Short: Yes.

Long: The usual SD of measured thresholds is around 6 dB (see http://www.ncbi.nlm.nih.gov/pubmed/24165302), which, assuming a normal distribution, indicates that 95% of the population is within 12 dB, and 99.7% within 18 dB. Thus, it is possible to measure -15 dB and interpret it the way you do.

In general, thresholds depend on the methodology, frequency, reference for the standard, the subject's state (e.g., age). And on the quality of calibration: Sometimes it's just the poorly calibrated system. If you consistently obtain such low thresholds, you might want to re-check your system.

Of my 16 years daily experience in ENT field since my MSc graduate in 1997 passing to my french DIS d'ORL graduate in 2003 uptill now, interpreting hundreds of audiograms to assess hearing even in normal cases for example persons seeking this hearing assessment as a document to catch a job, i saw few normal audiograms with normal hearing threshold at -10dB.

I think that the known variability of normal hearing thresholds -15dB to +15dB/+20dB (child/adult) need a recheck because it make a big confusion to some normal person. One-day, I met a gentleman 26 year old & came to ENT clinic complaining of decrease hearing, noted by himself after catching a job working with peers in an office. I asked him if there is a noise or a trauma exposure in his work. He denied all. Even there is no family history. ENT examination was within normal concentrating on ears (normal pinna, normal EEC & normal TM). I have done to him tuning fork hearing test which was totally normal (bilateral Rinne normal positive and central Weber). I tried to assure him but he still informed that he has hearing impairment & his proof was that his peers in work hear better than him. So to confirm my diagnosis that his hearing is normal, i have done to him pure tone audiometry (PTA), tympanometry and SRT/SDS which were bilaterally normal. Despite the audiogram was normal but the normal hearing threshold was at +20dB in both ears i.e. borderline normal & i explained that to him & that there is variable hearing normality like him & his peers in work but despite my explanation he pleases me to prescribe hearing aids to him (but i can not) to be able to hear as his colleagues in work who i think they have more accurate normal hearing than him. I followed this person few months later with another PTA which was also the same within normal. So, i advise rechecking this dilemma of normal range of hearing which can not resembling normal range of hemoglobin which can not be felt by persons but the ranging normal hearing thresholds between -15dB and +20dB can be detected by the person as in our above mentioned example. I think it is better to classify normal hearing range into 3 categories excellent hearing (-15dB to -5dB), very good hearing (-6dB to +5dB) & good hearing (+6dB to +20dB). I know that audigram is a quantitative measurement. I know also that SDS is test of quality but can not differentiate between my above suggested 3 normal categories because SDS depends of SRT of each category. So, my question can we differentiate between the 3 categories by fixing SRT of excellent group for all groups during applying SDS?

Also, sometime i met a patient with a unilateral hearing loss confirmed by audiogram with one ear hearing threshold at +20dB in the side of hearing loss & other normal hearing threshold at -5dB. No body says that this example is normal variability of hearing in the same patient but it is a real hearing loss & i considered that his base hearing was -5dB, so he has 25dB hearing loss in the diseased ear and i diagnosed it as mild hearing loss & i assessed him as usual. For my other ENT attendant in my previous answer i followed him few months later with new PTA to detect more improvement or deterioration of his hearing because there was a very rare possibility that he had a bilateral hearing impairment till +20dB above his base threshold which may be more accurate than this but fortunately his history & last PTA denied that.

Madjdak said everything. His answer is complete and I agree with him.

Yes, it is Possible.

0 dB isnt absolute zero.

For example:

10 dB =1*10-11W/m2

0 dB =1*10-12W/m2

-10dB = 1*10-13W/m2

etc...