I think there is some additional confusion between the order in which various scientific steps are carried out and the logical relationship between the various components.
We talk of observation coming first. But really we mean that the basis upon which all our logic, reasoning, theorizing etc rests is the real world observasions. Since the process is ongoing we don't really do any of this in any order. Testing is done, theoretical adjustments are made and around and around it goes.
When one is going to spend a lot of time and money on making observations you had better pick a good place to look. (like your kitchen). A good theory will help guide one in picking what tests to make. So in carrying out the process and picking a particular test the theory may proceed it.
However, as a logical construct, the theory must be based upon ALL the observations made. In that sense the observations made come before the theory. Even if they were made after the theory was fully formed and the observations only confirmed it.
Good expermintal work using a productive useful theory will be looking for ways to test the theory by seeing if any of it's predictions can be falsified.
The current relatavistic frame dragging experiment is a good example. Eighty years ago the theory was in place (general relativity). Now we have the capability to test it. Will frame dragging be observed or not.
Once the observations are made the resultant (perhaps same theory, perhaps modified or new one) will be based on the observations. Logically the observasions are "first" or take precedence. However, if frame dragging is observed as predicted then by the calender the theory came first and the observations that support it came after.