Dragging it back to WW2.
With the lack of navigational aids, the CEP, be it 100m or 400, was frequently dwarfed by the navigational error of not finding the right city. Until there were better navigation aids, there were times when CEP had to include navigation errors of miles.
In the analyses 'gross errors' were usually discounted. How an analysis was done was dependent upon the type of mission and many factors. In the example I gave above (H2X through cloud cover) any bombs that fell more than five miles from the target were ignored. This figure for bombs that were simply ignored is substantial. In the worse case scenario, bombing blind on H2X through 10/10 cloud, 41.5% of bombs fell more than five miles from the aiming point and were discounted.
US analysis was in some ways more detailed than British analysis, principally because bombing in daylight meant that the Americans could often see (and photograph) where the bombs from a Group fell. The British relied largely on individual bombing photographs, and anyway every aircraft carried a bomb aimer (bombardier) who used a bomb sight (no toggling on the leader at night) and every aircraft bombed individually.
The Americans used 'Group Pattern ' analysis in which a single set of error measurements were calculated for the bombs dropped by an entire Group. The object was to establish how well the bombs were aimed. The Americans established a 'pattern center', defined as the centre of a circle with 1000' radius in which most of the bombs fell. It was the displacement of this point from the aiming point which was used to calculate the bombing errors.
Here again many raids gave results which were rejected for analysis. Of 398 VISUAL raids by the 8th AF analysed from strike photographs in 1943, raids in which the famous pickle barrel might have been expected to feature, 113 (nearly 30%) were rejected from the analysis because
"no coherent compact pattern was formed."
The Americans had earlier attempted to use the Mean Point of Impact (MPI) to assess bombing errors. This was a method useful in some instances, like for example fighter bomber attacks on specific targets, but proved unsatisfactory for pattern analysis. The problem was that a relative few (2 or 3 from a Group of 21) stray 'sticks' of bombs falling outside the pattern might drag the MPI to the periphery of the main fall or even outside it.
It took some time for the British to realise that they simply couldn't navigate or bomb accurately by night, a lesson that in slightly different form had to be relearned by the Americans when they discovered that their results over continental Europe fell a long way short of those achieved on the practice ranges of the USA.
Cheers
Steve