This is unusual. Is there an injury history or something else going on? If so, you might consider upgrading to the dual sided vectors. Otherwise, it’s hard to say either one is accurate. You shouldn’t mess with scaling factors for things like a 52/48 imbalance, and if you have something more substantial there’s probably a history behind it. Any change you made to the scaling factor could render the numbers you get less meaningful. That’s a red flag to me in terms of power meter accuracy.
There are a lot of variables that go into wheel-on power measurement, and I agree w/ @tnordberg that the direct force power meter (Vectors) is usually going to be accurate. But if you’ve messed with scale factors, then the real answer is practically impossible to know.
My advice would be to use the Vectors as you sole source of power measurement (using TR’s powermatch) since there are so many sources of error involved here. Whether that measurement is accurate or not matters less than its consistency/precision.