I am having difficulty understanding what is meant in Question 9C "In
order for your estimate in part b to be the causal effect of being younger at
school entry to be valid (rather than just the effect of being born in August
vs September), what else do you need to assume?" Does is it mean why do we think this association can be causal?
I also do not understand this question. Is it trying to ask what we need to presume in any causal relationship or specifically that comparing age and ADHD vs month of birth and ADHD? Could you please address this question?
I am also not understanding this question. As well as 9B.
HI, All --
Sorry about the delay in responding to these 3 questions about #9; I was in an airplane, then on the way home.
First let's review the background. The hypothesis is that sometimes children who are younger than other children in their class who are acting age-appropriately are labeled as having ADHD because the are just doing stuff that e.g. 5-year old boys do (it's mostly boys) that 5.5- or 6-year olds do not, and there are a lot of 5.5- and 6-year olds in the class, to whom they are being compared.
So we think the real risk factor is being younger than others in the class. But Layton et al did not have a measurement of this real risk factor of interest, because they don't know the age that kids in their study actually started school. Instead, they had a measure of something that causes children to be younger than most of their peers: being born in August, compared with being born in September in a state where you e.g. have to be 5 years old by September 1 to start school.
We have learned about things that cause the exposure of interest, but are not perfectly correlated with it: examples were being randomized to be invited for screening when the exposure of interest was screening itself and having your birth hospital be randomized to the Baby Friendly Hospital Initiative when the exposure of interest was exclusive breastfeeding. And we have learned how to estimate the causal effect of an exposure of interest from the effect of one of these things that causes the exposure of interest. So reviewing that material should help you understand this question.
If not, post again! I'm home now, so I'll be watching, since I know the deadline is approaching.
Tom
Sorry about the delay in responding to these 3 questions about #9; I was in an airplane, then on the way home.
First let's review the background. The hypothesis is that sometimes children who are younger than other children in their class who are acting age-appropriately are labeled as having ADHD because the are just doing stuff that e.g. 5-year old boys do (it's mostly boys) that 5.5- or 6-year olds do not, and there are a lot of 5.5- and 6-year olds in the class, to whom they are being compared.
So we think the real risk factor is being younger than others in the class. But Layton et al did not have a measurement of this real risk factor of interest, because they don't know the age that kids in their study actually started school. Instead, they had a measure of something that causes children to be younger than most of their peers: being born in August, compared with being born in September in a state where you e.g. have to be 5 years old by September 1 to start school.
We have learned about things that cause the exposure of interest, but are not perfectly correlated with it: examples were being randomized to be invited for screening when the exposure of interest was screening itself and having your birth hospital be randomized to the Baby Friendly Hospital Initiative when the exposure of interest was exclusive breastfeeding. And we have learned how to estimate the causal effect of an exposure of interest from the effect of one of these things that causes the exposure of interest. So reviewing that material should help you understand this question.
If not, post again! I'm home now, so I'll be watching, since I know the deadline is approaching.
Tom
Thanks for posting! I had previously posted this separately but this might be a good place to review it:
I am a little confused about what question 9b is asking. If a child is born in August but is 5 years 3 months at enrollment, is the question suggesting that they are starting school at a different time (ie later than September) than September-born children? There's no other way for an August-born children to be 5 years 3 months at enrollment without starting after September, correct? Thanks!
I am a little confused about what question 9b is asking. If a child is born in August but is 5 years 3 months at enrollment, is the question suggesting that they are starting school at a different time (ie later than September) than September-born children? There's no other way for an August-born children to be 5 years 3 months at enrollment without starting after September, correct? Thanks!