Speaking as someone who did this for a living for a few years, I can only say that the "words of the pilots" are, unfortunately, not terribly useful. Some will be accurate, some will not be (and wildly so), and there's no way to tell which is which.
I was lucky enought to be working on modeling WWII aircraft, and research at the NASM Archives, the USAF Museum at Wright Patt, and the IWM garnered us reams of actual flight test documentation carried out with rigorous accuracy. This is what we used for our flight models.
Now the interesting thing is after you've seen a series of flight tests which all agree with each other, you can then go back and read what the aces said, and it was very interesting how wrong they were. That's not to say they were in any way untruthful, but they often attributed performance differences to the aircraft, when in actuality (after seeing the real data) what they saw was differences in pilot capabilities. For example, it's very easy to find high-ranking WWI German aces who will insist that their 109's could outturn a Spit. Sorry, no, not if the pilot of the Spit had any clue where the edge of his envelope was.
What's even more frustrating is that right after WWI, the US did quite a bit of serious testing of various WWI types. I found all of the orders and the status reports at the national archives. This was part of the Sarah something or rather collection that was brought to the archives from Wright Patt at some point. The frustrating thing is that every single enclosure (that actual test reports) had been removed. This particular collection goes all the way up to the '50s, and the bottom line is that someone out there has a goldmine in flight test data that they stole before the boxes made it to the archives. The folks at the NASM archives have made a few attempts to find out what happened to the enclosures with no result- it's one of the great mysteries of those seriously involved in aviation research.
Whoops, wandering far afield here. What I meant to say is don't use the words of the pilots. There's a better way. Go to
http://www.x-plane.org/ and get yourself a copy of x-plane, which does a pretty phenomenally accurate job of creating accurate flight models based on the real geometry and thrust output that you plug into it. In other words, if you can get the airfoil shapes (and I've seen quite a few that appear to have been accurately measured) and a good three-view (easy enough to get) x-plane should be able to spit out some fairly accurate performance figures. You'll have to make some assumptions for prop efficiency (guess low) and hp output, but I think this is the best you're going to get.