As a civvy, and, worse still, a journo, I can at least make totally unclass. statements, if my more learned brethren will forgive me (and hopefully correct me where I talk @rse).
With regard to EF's manoeuvrability, it needs equal or superior manoeuvrability to any potential threat in the low-speed/close in regime for all the reasons given above. It's debateable as to whether F-16/F-15/F/A-18 have sufficient agility to defeat Su-27/MiG-29 in this area, though until those aircraft get a decent MMI and Western-equivalent radar, etc. they are patently 'beatable'.
But the money on EF agility is being spent to give superior supersonic agility as well, acceleration to impart maximum accel and range to the missile at launch, and turn performance for the reasons referred to earlier. Therefore, in short, YES, THIS DEGREE OF AGILITY IS USEFUL and is worth paying for.
With regard to stealth, there are huge classification issues, but my understanding is that there is a philosophical difference between F-22 and EF. F-22 is designed for 'all-aspect' stealth - like the F-117, because when it was designed it was expected to be a MR aircraft (or to perhaps be required to be one). All aspect stealth is most important and relevant in the air-to-ground role, where you need to be able to manage your RCS in order to avoid detection. NB that F-117 is not invisible to radar, but that when it presents certain aspects RCS is so low that detection range becomes negligible. The cleverest thing about -117 is thus the flight planner, which calculates RCS being presented to known threat radars, and tailors angle of bank, etc. to minimise exposure. That's how, by moving a mobile radar to where it wasn't expected, or by using high-altitude lookdown, or by using bistatic radar, -117 becomes detectable (and, as demod in Kosovo, even 'downable').
Because EF was designed first and foremost as an AD aeroplane, (and specifically as a BVR fighter-interceptor) the effort has gone into minimising frontal RCS, to make the head-on aspect as difficult to detect as possible, and to try and ensure that the EF pilot sees the enemy on radar before he can be seen himself.
There has been some effort to minimise RCS from other aspects, but that has not been accorded the same priority, as far as I understand.
I would suggest that the important factors are MMI/workload/information, weapon and sensor performance (agility and accel will actually enhance weapon performance), combat persistance, pilot quality and training, affordability and manoeuvrability - but you wouldn't want to do without any one of those attributes.
PS: Whatever you do to an F-15 radar, and weapons wise, it's a very poor competitor JOUST-wise (it did worse than F/A-18 in the EF and rivals versus Su-27 with weapons and radar parity), and is also horrifyingly expensive - costing more than EF or Rafale, though less (natch.) than F22.
[ 17 August 2001: Message edited by: Jackonicko ]