This post on DPReview shows the difference in the raw capture with HTP on and HTP off. The histograms you see are from the actual raw data captured, before the camera got its mits on it and faked the "JPEG" histogram you see in the camera.
http://forums.dpreview.com/forums/read.asp?forum=1019&message=24943011
Using HTP when shooting raw is, in my opinion, not a smart thing to do, especially if it is your aim to "expose to the right". You will be underexposing the raw capture by 1 stop and the camera's internal processing will conceal that fact from you when it generates the internal preview image from which the camera's own histogram is created.
There may conceivably be a
tiny benefit to HTP with raw, if you use
DPP to process your raw files, but otherwise no benefit whatsoever. AFAIK DPP is the only software that actually understands what to do with files captured with HTP on. Other software, like Lightroom, just applies pretty dumb processing to "mend" the underexposed file. Since I only shoot raw I never use HTP, ever. As I use Lightroom for raw processing I have even less reason to use HTP.
If I shot to JPEG I might use HTP, but only in conditions with troublesome dynamic range and important details in the highlights and shadows. In flat lighting (overcast day), or low contrast scenes - e.g. a mid toned bird against blue sky or foliage - then turn it off.
To prove this for yourself shoot a scene in raw with identical exposure settings under identical lighting, once with HTP on, once with HTP off. Then use the free software available here -
http://www.cryptobola.com/PhotoBola/Rawnalyze.htm - to look at the true raw histograms.
Here is my own example of the raw histograms, one without HTP and one with HTP. All exposure settings were supposedly identical - 200 ISO, f/5.6, 1/200. Exposure was set to maximise my ETTR data capture, without clipping. The first raw histogram shows that I achieved my goal. The camera was on a tripod and the images were shot 5 seconds apart - just long enough to switch HTP on and fire again. If you look at the data reported by Rawnalyze you will see that for the shot with HTP enabled the ISO is shown as 200/100, meaning that the camera "pretended" to be shooting at 200 ISO but was "secretly" only shooting at 100 ISO, to effect a 1 stop underexposure.
I've also attached the two files as converted by DPP with no edits, Neutral picture style, 0 sharpening, just as shot. You can see that the brightness of the sky has been toned down, while the rest of the scene seems barely changed. HTP is doing its job here, but we do have a slightly underexposed file - i.e. not an optimum ETTR capture.