The above diagram shows our first guess at where the shared region is. But this is a mistake, for the Type 2 coincident boundary cannot be a lower limit of the true shared region. This Type 2 boundary coincidence occurs because Jeannine's segment here goes no lower, yet she appears to the left of this in the significant Type 5 coincidence. The true shared region almost certainly has its lower limit set by this remarkable Type 5 coincident match, and - again almost certainly - the upper limit of the true shared region by the equally unlikely Type 3 coincident segment boundary. Thus:

While we have a larger yet more probable shared region, we lose all but 8 matches, and we lose 3 of the people - Bob, Ed and of course MD was absent anyway. That is, there are 9 out of 12 people in this higher quality result. These absences are not critical from the genealogical perspective - the presence of JG, Allan and J from the Barrow line compensates for the absence of Ed; the absence of Bob is compensated by the presence of Jeannine and TPhelan from the USA Lousada line, while the absence of MD is compensated by the presence of Jeremy, J and Allan in the UK Lousada line. We have thus, via our judicious inclusion of Precision = 3 data, arrived at a richer analysis than by using only Precision = 7 data!