I wouldn't say that it skews the data. Rather, it is an important element of it. I do account for runners with multiple BQ's by eliminating the slower duplicates from the final dataset before doing the Monte Carlo simulation. The BQ you used for registration effectively increases the utilization rate for that qualifier and the BQ that you didn't use lowers the utilization rate for that one.

It is an important consideration because what falls out from it - repeated across all runners with multiple BQ's - are "faster" and "slower" qualifiers with different utilization rates.

The later BQ you use to improve your seed time is irrelevant to my calculations for '25 because it is done after the cutoff time has been established and you have already gained entry.

However, if I decide to do this again next year, that later BQ that you used for improving your '25 seed, your presumed '25 BQ at Boston, and any other BQ's you achieve will be compared for estimating the '26 cutoff time and the fastest of them will enter the final dataset.