You never know all the variables that will cross paths.What you are learning for that rifle may not apply to another rifle.Still,you are a handloader seeking improvement,that is cool.
Wild guess,the full length resize is giving the clearance for your round to find its own center as it expands.Perhaps thetre is a touch of roundness or concentricity error in the chamber.A randomly indexed neck sized round may be placed in a bind,forcing the bullet to not enter the leade square and straight.
On partial sizing,1)check the case body taper on a cartridge drawing.Determine case taper per inch,included angle.
Now,how much do you back off the die? .050?Suppose total case taper is .010 per in,and you back off 1/20 of an in,the diametral change in the amount you size the brass would be .0005.Now,if you only back off .010 to not bump the shoulder,the diametral change is only .0001..Make a mark on a piece of flat ground steel with a sharpie and run a .0001 dial indicator over itto see how small .0001 is.
Point:Partial sizing has very little diametral effect on the brass sizing operation unless you have extreme case taper.
Partial sizing 2)Take a peeled bananna,put it in your fist.Squeeze.What happens to the bananna?So,sizing down the dia of a case without bumping the shoulder,what happens?The shoulder moves forward.IMO,it is best the shoulder gets some (just enough) bump to establish uniform,correct head clearance .That is my ideal for full length re-sizing
I can see no theoretical advantage to partial sizing,where the shoulder is uncontrolled.The diameter of the brass is sized the same for practical purposes,only the shoulder bump is omitted.
Some neck dies are "neck and bump " dies,which do restore the shoulder dim,but do not alter case body dia.