The first relates to data management. I would like to create categorical
variables by partitioning a continuous variable into two or more equal size
groups, i.e. groups with equal number of observations. In SAS, PROC RANKS
does this quite easily with the "groups" option. Is there a way to do such
a transformation with Stata. "autocode" partitions at pre-set cut-offs but
that would require figuring out the median first and then inserting it as a
cutoff into "autocode."
The second relates to the "cluster" option of the "regress" command. This
produces Huber-White t-stats. A by-product is that the log does not
include an adjusted R-square. Can someone tell me why? Also, is there
anything wrong with obtaining the adjusted R-square by running "regress"
without the "cluster" option and reporting it alongside the Huber-White
t-statistics?
