100% agree, these results are in actual fact completely meaningless, to show any viable data they would need to be repeated a number of times (preferably on multiple different copies of the same mobo for each sample) the mean calculated, and subjected to a standard error test with error bars. Any alleged difference outside these error bars then needs to be proved by using a t-test to determine the p-value (probability of the alleged result occuring by chance).
Data this close that has not been analysed this way should be treated with a massive pinch of salt and ignored. Unless the difference between any test subjects in huge, there is no real evidence that there is any difference between the boards whatsoever. Buy whichever takes your fancy, the chances are that in your home rig it will be perform completely differently than the pointless one-off tests shown by the reviewers.