DFC
Since the zero-lift angle of attack of a cambered airfoil isn't zero, you can't equate a % change in local AoA into a % change in lift/force.
Plus one would typically have to account for the delta downwash at the tail, due to the presence of the wing, which is going to roughly halve the delta AoA seen by the tail anyway.
As to the original question - why would one design for a different angle of incidence for the mainplane and tailplane...
The main driver, I would suggest, for the wing incidence is to obtain the best compromise between the fuselage deck angle for minimum drag, and the airfoil AoA for best L/D. Constraints on this optimisation would include ensuring that any geometric limits for takeoff/rotation/landing/flare were not penalising. (Barring a Crusader-style wing!)
Having thus arrived at a ideal wing incidence (almost as a \'tail-off\' consideration) the tail incidence is constrained by the desire for a minimum drag configuration at high speed - where we may desire that our mid c.g. cruise weight configuration has faired elevators and the best efficiency for our tail in creating trim lift - and the desire for the most efficient tail for purposes of low speed trim to ensure our field performance is not unduly affected by Vmu or trim limits.
Chances are, we\'re going to end up with a small positive root incidence on the wing, and a near-zero tail incidence.
In considering the effects the \'longitudinal dihedral\' might have on longitudinal S&C, let\'s consider a simplified case : symmetric wing and tail airfoils, and a fixed H-stab. Let\'s assume the wing is set at 2 degrees, and the tail at zero.
In cruise flight, assuming the body AoA is zero, we\'ll have a wing AoA of 2 degrees, a downwash of about 1 degree, and hence a tail AoA of -1 degree, giving us a bit of download for trim. (Which is good, assuming we\'re a nice simple +ve static stability type).
If we introduce a disturbance x to the body AoA, the wing AoA will be disturbed by X also, while the tail AoA will change by X/2, approximately (downwash ratio of 0.5 assumed previously). If our aircraft is stable the pitch axis should restore to the undisturbed condition.
Now, let\'s magically redesign the aircraft, and increase the wing incidence to 4 degs. Since a wing AoA of 2 deg is what we need for trimmed flight, that means we\'ll be at a body AoA of -2 degs for cruise. The tail AoA will be -3 degs now (-2 body -1 downwash) which means that to be in trim we\'ll need some elevator. Lets assume the elevator does indeed get adjusted.
Now, disturb this aircraft by X degrees. Again, delta of X to the wing AoA, delta of X/2 to the tail AoA, which means, all other things being equal, the aerodynamic response will be identical.
The only thing we\'ve changed is that the tail is going to be at -3+X/2, not at -1+X/2. As long as the tail lift curve is linear over a sufficiently large range, and the disturbances are relatively small, there will be no mathematical differences. If the tail is significantly nonlinear, it may matter - but in this case I\'d expect the case where the tail incidence is minimised to be the best case - and that\'s not directly a function of increasing or decreasing long-l dihedral, but comes from a consideration of the trim requirements.
In short, I can see no stability arguments in favour of increased or decreased long-l dihedral; other factors matter more.