Of course any Oops are deleted, but also you need to delete runs that are outliers in the data. These are times which are too fast or too slow.

If your using a monte carlo simulator to crunch data using the Std dev. be aware of a problem that most off the shelf stuff doesn't handle, say for example i have a dog with a time of 5.058 and a average std dev of .087 thats a normal std dev for a split at the meadows at some point in the simulations there the simulator will spit out times like 4.85 etc which we know is quicker then the track record. so you need to set minimum limits .
Hope that makes some sense