This brings up a lot of questions about how people read and make sense of data, and what are the most effective ways of communicating about data. I'm not sure that this piece is concise and clear enough. If a reader already knows something about probability, sample size, and the nature of correlations, s/he probably gets it. If not, I don't think this is reader-friendly enough to help you learn it.
Samuel Clemens had it right when he said,"There are lies, damned lies, and statistics". Selective use of measurements can
Extremely weak data is the same as no data. It's better to admit this (and design a better experiment) than to say "That's all we have, so we will use it."
You've cooked up a useful illustration of a real statistical problem (thank you), but then--strangely--seem to completely sidestep what you really need to sustain your original column: a demonstration that the sparse sampling of the study is what led to extinguishing (what you think) is a real effect. I was really looking forward to some technical rigor to back up what appeared to be nothing more than your personal bias against what the study found, and am really not seeing any.
5
Thanks for that clarification. This is most helpful and now I understand your reasonng I failed to comprehend in your last article.
This article is a case of too much statistical analysis and too little economic thinking.
A key concept in economics is opportunity cost. When parents spend time with their kids, the children are losing time they could be doing homework and preparing for exams. Thus children who receive parental care will tend to end up becoming "out-of-the-box," conceptual thinkers. These children may do worse on standardized tests than those who do not engage in 'enriched activities.' Often you find 'students' whose capabilities far surpass those of the teachers and professors. This situation has often been described by Rousseau, J.D. Salinger, and other thinkers. I encountered it myself when I tried to teach my son the underlying ideas of mathematics which caused him to perform poorly on exams. Chris
A key concept in economics is opportunity cost. When parents spend time with their kids, the children are losing time they could be doing homework and preparing for exams. Thus children who receive parental care will tend to end up becoming "out-of-the-box," conceptual thinkers. These children may do worse on standardized tests than those who do not engage in 'enriched activities.' Often you find 'students' whose capabilities far surpass those of the teachers and professors. This situation has often been described by Rousseau, J.D. Salinger, and other thinkers. I encountered it myself when I tried to teach my son the underlying ideas of mathematics which caused him to perform poorly on exams. Chris
2
Ah legerdemain and reductio ad absurdum comments on the parenting issue once again. I could care less about the efficacy of the statistical regime employed by the authors of the "study" in question.
It is simply an axiom of human existence that time lovingly spent with one's children by both parents is more beneficial to the long-term health (emotional and physical), happiness, and success of those children than to not spend available parental time with them by leaving them alone to their own devices with or without their peers or with some third-party surrogate.
It is terribly unfortunate that some families find themselves for economic reasons unable to experience meaningful parent child time through no choice or fault of their own. It is also terribly unfortunate that some families find themselves for reasons of career (male or female), wealth accumulation and/or social status/climbing unable and therefore unwilling to experience meaningful parent child time.
These facts do not disturb the verity of the axiom nor does any study posited as an apology for parental self absorption.
It is simply an axiom of human existence that time lovingly spent with one's children by both parents is more beneficial to the long-term health (emotional and physical), happiness, and success of those children than to not spend available parental time with them by leaving them alone to their own devices with or without their peers or with some third-party surrogate.
It is terribly unfortunate that some families find themselves for economic reasons unable to experience meaningful parent child time through no choice or fault of their own. It is also terribly unfortunate that some families find themselves for reasons of career (male or female), wealth accumulation and/or social status/climbing unable and therefore unwilling to experience meaningful parent child time.
These facts do not disturb the verity of the axiom nor does any study posited as an apology for parental self absorption.
3
On the first day of stats class in grad school, the professor had us all say together, "your model is mis-specified!" They always are.
Thanks for the memories.
Thanks for the memories.
2
I like numbers too, and science is always a good thing. But when you are dealing with young people and parents, many of whom who have a myriad of baggage that they're acting out, numbers don't always tell the true story.
I'll say it again: Ask any old-time schoolteacher who isn't married to the latest education fad or philosophy of child-rearing, about parental involvement in their children's lives. She'll have 30-40 years of observation to go by.
