A lot of that is Hasek's sample being mostly dead puck era.
Roy played 88 games through the 1992 playoffs with a .908/2.53. Of course, among goalies with 20 GP from 1980-1992, Roy is 1st in save percentage. Hasek, in 6 GP, is .899/2.92, but he hasn't played a full playoff series yet.
Through 1997, Hasek catches up. Roy has 153 games, .917/2.38, while Hasek through 22 GP has .919/2.37. Through this point in time Roy has more Cups and as many Conn Smythes as Hasek has full series played (2 - a loss to the 94 Devils where he played very well, and a loss to the 95 Flyers where he was quite terrible.) But if you're going by averaging stats, they're the same goalie!
Then we hit the dead puck era. Because he has so few games played, and because he mostly deals with weak Eastern Conference teams, Hasek shoots up to .927/2.03 while Roy, through 253 games is at .918/2.30, finishing his career with playoff averages that are slightly worse than Ron Tugnutt. Roy adds a 3rd Smythe and a 4th Cup while Hasek wins his first Cup by posting a .920 (.914 vs the teams in the West, .942 against the weak Eastern representative in the Finals).
That's most of the case. They're about the same in the playoffs, and Lemieux has a really strong Canada Cup that was more impressive than Orr's 76 CC (plus other solid performances as an old guy in 2002 and 2004).
The other thing to argue for Lemieux, is to flip the complete season argument against Orr. Lemieux gives you more kicks at the can while producing at an MVP-ish level. Orr has 6 superb seasons from 69-70 to 74-75. Outside of that he doesn't offer a ton to a Big 4 debate. Lemieux, even in partial seasons, was capable of winning an Art Ross. He had 8 120-point seasons, and his 6 Art Rosses ties Gordie Howe for 2nd.