Logically the lighter MR2 with less transmission losses is the quicker car assuming everything else to be equal. But, whilst theories and logic are great in the classroom they don't necessarily hold true on the street as there are too many other variables to consider.
Traction depends on conditions. MR2s are tail happy, GT4s are grippy buggers. Perfect conditions favour the MR, the worse the conditions get, the more the advantage swings to the Celica. Do you want to race on a warm summer afternoon, or on a rainy night? If it snows I would bet everything on the Celica because there's a chance the MR doesn't even make it to the race! I've had a few 200+ hp rear drive cars and even the difference in traction from warm sun in the afternoon to a cool evening later the same day can mean the difference between a good launch or 20 yards of peeling. Incidentally the grippiest tarmac you will find on the street is the funny colour stuff near junctions and crossings, then compare it to the start line at Santa Pod where you can watch people walk on it and lose their shoes.
These are old cars. How quick they were when they came out the factory doesn't mean they are that quick now. Is an MR2 that has had a hard life still quicker than a GT4 that has lead a sheltered, cherished existatnce? Even if you take two brand new identical cars off a production line, one will be quicker than the other. Also sometimes the official figures are just not attainable in the real world, E type Jag, Pontiac GTO and Maestro Turbo all being well documented examples of this.
From the figures quoted on here it seems the MR2 is at best about half a second quicker to 60. That means the driver of the slower car in the other lane could go on amber and tell everyone he saw you off.
Is half a second all there is to it when choosing between two very different cars?