I'll say it again: Ask any old-time schoolteacher who isn't married to the latest education fad or philosophy of child-rearing, about parental involvement in their children's lives. She'll have 30-40 years of observation to go by.
4
What are ages of "the kids" that were studied? Do parents spend more time with, say, younger (infant/preschool) children than those in school full-time? And what about diversity factors? Socio-economic factors? Live-in extended family? It all seems pretty vague to me.
1
Read this and the first article as well as many comments. Regardless of the research design issues Wolfers maintains are there and the overwhelming attention to "parenting" time, the original research article is about mothers' time spent with kids. Isn't extending this to both parents (or perhaps other caregivers) problematic? Wolfers and commenters seem to be falling into another methodological trap.
2
Unless I am mistaken, what Wolfers is actually saying is that this study might at best accurately measure only average parenting time and that, almost tautologically, this correlates very closely to an average of outcomes (however measured).
To explain:
The unstated (and unsupported) assumption of the study is that measuring parenting time of many parents over a short period can be extrapolated to a measure of parenting time of each of those same parents over a long period; then a comparison of this time to outcomes is possible.
In fact, statistical theory says only that measuring the parenting time of those parents over a short period can be extrapolated to a measure of parenting time of many more parents over the same short period. Which gives you a pretty good measure of average parenting time--but cannot be compared to outcomes, except on average.
To explain:
The unstated (and unsupported) assumption of the study is that measuring parenting time of many parents over a short period can be extrapolated to a measure of parenting time of each of those same parents over a long period; then a comparison of this time to outcomes is possible.
In fact, statistical theory says only that measuring the parenting time of those parents over a short period can be extrapolated to a measure of parenting time of many more parents over the same short period. Which gives you a pretty good measure of average parenting time--but cannot be compared to outcomes, except on average.
Correlation & causation ?
I am uncomfortable with the Disney example.
I made a choice when my children were young, never to take them to Disneyland, Disneyworld, Eurodisney or any Disney.
If I were being monitored as a parent with children at Disney my facial expression would be one, not of bliss, but one of pure pain.
Any Disney park or product is not life as I see it, but covered with the lenses of "rose-colored glasses".
I did however, read to my children every night.
I am uncomfortable with the Disney example.
I made a choice when my children were young, never to take them to Disneyland, Disneyworld, Eurodisney or any Disney.
If I were being monitored as a parent with children at Disney my facial expression would be one, not of bliss, but one of pure pain.
Any Disney park or product is not life as I see it, but covered with the lenses of "rose-colored glasses".
I did however, read to my children every night.
3
Isn't the Disney trip just an illustration? Wolfers' other article states he, himself, just returned from a trip to Disney with the kids. Maybe he merely had Disney on his mind.
I took my child to Disney and when he was in middle school to Ireland. He also visited museums and lots of sights in NYC and Washington, local national parks, beaches, county fairs etc. He liked both types of things""real" and "fantasy." I don't think either hurt-- "to each his own." And after all, many classic books are fantasies so reading has that aspect, too.
I took my child to Disney and when he was in middle school to Ireland. He also visited museums and lots of sights in NYC and Washington, local national parks, beaches, county fairs etc. He liked both types of things""real" and "fantasy." I don't think either hurt-- "to each his own." And after all, many classic books are fantasies so reading has that aspect, too.
3
I totally support your last sentence calling for careful reading of data.
A mentor once shared there are 2 types of folks out there: those who are model-driven and those who are data-driven. Though you claim you'd rather stick to the numbers, your stance here appears to be model-driven, branding the study's choice of 1 weekday and 1 weekend-day for detailed diary entries as "arbitrary" and "unreliable."
For all the time it must have taken you to generate your gorgeous "illustrative examples," it's too bad you didn't choose to spend just a tiny bit more time to stick to the numbers and simply graph the study participants' diary data ("numbers") against parent/mother time with kids Sunday-Saturday data ("numbers").
That might have helped you clarify for me why a scientific claim doesn't add up.
