This subject came up in the recent thread about the church with a poor EICR. It was suggested that inspecting and testing ¼ of the installation every 4th year might be preferable to testing 100% every 4 or 5 years. I suggest that this is a misunderstanding of the principal of sampling.
The principle of sampling is that if inspection or testing of 20% of an installation reveals no defects, it may be inferred that the other 80% is in a similar condition. A statistician could calculate the degree of confidence of the assumption. The sample must be identified in advance and may be selected randomly, or systematically - say every 5th circuit, or every 5th socket. What is not permissible is to sample a sample, so it would be either 100% of sockets on every 5th circuit, or 20% of sockets on all circuits. If the sample has not been selected randomly, the same one should not be used repeatedly - so circuits 1, 6, 11, etc. on one occasion and 2, 7, 12, etc. on the next. Self evidently, the method of sample selection must be recorded on the EICR so that the next inspector may select a different sample.
If the sampled circuits (or sockets, or what have you) give cause for concern, the sample size may be increased: the purpose of this is to increase the level of confidence, not to find more defects.
If the sampled inspection and testing is satisfactory, the EICR is signed off and the normal inspection interval is recommended. It should not be reduced on account of the sampling. To test 20% every year is not sampling because in fact the whole installation is tested every 5 years and some of it, e.g. the origin and main panel, ends up being tested 5 times in 5 years.
Now I stand by to be shot down in flames!
