1.
Obtain
more and better data.
2.
Use
a data reduction technique like factor analysis to create a new hypothetical
set of linearly unrelated variables (factor scores).
3.
Transform
two (or more) collinear variables by dividing (multiplying) the two and using
the newly created variable in place of the two collinear ones.