This is weird. I have computers on both front and rear wheels. Front was there first, added rear for stationary trainer use. Took bike on road for twenty miles. Computers are off by .4 miles. Rear one measures MORE distance. Rear one is cheap Schwinn, but wheel circumference input is measured in mm, so input was accurate to 1 mm. Cateye on front has input only for 700cx28. Even though cateye was way more expensive, could it be possible it is less accurate than the cheapy Schwinn?

It all depends on how you calibrate the comps, whether you actually measure the loaded rolling distance (best), or just measure the circumference unloaded (more unreliable), or enter a nominal wheel size (most unreliable).

The computers simply count wheel rotations and multiply by circumference. If the circumferences are the same and the pickups don't miss wheel rotations, the computers will report the same numbers. If the numbers are different, it's the entered circumferences being different or the pickups occasional missing a rotation. It's most likely the former.