Instead, this back&forth demonstrates how science communication can be value- or model-driven, not data-driven, despite your closing your preference to "stick with the numbers."
A mentor once shared there are 2 types of folks out there: those who are model-driven and those who are data-driven. Though you claim you'd rather stick to the numbers, your stance here appears to be model-driven, branding the study's choice of 1 weekday and 1 weekend-day for detailed diary entries as "arbitrary" and "unreliable."
For all the time it must have taken you to generate your gorgeous "illustrative examples," it's too bad you didn't choose to spend just a tiny bit more time to stick to the numbers and simply graph the study participants' diary data ("numbers") against parent/mother time with kids Sunday-Saturday data ("numbers").
That might have helped you clarify for me why a scientific claim doesn't add up.
Instead, this back&forth demonstrates how science communication can be value- or model-driven, not data-driven, despite your closing your preference to "stick with the numbers."
4
The silly and tender moments you share together are what makes a family. Just tonight, my six year old daughter and I played catch in the living room and read books. We cuddled in bed and she stroked my hair. During that time, we joked and talked a lot and connected. I personally benefit from time with my kids every single day, it's not just about them. Kids are awesome if you take the time with them!
6
I think you picked the wrong horse in this race.
Much is to be made of this study, but perhaps the most important point of discussion ought to start with the most important data point: the so called "positive outcomes".
You mentioned these were judged by a "range of test scores." Ok, let's assume (because I don't know) that these test scores are a combination of physical, emotional, intellectual, and academic markers. I would still argue that even with a load of data, "positive outcome" CANNOT be reduced to properties that behave/don't behave according to one catalyst, i.e. mothering.
Putting it another way: I know some physically healthy, accomplished, charming individuals that I would not welcome in my home. (And anecdotally, those are usually the types whose parents DID NOT spend a ton of quality time with them--but that's just anecdotal). :-)
So much goes into parenting that I don't need a study to inform me as to what I can see with my own eyes. If i want my children to turn out the way *I want them to*, then the more time I spend with them. the greater the "outcome" *according to the goals I, as a parent, set.* Try and measure that!
The needs of children are forever morphing- from when they are babies until they are teens. Temperament, economics, special needs, predispositions all have to be accounted for as they are in families.
In the end, I believe that how much a child is loved trumps what kind of time is spent with them.
...and love cannot be measured.
Much is to be made of this study, but perhaps the most important point of discussion ought to start with the most important data point: the so called "positive outcomes".
You mentioned these were judged by a "range of test scores." Ok, let's assume (because I don't know) that these test scores are a combination of physical, emotional, intellectual, and academic markers. I would still argue that even with a load of data, "positive outcome" CANNOT be reduced to properties that behave/don't behave according to one catalyst, i.e. mothering.
Putting it another way: I know some physically healthy, accomplished, charming individuals that I would not welcome in my home. (And anecdotally, those are usually the types whose parents DID NOT spend a ton of quality time with them--but that's just anecdotal). :-)
So much goes into parenting that I don't need a study to inform me as to what I can see with my own eyes. If i want my children to turn out the way *I want them to*, then the more time I spend with them. the greater the "outcome" *according to the goals I, as a parent, set.* Try and measure that!
The needs of children are forever morphing- from when they are babies until they are teens. Temperament, economics, special needs, predispositions all have to be accounted for as they are in families.
In the end, I believe that how much a child is loved trumps what kind of time is spent with them.
...and love cannot be measured.
10
So, I was settling in to read an interesting technical argument about why, when there is a correlation involved, the sampling errors in data of this sort will not tend to cancel out. However, as far as I can tell, the argument is that in fact they WILL tend to cancel out, but if the sample size is too small an underlying correlation might be lost in the noise. Three things about this: 1) I think this is tantamount to saying the statistical methodology of the original study was valid, but their sample size might have been too small. If so, that claim--if it even counts as a claim--seems like a weak substitute for an actual meta-analysis. 2) This seems a far cry from the initial article, which seemed to state categorically that since the input data was noisy, the result of the study was not valid. (Let alone the headline, which implied that a conclusion opposite to that of the study could be drawn.) 3) This still reads like an opinion piece, not a statistical argument (scatter plots notwithstanding).
