I recently heard it claimed that objectivity in science requires direct measurement of all variables of interest, because we can only validate an indirect measure, or estimate, of some variable by comparing it to direct measurement; without such calibration, a proxy measure is empirically meaningless. I think the latter claim is false (and hence, so is the first). Suppose that in some empirical situation we have reason to believe there is a natural quantity X, the value of which we cannot measured directly, but we have some proxy measure Y which we hypothesise is proportional to X. We also find that the value of Y correlate with directly measurable variables A, B and C, which are believed, on sound theoretical grounds (i.e., parsimoniously consistent with all relevant evidence), to be determined by X. I think this would justify taking Y to be a useful estimate of X, absent evidence to the contrary. I suspect that there are many examples of this kind of reasoning in many different areas of science.
I...