And if you have a way of semi-realistically modeling the "chicken-factor" I'll gladly redo the whole tests under the core rules with this factor included.
You're right, it depends on the player--whether he's feeling "risky", whether he's become attached to his character through chargen, whether he'd rather have this character die so he can re-roll another one.
It's very subjective and hard to quantify.
As you can see, without the "chicken factor", enforcing the vanilla KIA rule actually increases the average number of skills as well as generation time, as this eliminates the large number of 20-years-old half-term characters with a skill or two.
That's because those who make it have survived and therefore have more skills. I don't think that accurately represents real characters used in a game, though, when the GM is enforcing the original Survival rule. As you say, it doesn't consider the "chicken factor"--when the player decides to end chargen because he thinks he's ahead of the game (or for some other reason--like he's attached to his character).
It does have a very real effect on the outcome, though.
If it were me, I'd simulate the chicken factor by simply rolling the 24 characters you rolled up for this excercise with the belief that each character is
your character.
If he's a good character, then don't be risky with him. End chargen. If he's not so good, of if you're feeling lucky, then go for it.
Just be consistent and true to the action of rolling up the character--that's what will return the best results in this analysis.
You could even enlist some friends so that a different sensibility other than your own was considered. Have every other gamer you know create two characters with you?
I could think of some ways to do this statistically, but I think the approach above best simulates what we're trying to achieve.