Hi Statalists,

I have run a lot of regressions in this format:
y1 = a1 + b1x1 + b2x2 + e
y2 = a2 + b1x1 + b2x2 + e
y3 = a3 + b1x1 + b2x2 + e
etc...

And i am interested in the significance levels of all a's. I do have the significance level and standard deviations for each indivdual a, but want to report it like this:
Top 10% of a --> a1-a100 (average)
10%-20% of a --> a101-a200 (average)

How do I test if these clusters of a are significant?

Example:
Decile a
Best 1 0.051
2 0.015
3 0.004
4 (0.002)
5 (0.007)
6 (0.013)
7 (0.019)
8 (0.026)
9 (0.037)
Worst 10 (0.085)