Rather than use two totally different cars you might get a more interesting result by using one car but with two different levels of weight, and two different levels of power. If you wanted to add variables, you could do different levels of downforce, and/or different tires. Note that the number of combinations goes up as 2 raised to the "number of variables" power (i.e. 8 for 3 variables, 16 for 4 variables etc.)
Another approach would be to use weight and power at three levels each. This gives 9 combinations. Since two points only define a straight line, but three points describe a curve, this gives a better picture of the effect of each variable.
If you want to get a little more complex, you can repeat the 9 combinations for two different levels of downforce, or with hard and soft tires etc. This would give 18 combinations. Then rather than have each driver try all the combinations, have each driver randomly assigned to do some portion (half or a third) But have every driver do the medium weight- medium power combination at each level of the third variable. You can then adjust each drivers times proportional to how far off they are from average on the medium-medium points. This adjustment would compensate for differences in the driver ability at each point.
Whichever way you go, you need to be alert for and control unintended variables. The most obvious one here is "practice effect". Each drivers times will improve somewhat as he drives, so the combinations that he tries later will look "better" than the ones he attempts initially.
It's also important to choose the levels of each variable carefully. If the hp difference is 10 sec/lap but the weight difference is only 0.1 sec/lap, the effect of the weight change will be lost in the "noise". Do some testing before setting the levels.
)
