Sure, you can do a transfer function of slid (did you mean skid?) compared to pump housing. Coherence is a measure of causality between input and output. Let's say you measure the COH = ~1 between the ?skid? and pump. In this case, the relationship between the two are the same but not sure what useful information that gives me or whether it helps me figure out the problem.
As Gary stated, it is used primarily in impact/modal testing to provide confidence in the cause/effect (e.g. impact to response) relationship.
For the case you mentioned (large difference between skid and pump), might imply looseness in bolting.
I might be wrong in my reasoning, others can elaborate.