Your explanation of noise isolation test is confusing. Is the test done by playing a sample noise from a speaker so that SPL at the ears of the dummy head is 100 dB? If so, it is not a realistic test. 100 dB SPL are not encountered that often in daily life. May be 85 or 90 dB inside subway trains and 80 dB while out on a busy street.

A better sample noise would be one where SPL of traffic, train and human speech are equalized so that they match average real life levels at the dummy head ears. For example, human speech at 70 dB, traffic at 80 dB and train sound at 90 dB.