Nolan:
I would go with the -xtscc- coded regression.
In addition:
1) if (as I I surmise) -lnGDPsquared- is the squared term of -lnGDPpercapita- I would look for the turning point location;
2) other things being equal, the only country that has a role in contributing to explain variation in the regressand is Ghana (coded=2);
3) other things being equal, time as categorical variable seems to play a role in reducing the incidence of malaria (it may well be that -year- shadows some other latent variable, such as, I'm guessing here, vaccination availabiity). In the light of this evidence, I would probably give linear time and squared_time a shot (instead of considering time as a categorical variable) ans, again, I will investigate a possible turning point.
I would go with the -xtscc- coded regression.
In addition:
1) if (as I I surmise) -lnGDPsquared- is the squared term of -lnGDPpercapita- I would look for the turning point location;
2) other things being equal, the only country that has a role in contributing to explain variation in the regressand is Ghana (coded=2);
3) other things being equal, time as categorical variable seems to play a role in reducing the incidence of malaria (it may well be that -year- shadows some other latent variable, such as, I'm guessing here, vaccination availabiity). In the light of this evidence, I would probably give linear time and squared_time a shot (instead of considering time as a categorical variable) ans, again, I will investigate a possible turning point.
Comment