I use a Roland Quad Capture for inputs.  The Quad has a built in compressor/limiter/gate.  I have no issue with setting the input sensitivity for myself as I believe I have good mike technique and once set I do not clip my input. Therefore, all of my compression is done post recording.   However, I have started recording less experienced vocalists, and when I set the sensitivity on a “typical” vocal intensity, I start recording and then they strike a sudden plosive or high note that makes my meters go red and results in a clip.  So I set the sensitivity lower and re-record to get a take that does not clip, but also has recording levels below -12db.
 
It was suggested that I use the Quad’s limiter to limit the peaks of the vocal input.  This rubs against the “rule” of applying compression post-record, i.e., in the mix.  What are your thoughts on the issue?

I like to limit some before and some after. I also set up 2 vocals tracks for the same input. One running hot and one backed off about -10db. That way if there are any clipped peaks, I could fill those in with those peaks from the quieter track that didnt clip.

I wasn't impressed by the Roland limiter when I had the VS-series in.  I found them hard to set and got some "chatter." Maybe I didn't try hard enough or maybe they got better.
 
But I almost always use a comp/limiter on mic'ed stuff.  It makes for a nice, controlled signal and you're right, it makes it much easier during the mixdown.  I use nice analog hardware comp going in (Warm Audio 76 or an Rupert Neve Design) for control and tone, and the ITB comps just to ride the volume.
 
@

There's no rule about compression on the way in. In fact lots of pro engineers will use dynamics processing on the way in and already commit to the sound they want then. I wouldn't use the. Compressor in the quad either.

You would be better off using a quality compressor/limiter on the way in patched between the Mic Pre and the audio interface.
 
It should be set for limiting  (not compression)  and the threshold set quite high so for a lot of the performance it basically does nothing much and lets everything through as normal. However a loud passage or phrase will catch the limiter and the limiter will just limit that loud section to a predetermined value.  If set well it should not clip either or even be obvious.
 
Many people think limiters should come after compressors but there are many situations where the reverse is true.  When you limit first (in recording) you end up creating a signal that needs much less compression after.  (in mixdown) It is still needed but the compressor won't need to work anywhere near as hard as if it was first.  It will sound transparent and lower ratios can be used.  (eg 2:1 anything over and you are over compressing)  Lower ratios make things sound bigger.  Then you have the luxary of turning the vocal down.  Maximum illusion, minimum voltage.
 

As it's a software compressor limiter, and gains no benefit to improving the gain at the A/D conversion, I would record without in Sonar, than record with in Quad's mixer. You can always apply some in Sonar when recording, but record the clean audio, then have the option to change tweak an unprocessed sound in the mix

 
 

I agree with Jeff, if a situation demands it, adapt as necessary.
 
On a side (but important) note, be sure to stress (and nag if necessary) the importance of them improving their art. The reason I say this is because I have a friend whose daughter has progressed through audition levels on The Voice and such, but never made it. I went to watch her perform a few months ago and she has an incredibly powerful voice but lacks stage presence and song choice to match it. What concerned me even more was when I sat down to speak with her afterwards she came across as "disinterested" in improving her art. It is very saddening to see such a thing in progress, but nip that one in the bud if you can.

Compressing on the way in used to be standard practice, especially for vocals.
 
The need for it is less in the digital age, but it can still be helpful when you are both the singer and the engineer and you can't watch meters and sing at the same time. Just keep the settings gentle, and if the limiter LED (which I'm assuming the interface has) is turning on often, back off the gain a bit.

Compression on the way in is good and nice for the better singers. Slight conditioning on the signal on the way in can be very good.  Limiting on the way in (with a high threshold) for out of control singers though.  There is a difference.

Your interface compressor could be useful for live monitoring. Very low latency.

I use a hardware compressor limiter on the way in and also Mic some sources twice.

I use interface effects when mixing live gigs or for other monitoring.