time variables in a regression

#1
Hi all,

I'm looking at a linear regression model using the data on wars in the 20th century. There are two time variables on the right-hand side: 1) year and 2) a dummy variable that equals 1 if a war occured after 1945. My question is: what is the coefficient on this dummy really telling us? It doesn't seem like a valid counterfactual to say: "holding all other variables (including year) constant, what is the effect of going from a pre- to a post-1945 era?"

Thanks in advance for your help. Also, I'm new here, so please let me know if this post is more appropriate in some other section of the forum.
 

vinux

Dark Knight
#2
Hi all,

I'm looking at a linear regression model using the data on wars in the 20th century. There are two time variables on the right-hand side: 1) year and 2) a dummy variable that equals 1 if a war occured after 1945. My question is: what is the coefficient on this dummy really telling us? It doesn't seem like a valid counterfactual to say: "holding all other variables (including year) constant, what is the effect of going from a pre- to a post-1945 era?"

Thanks in advance for your help. Also, I'm new here, so please let me know if this post is more appropriate in some other section of the forum.
What is your observations? I mean, is it a country level data?
I think the dummy variable make sense. I guess you are analyzing the wars after World War II (1939–1945).
 
#3
hi,
this is conflict-level data, and there are conflicts both before and after WWII (so the "post-WWII" dummy takes on values of both 1 and 0 in the data set). the "year" variable is the year the war started. the dependent variable is an index of war intenstity.



What is your observations? I mean, is it a country level data?
I think the dummy variable make sense. I guess you are analyzing the wars after World War II (1939–1945).