I paid $100 for a Sinclair concentricity gauge, and in the box was a little piece of paper like from a fortune cookie that said, "You will probably find that the problem is the expander ball."
I now know that the problem is using the expander ball AND the sizer in the same step. The expander ball goes in easy, but while it is in, the sizer makes the neck smaller. To get out the expander ball must pull through the neck. It the cartridge is being held by a conventional case holder that is asymmetrical, the neck is going to get bent. Not enough to see, but it can be measured. Probably another .004" of runout.
If one takes the expander ball out, sizes the brass, puts the expander ball back in, and expands the neck while pushing, only about .001" of run out is caused in that step.
Curiously, almost nothing can be done about crooked necks, but fire forming to a concentric chamber.
I have 6,000 rounds of .223 brass that my .245" bushing in my Redding S die gave .001" of additional runout. The only practical thing I can do is shoot it to fix it.
If I open the neck way big with a mandrel or expander ball, the sizing die will not straighten it, becuase before the die gets contact with the body, it has already been following and sizing the crooked neck.
What does it all mean?
Don't use the expander ball.
Use a decapping die to get the spent primer out.
1) In benchrest, the chamber neck is so tight, the brass does not reach the elastic limit, and so the neck is not resized.
2) In varmint class accuracy, if the chamber neck is as small as possible without necessitating neck turning, then it can be sized with an S die and bushing or with a Forster factory honed out die. If the chamber is SAAMI sized, use the Forster factory honed.
3) In deer hunting accuracy, use the Lee really good buy $10 dies with the expander ball in place.