I have a staggered Difference-in-Differences setting as below when examining the impact of laws implemented in different countries in different time periods: Array
When running the regression with Y is firms' asset growth, the result is as below Array
As we can see from the Table, the coefficients of the Leniency Law variable decrease in magnitude and losing the significant levels as well.
My senior friend told me one way to deal with this problem is to ignore column (6), but adding a dummy variable called US_dummy for every specification from (1) to (5). US_dummy equalling to 1 if this firm is in the US.
However, when I run the regression, I faced the omitted problem, which is explained here
I am wondering why I need to control for US_dummy variable like that and how to overcome the omitted issue here.
My solution is instead of controlling for US_dummy, I may control for US_pt, which equals US_dummy*pt. I am wondering whether I am falling into any fallacy?
Related Posts with What we should do when the coefficients of variables of interest quite different from that of whole sample?
Command collapse AND keeping the original datasetDear all, I was just thinking about this: When you analyze hierarchical structured dataset (for exa…
Overall title when using a by option with marginsplotI am using marginsplot (in V16.1) with a by option and getting a graph I like, but I can't figure ou…
RIF regression and decomposition of income inequalityI need help to have the command which is used in this paper of Rios Avila (2019) to have the results…
How to interpret the outcome of command “gllamm”?Dear statalists, I have gotten the outcome of “gllamm”,but I don't know how to illustrate,could you …
putexcel for multiple tables extending beyond Excel column ZI recently found myself with need to produce over 1000 cross-tabulations with their respective chi-s…
Subscribe to:
Post Comments (Atom)
0 Response to What we should do when the coefficients of variables of interest quite different from that of whole sample?
Post a Comment