ‘comparable to the quality of care provided in many other countries’. 25% of the centers were good quality, which is about double the figure for the US.

Which is to say, only a fraction of children have access to good quality daycare anywhere. Given the findings that parents are remarkably bad at measuring quality, does it help to talk about how children would hypothetically do in high-quality centers?

I'd honestly really welcome feedback here! I do not want to mislead people about what they can realistically expect to see, but other than that I'm very open to making changes to the article.