That exit poll also showed Perot voters splitting 34-28 Dole (with 34% abstaining), so I would think that result speaks more to flaws in the exit poll than Perot costing Clinton votes.
Yeah, the exit poll is wrong - if you do the math on the gender split, Clinton does better in the exit poll (44.76%) than he did IRL (44.43%). I've noticed the same thing happens with 2000 - the "revised" exits overestimated Bush in Florida and New Hampshire, so you sometimes get arguments that Nader didn't really cost Gore those states.
Speaking of Nader, he was on the ballot in CO in '96 and did rather well (almost 2%). So that would explain part of Clinton's loss.