There is always the possibilty of obvserver bias though. If you able to accurately assign the never-give-ups, middle and slacker groups before the process start

There is always the possibilty of obvserver bias though. If you able to accurately assign the never-give-ups, middle and slacker groups before the process started then the 'attitude => result' case would be pretty watertight. But if you only determine that during their startup progress then it still leaves 'result => attitude' as a possibility. Beyond the obvious metrics, the founders may have a feel for the way things are going that they can't articulate but influences their decisions. You could argue that 100% believing 'attitude => result' is the best way to think even if it is not so true. But it seems to me that each person involved in any endevour needs maintain their own cost/benefit analyses of the situation. If the costs mount and the chances of benefit shrink they are actually doing the right thing by changing path. I can see why this might not make for good coaching advice though.