In: Economics
A researcher wants to look at how daycare for children affects parents’ employment. He considers 5 possible research designs. In which of these should we be most worried about confounding/lurking variables?
Question 1 options:
In a randomized control trial of a policy providing daycare, with randomization at the individual level |
|
In a study that compares employment rates of parents before and after the introduction of free government provided daycare in one Canadian province |
|
In a randomized control trial of a policy providing daycare, with randomization by area of residence |
|
In an observational study that compares the employment rates of parents whose children are in daycare with those whose children are not |
|
In an observational study that compares employment rates of parents living in provinces that have free government-provided daycare compared with parents living in provinces without a government daycare program |
So a researcher basically wants to look at whether parents are able to dedicate their time for jobs and can afford the day care expenses at the time.
Randomized control trial should either have two options, one which is the control group which is not being affected and one where the policy is being implemented.
The first option with randomization at individual level should be the one which is most confounding as every individual will be different, one either has to select a complete province or a selected area, as the variables won't help figure out how has the impact truly been.
Second option is clear and precise as it classifies between two scenarios. By area of residence will also help determine the result. Fourth is clear in comparing children who are in daycare vs those who are not. Fifth is also appropriate as it distinguishes clearly between government provided free day care and how it will impact employment.
Thus the first option is the most confounding one where there might be more variables which would not help in getting a meaningful answer