Don't get me wrong, there are many ways to make a CJ/YJ flex VERY well as you seem to have found out... but it is a matter of simple math to see how the wider the track width and the shorter the wheelbase, the less flex a vehicle has for a given ramp score.
Using your example - you scored an 800 on a 22 degree ramp with a 92" wheelbase. That means you traveled 73.6" up the ramp. With your new 97" wheelbase, you would have to travel 77.6" up the 22 degree ramp to get the same score of an 800. BUT, assuming no other changes besides wheelbase, and the same track width, a score of 800 with a 97" wheelbase vs. a score of 800 with a 92" wheelbase vehicle means that the 97" wheelbase vehicle has more flex (it traveled further up the ramp, so the axles are at more extreme angles in ralation to one another).
To put this into "so silly it is obvious" terms, imagine a vehicle say a 40" wheelbase that has a track width (sidewall to sidewall for example) of 100". a vehicle like that could quite possibly score a 10,000 on a 30 degree ramp... Even at a 1000 on a 30 degree ramp, that vehicle would hardly be flexing the suspension at all [img]/wwwthreads_images/icons/smile.gif[/img]
As you can see, the shorter the wheelbase and the wider the track width, the better the ramp score - but not necessarily meaning better flex.... That make sense?
As to your Jeep's example of such a great increase in ramp score, you obviously made other changes to get that much of an increase. But to further prove my point, imagine your same suspension setup that scored a 1343 on a 25 degree ramp if you shortened the wheelbase back to 92". Your RTI would jump up to a new score of 1410 [img]/wwwthreads_images/icons/wink.gif[/img]
On a related note, RTI just cannot account for the width factor and give a relevant number for comparison between a variety of vehicles. RTI should either be changed to account for width, or a new way of measuring articulation should be used: such as angle finders on each axle measuring the degree of the articulation of each axle, then finding the difference in the angles and calling that your "score". This would eliminate all of the "cheating" methods commonly used in RTI testing, and tell you who REALLY has more flex [img]/wwwthreads_images/icons/smile.gif[/img]