8
This is not correct; measurement error does not cancel out in a larger sample. As long as there is measurement error, the estimated correlation will (in most cases) be lower than the true correlation. For more information, look on Google or Wikipedia for "errors-in-variables" or "attenuation bias."
1) is not right, although the article does leave that impression. Measurement error in an independent variable like parenting time actually does introduce bias, not just decrease statistical power -- this kind of error pushes the correlation downward, no matter how large your sample size is. So the author's basic criticism, that the study methodology is guaranteed to underestimate the true correlation between parenting time and outcomes, sounds correct.
See: http://en.wikipedia.org/wiki/Regression_dilution
See: http://en.wikipedia.org/wiki/Regression_dilution
I suspect family wealth may show a stronger correlation than time spent. It's not that children are paid to have good outcomes but rather that wealth can improve the quality of the time spent. Even if the time spent has no direct bearing on spending, like a trip, the adult's attitude and concerns about money can reinforce success or failure.
I mention wealth for two particular reasons. First, President Obama gave a speech about 18 months ago and he said family wealth was shown to be the most important predictor of family success. Second, the poorer half of the population has gotten so much poorer since 1995 going from a 4% share of family wealth to just 1%. This has created an economic crisis that has harmed both family formation (marriage) and child rearing.
I also suspect that wealth is most important not in the sense of having an abundance will significantly improve outcomes but rather having too little can cause worry and hardship that leads to a breakdown in the role model function of the adults. More study is needed.
I mention wealth for two particular reasons. First, President Obama gave a speech about 18 months ago and he said family wealth was shown to be the most important predictor of family success. Second, the poorer half of the population has gotten so much poorer since 1995 going from a 4% share of family wealth to just 1%. This has created an economic crisis that has harmed both family formation (marriage) and child rearing.
I also suspect that wealth is most important not in the sense of having an abundance will significantly improve outcomes but rather having too little can cause worry and hardship that leads to a breakdown in the role model function of the adults. More study is needed.
6
@Eugene Patrick Devaney: Not totally disagreeing with your point, but I've seen, over a period of decades, 1% families where both parents were so busy with their demanding jobs that raising the kids sort of got outsourced to school and the nanny/housekeeper. I've always felt sorry for those kids (now adults) because they definitely got the message that money was more important than they were.
4
Mr. Devany, I would argue that wealth is not as much of a factor as you would suggest.
1) If someone is poor, does that take a person's agency away? It is free to go on a walk, and observe flowers, nature, etc. A parent can take his or her child for a free 10 minute walk and point out to the child kindly the things of nature such as flowers, and ask gentle observational questions about the environment.
2) A parent is role model in how he or she chooses to behave. If one isn't sure of how to behave, one can seek out other role models for support.
3) What has worsened is the cultural milieu, which has put pressure on new parents in 2 ways that make it harder to parent well
a) endless bombardment from media (smart phones, TV) which Is distracting and, again, if one does not choose to focus on the child but instead chooses to focus on media time with the child will be lost
b) active cultural approval of single parenthood for never wed parents as well as cultural approval of adultery which causes trauma to children of divorce, thus decreasing the emotional resource available to children if there is only one active parent, and actually draining resources fro the child or children in the case of adultery by traumatizing the faithful spouse as well as the children.
1) If someone is poor, does that take a person's agency away? It is free to go on a walk, and observe flowers, nature, etc. A parent can take his or her child for a free 10 minute walk and point out to the child kindly the things of nature such as flowers, and ask gentle observational questions about the environment.
2) A parent is role model in how he or she chooses to behave. If one isn't sure of how to behave, one can seek out other role models for support.
3) What has worsened is the cultural milieu, which has put pressure on new parents in 2 ways that make it harder to parent well
a) endless bombardment from media (smart phones, TV) which Is distracting and, again, if one does not choose to focus on the child but instead chooses to focus on media time with the child will be lost
b) active cultural approval of single parenthood for never wed parents as well as cultural approval of adultery which causes trauma to children of divorce, thus decreasing the emotional resource available to children if there is only one active parent, and actually draining resources fro the child or children in the case of adultery by traumatizing the faithful spouse as well as the children.
