The Marine Corps is considering non-cognitive testing to identify the best officers. Can this testing achieve its intended result?

I see this as an approach in augmenting predictive indicators in the selections process. As this helps refine the selection process it seems warranted. However as there are many contributing factors not so easily quantified such as charisma I think there is still the need for discourse in determining the best qualified candidates. Also I think there is a big difference in how one performs on a "marshmallow test" and how one performs under fire thus before I could sign off on this I would need to see the math behind any such regression analysis that determines the goodness of fit for intrinsic character traits such as integrity, courage, or loyalty with any probability. Also as their are many variables in ones career how is the performance of selected individuals measured against the selection criteria. With all the nuances it seems to me like you could end up comparing apples to oranges in measuring the results of individuals careers. This approach could also be over complicating the process - just look at Lee and Grant and at who won the war. I also think this approach also has concerns as to pain points this method is trying to address and how it might conflict with the Marines end game of winning wars. While I think good officers win wars the consideration of how this process could negatively impact such ideas needs to be evaluated. Maybe this approach will produce an inflexible and to regimental officer corps - who knows. I think there is also the war factor - maybe due to the nature of war good officers do bad things to either protect their troops or complete the mission thus redefining who they are. Not saying its good or bad just questions in measuring non-cognitive skills with relation to a very complex multiple regression. The guy or gal that can answer all of this in a 100 question "test" - genius.