Observational data trumps experimental data
trumps model data is a generalization rightly pointed out by Matt, in the sense of root cause analysis.
Not exact matches
And there I say that the mark in stone or the carbon - dated tree stump from under the glacier
trumps the
model simulation every time, because these are empirical
data while the
model simulation is not.
In the real world
data,
trumps models, especially those with GIA adjustments.
Observations
trump models every time, no matter how much you try and manipulate the raw
data.