We know we can't expect hard data, like scientist would want. But, we have to do with what we have. As Stanton Freedeman said, we are not talking about hard evidence, but an overwhelming proof.
If we accept that data is not perfect and use statistics, than there is actually an incredible amount of useful physical data out about UFOs.
My hypothesis is that an average witness does not understand engineering and physics well enough to be able to fake his report. So if an non-expert witness is correctly describing a little known physical process, his testimony might get, say, 70% relevance score, as opposed to, say 10% relevance. And when you have 3-4 witnesses observing the same physical effect, than its pretty safe to take that effect seriously.
A good example would be that some UFOs that landed, left inundations in soil. And from these inundations their wight can be calculated. Like in Zamora case, where several investigators who debriefed him, prised Zamora's character. In that case inundations were left in soil. Similar inundations were left in the Raldensham Forest incident, although craft was much 3-4 times smaller by its diameter. Cases with inundations in the soil go into dozens and from them we know that a "specific weight" of an average UFO is about twice as large as for an aeroplane that would fit into the same circle.
From their heavy weight, and not just weight, we can work out that UFOs are not using aerodynamic effects to fly, like aeroplanes do. This analysis might sound trivial, but it takes us one step further from total ignorance.
There are some physical aspects of UFOs that are not widely known to general public through media. These little known clues than can be used as a keys to prove or falsify if witness is genuine. Ray Stamford and Derrel Sims both use this technique to protect their valuable time from witnesses that can't contribute quality reports.