If you take the various derivatives and their contributions to things like the frequency and damping of the various aircraft 'modes' - Dutch Roll, Phugoid, etc., you'll find that with some small simplifications you can characterise most of the characteristics with relatively simple expressions (sometimes ratios, or difference, or whatever) of the derivatives.
Experience and HQ research shows that certain characteristics are favourable - much of the MIL-F-8785 requirements comes from that kind of background. So with a target range for, say, Dutch Roll Frequency, and knowing what derivatives dominate this parameter, I can derive desirable values (or ranges of values) for the derivatives.
The complication is that it's rare to be able to satisfy all the design requirements at once, even if all you consider is HQ, and if the context of the whole design is considered it becomes as always a matter of compromises.
For the specific of dihedral/anhedral, it can be shown that the ratio between dihedral effect Cl-beta and directional stability Cn-beta is important in determining both Dutch Roll and Spiral stability; depending on which is more important and on the directional stability, there will be an optimal value of Cl-beta, which may influence the choice of geometric dihedral/anhedral.