Hi everyone,

my data contains information of approx. 500 firms over a 7-year-period (quarterly observations). I am using a generalized difference in difference approach to measure how specific events influence median firm ratings. I created a dummy variable that indicates whether an observation is treated or not treated. Also, I use time and firm fixed effects and I include control variables which control for time-variant characteristics of the firms. My model looks like this:

Median Ratingit = a + 𝛽1Event dummyit + ∑𝛽kControls + timeFE + firmFE + 𝜖it

I am now considering weighting the quarterly median ratings by the rating count that was used to generate the median ratings since the observations are more reliable if there were more initial ratings. I found following information on -aweight- (Stata 16 help for weight):
Typically, the observations represent averages and the weights are the number of elements that gave rise to the average.
I have two questions:
1) Can I use this code even though I use the median rating instead of an average?
2) Can I use this code if only the dependent variable represents a median and the rest of my variables do not?

I hope the setting is understandable - thanks in advance for any help.
Sam