I am referring to the so called "Galton's problem" according to which there is spatial dependence between countries with regard to cultural differences. That is, there is diffusion of cultural values between countries. Does the fixed-effect method for analysing time-series data address this problem? I guess we need to assume that this extraneous influence is constant over the observed period? Can time-effect variables (dummies for each round of European Social Survey in my case) included in the model control for a possible change in this effect of foreign influence?
Thank you, Rajesh! I don't expect that the FE may really "capture" the spatial dependency but isn't it at least less problematic since it is not actually a cross-sectional analysis? The article seems good but it is too high level for my statistical knowledge.
Plamen
a fixed effects dummy for each country will get rid of the dependence by treating it as a nuisance but it brings no understanding of that dependence
You can use spatial panel data models such as Baltagi and Dong (1999)), or use panel data model with interactive fixed effects as proposed by Bai (2009) if the spatial dependency has a week form.
- Badges
- Science topic