Old post about REM limitations, still relevant though:
If you actually wanted to challenge the veracity of REM measurements, there's a few ways you can make a good case. Namely:
1 The original Audiogram: how accurate is it (and therefore your subsequent target)? Threshold measures are repeatable for some people, but not for everybody. Anyone on here who regularly tests hearing will know that there are patients who 'wander'.
You have also got to consider whether the test was done in noise, with ear-tones or cans, how long ago etc.
2 The probe tube - has to be placed in the last 6mm of the ear canal - not impossible, but it's one of those jobs where an extra hand would be useful, especially if you have a programming lead on the aid. Due consideration for the condition of the ear canal, whether the patient has wax, a cold, a middle ear infection.
3 Target and stimulus tone. People swear by automated speech weighted noise and pseudo random noise. They will also say that a NAL target is better than a DSL. Personally I have massive reservations about using ANY form of stimulus that isn't real speech. Hearing aids are inherently designed to filter noise from within speech channels: if you reproduce a signal or modify it, it contains noise - the aid WILL turn itself down. The best results will always come from straight speech input: not everybody uses this.
4 Loudness experience and patient expectation. Some patients just don't like where the prescription sits, even when you've successfully negotiated all the above. Dealing with this can be difficult: the patient may feel left out of the programming loop or they may just want a different level of loudness than you've arrived at. You have to bear this in mind and soften or louden the settings accordingly. However, it is usually possible to wean them to nearer their prescription over several visits. Other patients sit miles away from their target and are quite happy.
5 Temporal aspects within signals that you can't 'see'. Different manufacturers use different attack and release times as part of their processing strategies, they will make the aid sound different or unnatural, BUT you can't necessarily visualise the response - even on the 3D systems it's not that clear.