To avoid bias from parents who stopped having children after reaching a preferred sex ratio, the researchers did not count each woman’s final birth in the analysis.
Uhhhh, I’m no statistician, but doesn’t that introduce a bias into the statistics? It means that you (potentially, depending on behaviour) have fewer examples of balanced-sex families.
Yeah, I can see the bias they are trying to account for, but fail to see how excluding a birth from the statistics helps combat this in any way. Just because you stopped having kids doesn’t have any effect on the gender of your last kid.
It does, but that may not necessarily be a bad thing. It largely depends on what the overall dataset looks like.
It’s not unusual to tweak your dataset in response to certain biases, especially if there is a known bias at play (for example, I’ve actually met plenty of parents who keep having kids hoping for a boy/girl, and then stop once they get what they initially wanted. As creepy and weird as that is to me, it’s definitely a thing).
This does seem a bit blunt of an approach, however. I would’ve preferred a survey question as part of data collection where parents are asked if they were “trying” for one sex over another, if they wanted “one of each”, etc etc., and then using that info to weight the data.
But without reading the article myself, my assumption is they just used a readily available dataset (such as medical records) rather than recruit participants directly. But I could be wrong, didn’t read it after all.
Uhhhh, I’m no statistician, but doesn’t that introduce a bias into the statistics? It means that you (potentially, depending on behaviour) have fewer examples of balanced-sex families.
Yeah, I can see the bias they are trying to account for, but fail to see how excluding a birth from the statistics helps combat this in any way. Just because you stopped having kids doesn’t have any effect on the gender of your last kid.
It does, but that may not necessarily be a bad thing. It largely depends on what the overall dataset looks like.
It’s not unusual to tweak your dataset in response to certain biases, especially if there is a known bias at play (for example, I’ve actually met plenty of parents who keep having kids hoping for a boy/girl, and then stop once they get what they initially wanted. As creepy and weird as that is to me, it’s definitely a thing).
This does seem a bit blunt of an approach, however. I would’ve preferred a survey question as part of data collection where parents are asked if they were “trying” for one sex over another, if they wanted “one of each”, etc etc., and then using that info to weight the data.
But without reading the article myself, my assumption is they just used a readily available dataset (such as medical records) rather than recruit participants directly. But I could be wrong, didn’t read it after all.