The independence assumption for Rojo’s scenario is reasonable as one runner having a good race doesn’t (shouldn’t) impact another runners performance being good or bad (i.e. blowing up or not). However, assuming independence when the probability is defined as making the team doesn’t work. As it doesn’t realistically model the circumstances mathematically.
Assume a race of N total runners. Each runner has a probability density function assigning a probability to each of the potential N finishing places. The probability can’t be zero for any Nth place and the probabilities have to sum to 1 for each runner. Additionally though, the probabilities for each finishing place, 1 through N have to sum to 1 over all runners (assume no DNF’s).
Under these conditions, if 4 runners have a 75% chance of making the team, that’s because they each have a 25% probability of finishing 1st, 2nd, 3rd, or 4th. This is a uniform distribution btw, and in general scales based on the number of entrants, where to maintain uniformity across the N finishing places and participants, the probability assigned to each of the N participants finishing in places 1 through N would result in a probability of (N-1)!/N!.
The distribution does not need to be uniform though. In fact you’d likely have severely right skewed pdf’s for the favorites and severely left skewed pdf’s for the runners who just dipped below the qualifying standard.