Dear Statalists,

i have done an Event Study as described by Princeton DSS (
HTML Code:
https://dss.princeton.edu/online_help/stats_packages/stata/eventstudy.html
) . Different from the DSS example i have CAR(s) for various event windows [-2;2] [-1;1] and [0].

When looking at the t-statistic results i see the most of my events are not significantly different from 0 for any level from 10%-1%. I am new to statistics and therefore wonder. Before continuing with creating my regession of CARs onto other independent variables, will i need to to 'drop' all of the insignificant events?

When i use the
Code:
reg CAR_MM2 if dif==0,robust
it shows me that my overall sample of events is significantly different from 0.

How do i go from here?


More infos about my sample:
- The deals are inbetween 2002 and 2016 and multiple firms have multiple events (however i believe the DSS instructions dealt with this problem)
- Should i do any 'non-parametric' tests, such as Wilcox or Friedman? I am not sure these kind of tests apply here.
- I do not have CRSP / U.S. data and therefor cannot use Eventus, or any other direct eventstudy command (estudy, eventstudy2) to do my analysis, many of which include these kind of robustness tests


Please excuse if there are still open points / unclarities. I am happy to provide more informatino as well as a sample of the data. Since this seems more as a statistical reasoning rather than a 'how to compute in stata' problem i did not add it yet.

Thank you!!

Best regards,
Max