My initial comment would be that the room has to be quiet and controlled enough for this determination. A noise floor above ~35 dB SPL will mask the subtle second-order differences between cables or tubes. If the room isn't reasonably treated, you may be hearing the room when you think you're hearing a tube swap. In other words, the room's contribution is constant, so it becomes perceptually invisible even when it's dominating.
All that said, given a decently controlled room, I think that tubes (especially input and driver stages) will generally move the needle more than incremental cable upgrades past a modest baseline.

