An interesting thread this.
Just my tuppence worth immediately:- I tend to find the culprit for floaty touchdowns is too fast an idle and/or a propellor with too high a pitch . IN GENERAL.

I hold a full-size pilots licence and I like my models to have some recognisable characteristics of full-size aircraft.
My primary concern is longitudinal stability. The easy test for this is to stall the aircraft at altitude. Does the nose drop as the speed bleeds off so that you have to apply more up-elevator?
If the answer is yes, then the aircraft is logitudinally stable (at low speed). It doesn't have to be much, just some, is all.
This helps during the approach.
Light aircraft technique is to exploit the secondary effects of throttle and elevator by way of longitudinal stability so that ON THE APPROACH, rate of descent is controlled by throttle and airspeed is controlled by elevator (trim). Approach speed is usually 1.6 times the stalling speed or thereabouts, so WELL AWAY FROM STALLING.
Because we don't have instruments and we can only estimate the wind speed relative to the models stalling speed, I like to set up the circuit so that the plane is flying with the throttle set midway and the throttle trim set toward minimum (so that when the throttle is closed, the engine is at slow idle). I then allow the model to stabilise in airspeed by flying it level, using elevator trim. As the model turns base leg, I reduce the throttle by one or two clicks only and allow the model to descend, maintaining the same attitude. This equates to the model having the same airspeed. Because the model is in trim, only gentle aileron input usually is required just to line the model up and trickle down the approach. Because the throttle trim is already set to low, as the model is above the threshold, I close the throttle and flair simultaneously. The act of raising the nose washes off speed and closing the throttle to slow idle at the same time keeps the descent going, so the model usually just settles straight on.
Things to avoid when using this technique are premature flairing and general pitch changes during the approach, usually diving for the threshold "trying to avoid a stall". this just results in your airspeed being all wrong just at the point when it should be "just right". Trainers will approach at about a fast jogging speed, so any headwind will make this look very slow.
In preference, I like to stand close to the touchdown point during the approach, so that I can get a good idea of the model's direction and descent path (like, "is it heading for my feet?") but many fields won't permit this, herding us together into flightlines or pilot pounds so that we don't hit each other.

Now, this all doesn't work if the model is neutally stable in pitch, because all that happens when you close the throttle is that the model continues on its way, just slowing down until it stalls, so it has to be controlled in attitude constantly. Consequently there's a tendecy to apply unintentional aileron input when controlling the elevator during the approach and it results in less accuracy in the touchdown phase.