I entirely agree with fattdad in that just spring models do not work properly particularly for dishing action, edge or surrounding effects (not to add actual nonlinear or irregular behaviour). I was mostly saying (and implicitly subjecting to critic) the fact that, in effect, a modulus of subgrade reaction is (here at least) being proposed (even by geotecnicals) as what you are expected to be doing ... not even a Young modulus is given, many times.
I also entirely agree that considering (in my view, too much shortsightedly) each footing separatedly os entirely inadequate; I had in Mathcad the opportunity to integrate the effects of foundations for the same set of flexible loads on (floating) piles, grill, footing and mat on elastic halfspace and overall settlement didn't differ but about maybe 10%. Hence, what matters for settlement issued where controlled by an elastic behaviour is Magnitude AND Position of the loads. The only modifier to that is stiffness added by the foundation and superstructure in its interaction with the ground. So if you want any degree of accurate evaluation of settlement you need to go for the whole set of footings or applied loads, never a footing by itself.
Contrarily, this also must expose other blatant misconception on these kind of works, that is: that, except for very good soils, you need to extend the depth of the tests far less for footing works than for mats. Doing so you forfeit the required info for proper evaluation of the settlements, and you will be getting unconservative appraisal of the settlements on dishing action and edge effects. Remember, the pressures exerted by individual footings and piles integrate in the terrain and subject it to one overall average "same" field of new behaviour, hence you getting akin settlements at the interface.
All what above subject and referred to the particularities of the foundation systems themselves, of course.