Ixperiencitness experiments are experiments determining what conditions produce what ixperiencitnesses. Because the ixperiencitness that a particular itobody produces is not usually visible to other itobodies there has to be tools that are used and can be trusted to give relevant information for determining or helping to determine this. Science is full of such tool like telescopes, microscopes etc the information from each tool has to be understood to be used in the correct way.
If you know that a specific grouping of structures and functionings produce a specific ixperiencitness any device that can determine these structures and functionings can be used as a tool to see the ixperiencitness in an itobody where it in fact does not actually sees the ixperiencitness in the itobody. This type of indirect observation occurs all the time in scientific observation and experimentation. This process is successful because the scientific structure that supports it is accepted by most of the scientists. It is accepted by most of the scientists because it forms a coherent scientific structure or theory which is based on experimentation.
When you look up at the stars at night you do not actually see stars, planets, galaxies, and photons, you think that you are looking at these things because of our current scientific theories that you have learned about over your life time. Young children, people in the past, and people in the present that do not know about science do not see these things in the sky. They have their own theory or just see dots of light in the sky (if they have the tool (working eyes) to see with).
Some thinkers have argued that you cannot trust experiments that are based on anything other than ones own senses. The problem is that we cannot always trust own senses. That is why science is based on many people doing many experiments as often as necessary which is as often as desired to convince someone. We need theory to explain the information that we get from the senses.
When the theory has anomalies, inconsistencies etc. with the experimental data there may be problems with the experiment or with the theory (or even the experimenter when he is faking the data or incompetent).
A rudimentary theory has to be created so that experiments can be designed to test the theory and to variations of this theory.
The best plan is to start with the simplest theory possible and then modify it as necessary when inconsistencies are found