I'm a huge fan of speed comping. As a result, I've been doing quite a bit of it in recent projects.
 
I'm finding that more often than not, the first take has the greatest number of useable sections, with the second take next. There may be a section here and there from a later take, but it seems fairly rare. In the most recent song I posted on YouTube, almost all of it was first take.
 
Has anyone else noticed a similar trend with their comping?

I tend to do 3 to 4 takes of vocals and the same with bass............with guitar...............I won't even go there :-).
 
But yes I agree. I get most stuff from the first 2 takes. 

Same here, I learned a long time ago not to be so quick to erase a part I was thinking I sucked on.. My basic template for songs has 3 tracks for each part, Vox, Guitar and bass. 
 
So I warm up without record on and get levels set, Then start recording with Vox 1 all the way to the end. Not happy, but I toggle record to Vox 2 and mute Vox 1 and carry on This track I'll stop at the point I screw up slip edit back to a logical cut point before the screw up.. Toggle record to Vox 3, Set the now time a few measures before Vox 2 ends and carry on until next screw up. I flip between Vox 2 and 3 until finished. 
I then compile a whole track using cut, delete and slide Vox 3 up into Vox 2. I set snap to a Quarter note or whole measure depending. Sonar seems to do a good job of combining the takes at the cut points without artifacts. It looks like it draws little crossfades if you zoom in.  
 
Any how, I leave it alone and go work on something else or sleep on it. I come back and listen to Vox 1 and Vox 2 and carry on from there. Sometimes Vox 1 was best and only needs a few parts cut and pasted from Vox 2 to polish up... This is the same for Guitar and bass, Mandolin, Dobro,,, everything real.. 
 
Midi I just play as good as I can and edit the one track. Cheating is much more fun with MIDI. It is allowed by me because I am not a keyboard player, or drummer.  I guess we have tools for cheating with audio too, but I am not a fan of audio cheating at this point. If I can't sing it, or play it, then why would I try and record it? 

for whatever reason I've been getting my best performances- on the average- from the second take.  Now I will say that sometimes when I'm sketching a song and the vocalist does a scratch vocal track I tend to gravitate toward that take.  There sometimes seems to be a kind of un-seen variable to some first takes- the 'spirit' of the performance.  The performer can re-create the notes but not the stuff that was going in the psyche at the time. 
 
Thanks for bringing this up man!

For me, it's also the second take followed by the first take generally. Though sometimes I may totally redo the vocals a couple of days later if I'm not satisfied and on those takes it's also the second and first takes that have the keepers in them.

There seems to be three types of session performers for me.
1: come in and nail it in the first couple of takes and there's lots to work with
2: hunt around for the part and then slowly craft it into a working line, later takes are the most refined
3: kind of like Nuke LaLuche in Bull Durham, all over the map and I end up comping the crap out of it and 'building' the part lol!

My first instrument tracks are usually the the keepers, but it take 3 or 4 takes for my voice to warm up.  It's remarkable how several of us are using the same or similar techniques.  Craig, I love the new comping mode and think it's one of CW's best improvements in years.

Featherlight
There seems to be three types of session performers for me.
1: come in and nail it in the first couple of takes and there's lots to work with
2: hunt around for the part and then slowly craft it into a working line, later takes are the most refined
3: kind of like Nuke LaLuche in Bull Durham, all over the map and I end up comping the crap out of it and 'building' the part lol!

Interesting...at one point in my life I did a lot of production, and found that certain performers followed a "performance curve." I wrote an article about it...I think you'll find it interesting, as it further explores what you're talking about.

being spontaneous isn't something we can practice,
it happens, its original and once we practice something or get deep in to takes we lose that edge, that raw originality we had in the early stages of creating passages with long brush strokes.
 
I tend to stick to the earlier takes as that's where the magic is, once we polish it can sound to over produced.
I find some of my best material came as a young songwriter when I had reckless abandonment in arrangements.
not adhering to any special formula's. 
verse / bridge / chorus.
 
 sometimes its as if you are trying to capture your life story in a 3 minute piece.
its hard to tell a story that happen 20 years ago with fine detail., you tend to ad-lib.
the same can be said for a 4th or 5th take.
 

As I am laying down tracks solo more often than not, I will often cheat and do a take till I hit a snag, back up 2 bars and continue. Not sure how to classify that though? Three partials??