Two unknowns used together randomly will always be an variable unknown. If you are searching for truth, you need to account for and test every thing that can affect the results, and keep track of it all. Sometimes errors can work for you rather than against you. If your spindle is out .001" and the mounted tool is also out .001", you can mark the high spots and then place them opposite each other to cancel the error. Lack of keeping track of things potentially causes larger errors, while keeping track of them helps to eliminate errors. "Faith" is not a word used by true metrologists and fussy machinists. In general, it is assumed that errors stack up. Simplicity is bliss...
Edit: By far, most work simply does not matter very much. Close enough is close enough. When testing your new lathe to see if it meets specifications, now we are talking important. If you don't, you will never really know what you have there. It is a good plan for you to test it.