Although you will still find Scatchard plots frequently in the literature to calculate binding constants, those are not recommended because the linearization introduces distortions in the weights of the values being fitted. It is much more exact to use a non-linear regression analysis of the data.
This is an excellent review about that: P. Thordarson, Chem. Soc. Rev., 2011, 40, 1305.
"Scattered" light when measuring fluorescence consists of several components.
1. Raman scattering from molecules of the solvent (spectrum of solvent can be measured and then subtracted from the fluorescence spectrum of the solution).
2. Rayleigh-scattering of the dissolved molecules (significant only in the field of the exciting light).
3. The light reflected from the walls of the cell, from the cell holder, etc. In normal practice, this kind of scattering dominates (to collect light from the center of the cell, use the crossed polarizers for the excitation light and the fluorescence light, a double monochromator
First of all you have to work with quantum yields due to the inner filter effect. To simplify the equations, the best choice is to excite only ligand. You should do two series of experiments, one versus ligand concentration and second versus protein concentration. The second one is necessary for estimating of Qb (quantum yield of totally bound ligand) from the plot: 1/Qr against 1/P, where Qr is the quantum yield of paricular samples and P is a total concentration of protein in the each sample. It should be a straith line. Extrapolation to infinite concentration of proteins gives you 1/Qb. Then follow the theory. Take a look on my paper: Journal of Fluorescence, Vol. 13, No. 4, July 2003.