I am trying to predict what are the determinants for foreign direct investments in Lithuanian regions from 2011 to 2017.

My dependent variable is foreign direct investments per capita in Lithuanian regions. And I have a variety of explanatory variables such as unemployment rate, labor cost, infrastructure quality and so on.

However, the data is really not normally distributed. Data does not show any linearity.

enter image description here

That is why I cannot use the multiple linear regression.

enter image description here

I found out that for the data which has a Poisson distribution, two regressions are suitable: Poisson and Negative binomial regression.

According to my calculations the dispersion is higher that the mean, so I should use the negative binomial regression.

Please correct me if I am wrong.

Also, how do you account for changing variables in time? (From 2011 to 2017)

  • 2
    $\begingroup$ The variables do not need to be normally distributed to perform a linear regression - the residuals do (I am not advocating a linear regression here, but this is something you should understand). $\endgroup$ – mkt - Reinstate Monica Jul 13 '18 at 8:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.