I've noticed this phenomenon in two groups I joined from on-line ads (one I stayed with, the other I didn't). The DMs instructed the players to roll for stats using the standard method. When I rolled each time, my highest stat was 15-16 and my lowest stat was 8-9.
Then I noticed that some players, usually those who had played with the DM in the past, had much higher stats.
In the first group, people "rolled" at home. I thought my character's stats were decent (16, 15, 14...), but they were the lowest in the group. One guy had used a computer program to roll his character. I knew because the program left the record on his sheet that he had rolled 26 times to achieve all his 18's, 17's, and 16's.
In the second group, I rolled my character's stats: 15, 14, 13, 12, 10, 9.
After we were done, I looked at the guy next to me (he had arrived with the DM). His stats: 13, 14, 15, 18, 18, 18 (in that order).
When I expressed my disbelief, the DM said, "I saw him roll". (I suspect the guy rolled d6 24 times, and then put his rolls in groups of four and THEN dropped the lowest.)
When another player started (an old member of the group), I found out he had a 17, 18, and 20. One new guy (who had rolled honestly) had a couple of low stats, a 6 and a 7. Because he was a noble from Conan RPG, he put the decent scores in his mental stats. When play began, the DM mentioned how weak the character was. He seemed surprised, even though he was there when the stats were rolled.
And of course, no one with 18's remarked on how lucky they were to have rolled such high scores. There was never any mention, either in or out of character dialogue, about how exceptional their characters were compared to everybody else.
Why this self-deception? It seems entirely common in the D&D community. If people want super-high scores, why don't they just use a different means of stat generation? Now, if I hear people rolled for stats, I assume something shady is going on.