The statements are not random in the sense I picked them up on the internet; they are just randomly selected within the sample, which lends legitimacy to the conclusion. Finding random statements on the internet doesn't mean anything. You need to find a similar Mexican leaning sample as well as a similar context. If you can find 10 random examples of Hispanic (Mexican leaning) supporting blacks in this sample or another like, I will concede, but everything points to replication and generalization across the group. Not every generalization is untrue by the virtue of being a generalization.
Now, in this sample, there is no more or less reason to think the Mexicans should show more or less disdain thanwhites. In my opinion the average Hispanic comment seems more virulent than the whites' (you may look at the Facebook post yourself for confirmation) ), which is what scientists have found to be the case as far as bias and resentment is concerned too. The fact random labs correlate with what I found is eerie and presupposes the observation is true.