Dear Stata-fan
I am working on my master's research and my current regression is: xtpoisson AV_Number of_Farm businesses OV1_Highly Educated Capita_OV2_AcademicInstitutes ln_OV3_Market Potential Capita_OV4_Chemical Businesses OV5_Government Subsidies OV6_Expenditure_Health Care CV3_GDP_Growth CV5_Inflation CV6_Business Taxes, fe exposure(CV2_Population) vce(robust).
Now I have already done several tests or choices based on the literature (FE/RE, Poisson/negative binomial, multicollinearity, heteroskedasticity). Then I also did the RESET test based on this example (https://personal.lse.ac.uk/tenreyro/reset.do) where I found a p-value of 0.0512 and chi2 of 3.80. So I can conclude that there is no evidence on model misspecification (although this is only just above 0.05 significance and so should be looked at with caution?).
I myself read several things about residuals but I haven't yet found anything concrete about this that works in stata and works with an xtpoisson regression. In the later stages of my research I am going to run other types of regressions (looking for lagged effects, and looking for differences between western and non-western countries)
My question is if anyone has any advice for next steps to carry out, and which tests I can still specifically carry out, as my supervisor is absent this week and I do want to continue working on my research.
Thanks in advance for any possible advice
Ward Bruurs
I am working on my master's research and my current regression is: xtpoisson AV_Number of_Farm businesses OV1_Highly Educated Capita_OV2_AcademicInstitutes ln_OV3_Market Potential Capita_OV4_Chemical Businesses OV5_Government Subsidies OV6_Expenditure_Health Care CV3_GDP_Growth CV5_Inflation CV6_Business Taxes, fe exposure(CV2_Population) vce(robust).
Now I have already done several tests or choices based on the literature (FE/RE, Poisson/negative binomial, multicollinearity, heteroskedasticity). Then I also did the RESET test based on this example (https://personal.lse.ac.uk/tenreyro/reset.do) where I found a p-value of 0.0512 and chi2 of 3.80. So I can conclude that there is no evidence on model misspecification (although this is only just above 0.05 significance and so should be looked at with caution?).
I myself read several things about residuals but I haven't yet found anything concrete about this that works in stata and works with an xtpoisson regression. In the later stages of my research I am going to run other types of regressions (looking for lagged effects, and looking for differences between western and non-western countries)
My question is if anyone has any advice for next steps to carry out, and which tests I can still specifically carry out, as my supervisor is absent this week and I do want to continue working on my research.
Thanks in advance for any possible advice
Ward Bruurs
