From my reading of the article, it seems like they tracked the number of laptops which failed and needed repairs, not the number which failed and were repaired.

From my reading of the article, it seems like they tracked the number of laptops which failed and needed repairs, not the number which failed and were repaired. So no, higher retained value doesn't explain it. Even if they were looking at how many were repaired (where lower retained value could tilt repair percentage downward, by making people disinclined to repair them, as you suggest) this would merely make other brands of laptops less reliable (with lower numbers caused by fewer damaged laptops being repaired); correcting for it wouldn't improve Mac/IBM/Lenovo's numbers, though it could damage the others, making them seem better by comparison.