1
This essay has a "one size fits all" approach to parenting. For the last 17 years, I have been watching 5 families of grandchildren grow up. Styles of parenting vary hugely. Some families encourage and even require independence: "Your homework your piano practice etc. is your responsibility. I will help when you ask, but these things are your job, not mine." Other families monitor and correct every assignment. Some are in between. I suspect that the kids will all be fine. Think of all the normal, reasonably happy adults you know. Do you think they were all parented the same way?
One should also realize that not only do parents bring up children, but children bring up parents. What the parents do affects what the kids do, and what the kids do affects what the parents do. Some kids don't need monitoring and resent it. Other kids need micro managing. "One size does NOT fit all".
One should also realize that not only do parents bring up children, but children bring up parents. What the parents do affects what the kids do, and what the kids do affects what the parents do. Some kids don't need monitoring and resent it. Other kids need micro managing. "One size does NOT fit all".
16
Wisdom!!!!
Did you read the article? It doesn't say anything about one correct way to parent. It's just saying that a particular statistical result is biased. It doesn't make any claim that more parental time is necessarily good, just that we might get that statistical result EVEN IF more parental time is good.
3
"Typically — though not always! — measurement error results in weaker correlations and so it may account for many nonfindings, like the failure to find a relationship between child outcomes and time spent parenting."
So, once again, the argument Wolfers makes is speculative. Measurement error might matter; then again it might not. Thanks. Notice in the first column he was more definitive about the study's lack of validity. After being spanked by readers, he backs off. It is also interesting that of the hundreds of economics studies that suffer from measurement error, he picks one particular one, done by sociologists, that contradicts what appears to be a strongly held belief. Perhaps he should subject his own work to this kind of innuendo.
So, once again, the argument Wolfers makes is speculative. Measurement error might matter; then again it might not. Thanks. Notice in the first column he was more definitive about the study's lack of validity. After being spanked by readers, he backs off. It is also interesting that of the hundreds of economics studies that suffer from measurement error, he picks one particular one, done by sociologists, that contradicts what appears to be a strongly held belief. Perhaps he should subject his own work to this kind of innuendo.
17
The problem with this discussion is its over reliance on quantity as opposed to quality of time with kids.
Those who make time to attend virtually all of kids' performances in plays and sports events as well as helping coach sports teams will find that their kids will never forget this attention few of their peers are lucky enough to receive.
Incidentally, if you would like to rear kids who develop into accountable but empowered adults who are productive members of society, take the trouble to source the best primer on parenting ever written "How to Talk So Your Kids Will Listen & How to Listen So Your Kids Will Talk" by Masler provides a pragmatic guide to developing highly productive kids with great social skills.
Earlier this was out of print but it is available in used book stores and as an ebook.
Both of our kids finished college with science degrees in four years, are at the top of their respective fields, and have lifetime support circles from their college buddies.
Those who make time to attend virtually all of kids' performances in plays and sports events as well as helping coach sports teams will find that their kids will never forget this attention few of their peers are lucky enough to receive.
Incidentally, if you would like to rear kids who develop into accountable but empowered adults who are productive members of society, take the trouble to source the best primer on parenting ever written "How to Talk So Your Kids Will Listen & How to Listen So Your Kids Will Talk" by Masler provides a pragmatic guide to developing highly productive kids with great social skills.
Earlier this was out of print but it is available in used book stores and as an ebook.
Both of our kids finished college with science degrees in four years, are at the top of their respective fields, and have lifetime support circles from their college buddies.
5
As I commented on your previous article regardless of the measurement parental presence is not parenting. Whether active involvement or modeling behavior parenting is not a strictly passive experience that can be measured in such a simplistic model. I am weary of studies that attempt to measure quantitatively a qualitative outcome. I will console myself with watching my successful young men.
5