Would it be inappropriate to use ExpoStats (or more specifically, the informed variance prior) for noise dosimetry data, given its based on airborne contaminate data?

If not, would it be reasonable to build a prior in the same way ExpoStats was build, that is with a large number of dosimeter data sets in different environments?

I would appreciate your thoughts
So for basic statistics noise data is usually in dB(A). This means that we can not simply add, subtract or get the AM. eg
|SPL dB(A) 79.7| 86.5| 82.5|Average|82.9|
|SPL log data|93325431|446683592|177827941|Average|83.8|

However, if the Bayesian uses logs then this would rectify the issue, can this be confirmed?

Are there other methods for manipulating data sets for noise?

Does anyone use the following method?
Another option suggested by the AIHA - A Strategy for Assessing and Managing Occupational Exposures 4th Edition - ISBN 978-1-935082-46-0.

Bayesian analyses can prove powerful here as well. Using information gained
in the Quantitative Assessment portion of the analysis, and combining that with
quantitative findings can lead to reliable overall decisions. Bayesian analysis software
has typically been used to address airborne contaminants, which tend to distribute
lognormally. Noise exposures also are lognormally distributed when expressed as a
percent of dose. Given that TWA values are normally distributed, as they are based
on a lograrythimic scale, they must be converted to percent dose before Bayesian
analysis can be performed"

you can put either what you call SPL log data (10^(DB/10)) or percent dose in any lognormal data analysis tools ( IHSTAT, IH Data analyst, Expostats).

Since ISO states that the long term energy average is the risk metric for noise, I would probably use expostats to estimate the AM of either the 10^(DB/10) , and then transform back to dB, or directly the AM of the percent dose.

Warning: AM as used above is artihmetic mean of the lognormal distribution underlying the submitted data (quite different from the sum of the data divided by n, which is the point estimate of the AM for a normal distribution).

I think others have also interpreted noise exposure data similar to airborne chemicals using the 95th percentile.