Robbie
In re-reading your post, I would prefer no EV compensation at all.
Shoot as the camera would meter with all settings neutral.
The idea behind the test is to see how the camera meters each scene with no adjustment. What I have found in thepast, is that with completely neutral settings, between about 25 and 230 greyscale on the histogram each 1EV change in exposure is almost perfectly linear at 45 greyscale. the camera should meter at between 110 and 125 greyscale if all is correct, although none of the cameras are quite perfect.
One thing to be careful of, and I found this with my K50/1.4 is that you will probably need to be in the shade, or on a cloudy day, because full sunlight will really push the shutter speed up, at F1.2, although I think you can set ISO a little lower on the K5,
The end result, should look like the following, which I have posted before.
As you can see, the *istD starts off wide open quite good, droops a little, and then rises constantly as you stop down to about +.7ev at F22. Dealing with a -.5+.7 EV drift over the entire exposure range is not too bad. The K10D is a lot worse starting at -.7 drooping to -1EV and rising to +1.7 EV at the top end. Putting the Jinfinance split image into theK10D resolved most of the large aperture errors, but did not address the intermediate and small aperture errors. In theory this could be better centered with a -.7ev shift on the entire exposure, but sadly the K10D lacks the ability to add EV compensation in manual mode. Both the K7 and *istD permit EV comp in manual. Some say manual is manual and you don't need EV comp, but in cases where a lens shows a consistent tendancy to under / over expose at all apertures, it is useful
The K7 overall is better but still no where near as good as my "gold standard" the *istD