The point of the lower test current is to enable RCD protected circuits to be tested without tripping them, unfortunatly it leads to values which don't match the real world as closely as the 25A test for a few reasons
1) you are effectivly taking a measurment and scaling it up (ok this current drops the voltage by x ... if I dropped it all the way down to zero (direct short between phobe points)... what current would I have?... kinda thing) 25A is much closer to real fault currents than 15ma so its more accurate, of course it'd be better to use the full fault current for the perfect reading, but that would be inconvenient
2) A higher current might clear a very poor connection that a lower one wouldn't, so your 15ma test says earth ok, the 25A test blows out the strand of wirewool laying across the bad connection. The reverse is true also, 15ma test indicates poor earth, 25A overcomes the oxididised joint. This sort of thing is more pronounced the older the installation
Use 25A test on none RCD circuits, use 15ma test on RCD circuits, the Z's arn't as critical there as they are without the RCD, and if the real fault current blows the earth clear, then hopefully the RCD should have been triggered (though when you reset it you might not have an earth, though an RCD does go someway to making a lack of earth situation a little safer)
(just incase you are not aware an ELFI tester works by measuring voltage, applying a current, and measuring voltage again, using the volt drop obtained and the test current and popping them into ohms law gives you supply impedance, and if you want PFC, then ohms low again with the supply voltage and the newly found Z's will give you that)
