First, what did you test, and how? That is, exactly what did you test, and precisely how? Details are incredibly important in something like this. You can see changes in peak frequencies and heights from rather minor changes in the way the piece was supported and activated, and a few centimeters change in microphone position can alter the entire shape of the spectrum, depending on the room. If you are careful to do the tests the same way every time you can get data that can be compared with your own data set, but comparing it with other folk's data can be problematic. I've learned to read my own charts in some respects, and get useful information, but it's not as useful as I'd like. Other people's data tends to confuse me, but then, I'm easily confused.
'Free' plate data is not good at predicting the resonant behavior of the completed instrument in any precise way. As Trevor Gore will tell you, particularly in high performance instruments, small differences can loom large, so you will miss stuff that would be good to know. I believe that the 'free' plate mode shapes have some utility in predicting tone 'quality' in some respects, but I'm not sure how you'd measure that. I could be fooling myself.
The most robust information you're likely to get on a completed instrument is the resonant pitches of the lowest modes. That can give you some idea of the character of the sound, but not much of an indication of how good it is. The exception there is when those modes stack up in such a way as to cause a 'wolf' note, but if there's a wolf you won't always find it in a spectrum chart either.
I would encourage you to keep making measurements, though. Maybe you'll be the one to see the correlation that matters, or figure out the best 'standard' method of making these measurements. If nothing else, you might learn what doesn't work, and that's useful too.