Hello sir, thank you for the video, very informative! I was wondering if you wanted to investigate the returns to education, for example, across the different regions in the dataset (reg662-reg669) would you divide the dataset into sub-samples or would you just include the regions as dummy variables. I am just struggling to understand if you include the regions as dummy variables how we would be able to analyse the differing in returns to education. Thanks
First, reg662 to reg669 are all dummy variables. We include them in the model as control variables to keep the geographic factors fixed because they affect the outcome variable. After ruling out the regional effects, the education effect is more "pure" or ceteris paribus. Including the regions means "If we are from the same region, what is the return to education?" Second, if we want to find the return to education in different regions, we can use the subsamples for different regions. Third, if there is one dummy, say we want to find the difference in return to education between south and north, we can add the interaction term between south and education to the model.
i love you Bob, becaus of you i solved my stata project problem.
Very informative. Great explanation. Thank you so much.
Hello sir, thank you for the video, very informative! I was wondering if you wanted to investigate the returns to education, for example, across the different regions in the dataset (reg662-reg669) would you divide the dataset into sub-samples or would you just include the regions as dummy variables. I am just struggling to understand if you include the regions as dummy variables how we would be able to analyse the differing in returns to education. Thanks
First, reg662 to reg669 are all dummy variables. We include them in the model as control variables to keep the geographic factors fixed because they affect the outcome variable. After ruling out the regional effects, the education effect is more "pure" or ceteris paribus. Including the regions means "If we are from the same region, what is the return to education?" Second, if we want to find the return to education in different regions, we can use the subsamples for different regions. Third, if there is one dummy, say we want to find the difference in return to education between south and north, we can add the interaction term between south and education to the model.
9:57 the 95 percent CI of the 2SLS estimate almost covers that of the OLS estimate