Hi all,
I have 3 waves of panel data on children whose development I analyse. I look at the effects of parental health shocks (a binary indicator) on a continuous development score in linear regressions and a binary indicator of clinically troublesome development in logit regressions. Both regressions are fixed effects with clustered standard errors. There are some variables in the data which the literature indicates may lie on the pathway of effect between parental health shocks and children's development, i.e. once parents have a health shock, they may experience a loss of wealth, and thus children's development may suffer as their economic environment is worse.
There are also changes in parental health behaviours and the relationship with the child.
I would like to analyse the importance of the variables that I have in the data and suspect may be responsible for influencing children's development as a result of parental health shocks in the relationship I find between health shocks and development. I had thought I could use SEM and GSEM in Stata to do this, but the more I read on this the more confused I get. In the manual only example 42g seems to get close to what I want (on page 441 of https://www.stata.com/manuals/sem.pdf) but doesn't include fixed effects. I'm not sure if this is possible in sem or gsem with fixed effects and if I should instead be using Richard Williams "xtdpdml" command, or something else entirely.
Can anyone give me any advice on what type of model I should be considering for an analysis that goes from:
Health Shock --> Children's Development
to
Health Shock --> Economic Declines --> Children's Development
Where all my variables are observed and I am running longitudinal fixed effects regressions?
Thanks in advance,
John
I have 3 waves of panel data on children whose development I analyse. I look at the effects of parental health shocks (a binary indicator) on a continuous development score in linear regressions and a binary indicator of clinically troublesome development in logit regressions. Both regressions are fixed effects with clustered standard errors. There are some variables in the data which the literature indicates may lie on the pathway of effect between parental health shocks and children's development, i.e. once parents have a health shock, they may experience a loss of wealth, and thus children's development may suffer as their economic environment is worse.
There are also changes in parental health behaviours and the relationship with the child.
I would like to analyse the importance of the variables that I have in the data and suspect may be responsible for influencing children's development as a result of parental health shocks in the relationship I find between health shocks and development. I had thought I could use SEM and GSEM in Stata to do this, but the more I read on this the more confused I get. In the manual only example 42g seems to get close to what I want (on page 441 of https://www.stata.com/manuals/sem.pdf) but doesn't include fixed effects. I'm not sure if this is possible in sem or gsem with fixed effects and if I should instead be using Richard Williams "xtdpdml" command, or something else entirely.
Can anyone give me any advice on what type of model I should be considering for an analysis that goes from:
Health Shock --> Children's Development
to
Health Shock --> Economic Declines --> Children's Development
Where all my variables are observed and I am running longitudinal fixed effects regressions?
Thanks in advance,
John
