Arriving at a hypothesis happens one step at a time.
Part of our work is constructing and refining the cladogram. Lots of new theropods, including early birds and other feathered dinosaurs, have been found in recent years. Every time one is discovered, there's an extensive analysis of its anatomy and the people who are studying it analyze all the evidence to see where it fits in the cladogram. So we are continually refining and testing hypotheses of relationships. This work is done by many different people so the analysis is repeated, just like an experiment.

New information comes to bear on our ideas all of the time.
Many new specimens give us valuable information about the movements that were possible in the arms and legs of these ancient animals. This helps us to test ideas about mobility and the ability to perform movements like the flight stroke. We have a pretty good idea of where in dinosaurian history a stroke very much like the flight stroke evolved.

Sometimes, we have to re-evaluate our ideas about posture, or about modes of walking, or about how fast an animal could travel or move its arms and legs. That's not unusual. New information from bones, soft parts, trackways, and other fossil evidence comes in all the time. They all contribute to the bigger picture.

Comparing footprints and skeletal structures of the of the living bird Rhea americana (A, B and C) with those of a theropod dinosaur (D, E) provides clues about the structure and function of the dinosaur's limbs.

Being a paleontologist requires knowledge of many fields.
People often ask what you have to study in order to be a paleontologist. I guess the answer is everything! Sometimes chemistry or physics is really useful; sometimes biomechanics or ecology or sedimentology or histology you just never know. The great thing about our field is that if you run across a problem that you can't solve, there are lots of experts you can work with who have the know-how in that field.