Dare I suggest that you do some simple arithmetic? The whole genome is about 97-98% what was called 'junk DNA', and if about 4% of the non-coding DNA differs between humans and chimps, then you can work it out. That's where I got my 5-6% from.
The entire genome of humans and chimps includes the so-called 'junk DNA'. The junk DNA is the vast majority of the genome. Hence, it is impossible for the junk DNA of chimps and humans to be so different that they are not even comparable. I googled your phrase, and only found it on creationist websites. I had at some of them but couldn't find any links to papers backing up their claims.
One difficulty in comparing chimp and human DNA is this. If you have a gene that has been moved, or perhaps a deletion, how you count difference. E.g. if we have the sequences:
AGCCTAGACTCAGGGCAT
AGCCTAGACTCAGGGCAT
Then, they are identical. If we hae a base substitution mutation, then we may get:
AGCCTAGACTCAGGGCAT
AGCCTAGACTCA
CGGCAT
Clearly, the two sequences now only differ in one place. If we have a deletion mutation however, we might get:
AGC
CTAGACTCAGGGCAT
AGCTAGACTCAGGGCAT
If you then start at the left and count base by base, they appear to be very different. A=A, G=G, C=C, C<>T, T<>A, A<>G, and so on. But, we know that the only difference is the deletion of the second C. So, is it then similar? But, if this is a coding gene, then because it's three bases per amino acid, it may now code for a number of different amino acids, and the protein it codes for may be very different.
So, a bit like accountancy, it may be that you can use the measure that gives you the answer that you want.
In terms of measuring evolutionary differences and divergence, it's the number of mutations that is important. And hence the deletion example above should be seen as one base different. Some of these mutations may be whole genes moving around (including inactive genes in 'junk DNA'), so that makes it harder to work out the true genetic difference. It's not a simple counting exercise, it's a matter of working out the mutations that occurred from the differences between the two genomes. E.g. if we produce the mutation:
AGCCTAGACTCAGGGCAT
GACTCAGGGCAT
AGCCTA
Then there has been a single change, but again the bases now count differently from the left.
If your intention is to measure them to be as different as possible to claim that chimps and humans are not closely related, e.g. if you are writing for 'Answers in Genesis' then counting base by base will give you the answer that you want. But, that would, I believe, be intellectual dishonesty and open you up to criticism.
The current scientific understanding is that we are finding out that there is biological function in at least some junk DNA. The fact that some (but not all) junk DNA is conserved over very long time periods means that it must have some function important for the organism. (By conserved I mean that mutations in these regions of the junk DNA are selected against so that the mutation rate found is lower than would occur if mutations were meaningless.)
http://www.news-medical.net/health/Functions-of-Junk-DNA.aspx There is research that suggsts what these functions might be. E.g. the link in the very first post of this thread.