Hi all,

I've a question which regression model to use? I've the following model:

Taxavoid= PC + Before/after + PC * Befor/after
tax avoid = continious
where PC is ratio variable of political party/ i coudl have used 1 or 0 (REPvsDEM) but, I use ratio, more info. PC= REP / (DEM+REP) so ratio on republican contribution.
Before/after =eventdate 1 is after 31.12.2017, before 0.
my range of financial data I use of the companies is 2015-2019.
My res question is if there is more tax avoid after 31.12.2017? and
Hyp2 this effect is more on REP

just an simple research, but I was wondering if I have to use Fixed effects, Random effects or ols cluster perhaps. please help me.
And which one that is? I've done huasman and was significant so no random, but more I was wondering on my ind variabele PC as this will be fixed, so cannot be measured in the regression. same for industry by doing a fixed. I think best is clustered in this case as i know i have panel data