Besides, testing their theories in this way, theoretically even if they are half-right, the AI system created could find where the psychology theories lack, and they do, I am convinced. I had contemplated a way to tag each word of a sentence spoken and captured by voice recognition with an emotional value based on tone, abruptness, volume, and inflection, and pitted against the normal conversation cadence, and average of the rest of the discussion. This could be used by intelligence to determine levels of chatter, or which conversations to record, and even predict future actions by that individual, based on probability of the overall society from which they were immersed.
A conversation, even perhaps one online could also give clues, or perhaps following a conversation on an Internet Forum, blog, Facebook page, etc. The only problem with this concept is that it would lead to unfortunate false positives, but if the Telephone Number List AI system learned in hindsight, then over time it could get rather good at the probability. In fact, using this system, we could also find the best leaders for various tasks - Warrior Squad Leader, Fighter Pilot, Administrator, General Patton type, and the best type of leader could be put into their best suited roles for the best success outcome.
No, it would never be absolute, but it would be better than experiencing leadership failure so often, which damages the group. i.e - "boss yells at worker, worker goes home yells at mom, mom yells at boy, boy kicks dog, dog chases cat, cat claws sofa" - as someone recently put it. Personally, to me it would appear that if each individual understood the use of emotion better, they might do better (and per my theory, a society does better when each individual does better), but again that might be asking too much as per my; "they are just humans - what can we really expect from them" theory.