Well, the debate as to whether or not ability scores should count when calculating CR/EL has raged for quite some time, but I have finally obtained the proof to end the debate once and for all.
The verdict: DO NOT count ability scores as doing so WILL eventually lead to a Total Party Kill!
The proof: There are several battle combinations I could use here, but I'll go with the one that offers the most conclusive and vivid proof.
Standard Level 2 Party V. Troglodyte Zombies
Why is this the best encounter to use as proof? Simply because I've played it several times (in the same night, trying to get the party through it alive), have seen what it can do, and know everything about it.
On one side we have a standard Level 2 party rolled up using the standard 4d6 method in the PH. We got a Fighter with 17/15/16/13/14/11 using a longsword with Weapon Focus, Improved Initiative, Endurance, and Diehard. We got a Cleric of PELOR with 14/11/13/13/16/14 using a heavy mace with Improved Initiative, Combat Casting, and +2 to Turn Undead checks as well as both the Healing and Sun Domains. We got a Wizard with 12/12/15/16/11/10 using a dagger with Toughness and Combat Casting. We got a rogue with 15/16/15/12/10/14 using a short sword and a composite shortbow with Acrobatic and Nimble Fingers. Basically we got your average run-of-the-mill party. Fighter dude with a big sword, good cleric with her deity's favored weapon, wizard with a pokey thing and spells, and a rogue with a light poker weapon that's made for searching for and disarming traps more than anything else. This is basically the iconics right here, and with damn good ability scores.
On the other side . . .
Zombie; Troglodyte Zombie: CR 2/3, EL 0; Medium Undead; HD 4d12+3; hp 29; Init -2; Spd 30 ft. (can’t run); AC 16, touch 8, flat-footed
16; Base Atk +2, Grp +3; Atk +3 melee (1d10+1, greatclub) or +3 melee (1d4+1, bite) or +3 melee (1d6+1, slam) or +0 ranged (1d6+1,
javelin); SQ Single actions only, damage reduction 5/slashing, darkvision 60 ft., undead traits; AL NE; SV Fort +1, Ref -1, Will +4; Str 12,
Dex 7, Con --, Int --, Wis 10, Cha 1.
Skills and Feats: Toughness.
Undead Traits: Immunity to all mind-affecting effects, poison, sleep effects, paralysis, stunning, disease, death effects, damage to its
physical ability scores, fatigue, exhaustion, and any effect that requires a Fortitude save (unless the effect also works on objects or is
harmless); not subject to critical hits, nonlethal damage, ability drain, or energy drain; not at risk of death from massive damage, but when
reduced to 0 hit points or less, it is immediately destroyed.
Now I have done this battle two ways: my modified way and the official UK way. My modified way, which works CRs closer to the party's level, STILL results in a TPK here, and using the official UK way, it's just downright nasty.
My way: I don't factor the party's ability scores, but I take the CRs directly from the most Appendix itself, Version 5. I use the old Version 4 method of determining overall EL, though, taking the base EL and increasing it based on the number of enemies. This results in a Level 2 Party at EL 5 each facing SEVEN Troglodyte Zombies to have what is a "standard" encounter (EL 5 in this case). Well, after playing the battle multiple times, there is WAY (short of cheating the dice rolls and/or the monsters rolling bad every time while the players roll great every time) for the party to win this. Between the 29 hp and massive attacks, one slashing weapon simply isn't enough. Don't even try to say the battle was stacked due to only having one slasher, because it isn't. Even the treasured iconics Tordek, Jozan, Mialee, and Lidda use this weapons setup, the only difference being that Mialee has a rapier instead of a dagger; that right there shows that traditionally only the fighter really gets the slashing weapons while spellcasters get bludgeioning weapons (and sometimes piercing weapons) and rogues get piercing weapons. What does this mean? A standard party can't win, period, as the zombies are pretty much immune to all of the party's attacks. Also, as far as Turn Undead goes, there is less than a 50% chance of getting it to work, and the HD are high enough to make it so only one or two get turned if it does work; the Sun Domain doesn't do much to add to that, and the zombies will quickly pound the fighter to mush.
The official UK way: This way is even worse. You see, UK counts the ability scores for the party, something I warned him was a bad idea. He never listened. My party CRs doing it his way are 4.3 (4), 3.8 (3.5), 3.3 (3), 3.9 (3.5); that makes them CR 14 and thus (with four members) EL 16-4=12. Also by his numbers, the same seven troglodyte zombies are CR 2/3*7=4-2/3 (4) and thus (with seven members) EL 9-5=4. See the problem now?! With my way even, the encounter was considered a standard encounter using 25% of the party's resources. Using the official way, this encounter was EL -8, supposedly "Easy" and using only 6.2% of the party's resources. By UK's estimation, the number needed for a normal encounter, well, we'd need 27 troglodyte zombies! CR 2/3*27=18 (18) and thus (with 27 members) EL 17-9=8. It doesn't take a rocket scientist to figure out that 27 of those things could flatten any normal Level 2 party in one round flat (well, two rounds flat because they gotta get there first, heh).
The answer: I took away the massive ability score penalty these things get and then calculated it using the Version 4 numbers to EL method and without counting the party's ability scores. The troglodyte zombie became CR 1.5 and thus EL 3, with TWO being an EL 5 encounter. The party finally won, but had to use that once/day Sun Domain power to do it.
Note: I can find absolutely no encounter anywhere that is "off" due to not counting ability scores while I can find massive numbers of encounters that are "off" due to counting ability scores.
@UK: Checkmate, old friend.
P.S. No, I'm not being a smartass. Me and UK have debated this in a friendly way for over a year. I'm just happy to finally find irrefutable proof on my side.
The verdict: DO NOT count ability scores as doing so WILL eventually lead to a Total Party Kill!
The proof: There are several battle combinations I could use here, but I'll go with the one that offers the most conclusive and vivid proof.
Standard Level 2 Party V. Troglodyte Zombies
Why is this the best encounter to use as proof? Simply because I've played it several times (in the same night, trying to get the party through it alive), have seen what it can do, and know everything about it.
On one side we have a standard Level 2 party rolled up using the standard 4d6 method in the PH. We got a Fighter with 17/15/16/13/14/11 using a longsword with Weapon Focus, Improved Initiative, Endurance, and Diehard. We got a Cleric of PELOR with 14/11/13/13/16/14 using a heavy mace with Improved Initiative, Combat Casting, and +2 to Turn Undead checks as well as both the Healing and Sun Domains. We got a Wizard with 12/12/15/16/11/10 using a dagger with Toughness and Combat Casting. We got a rogue with 15/16/15/12/10/14 using a short sword and a composite shortbow with Acrobatic and Nimble Fingers. Basically we got your average run-of-the-mill party. Fighter dude with a big sword, good cleric with her deity's favored weapon, wizard with a pokey thing and spells, and a rogue with a light poker weapon that's made for searching for and disarming traps more than anything else. This is basically the iconics right here, and with damn good ability scores.
On the other side . . .
Zombie; Troglodyte Zombie: CR 2/3, EL 0; Medium Undead; HD 4d12+3; hp 29; Init -2; Spd 30 ft. (can’t run); AC 16, touch 8, flat-footed
16; Base Atk +2, Grp +3; Atk +3 melee (1d10+1, greatclub) or +3 melee (1d4+1, bite) or +3 melee (1d6+1, slam) or +0 ranged (1d6+1,
javelin); SQ Single actions only, damage reduction 5/slashing, darkvision 60 ft., undead traits; AL NE; SV Fort +1, Ref -1, Will +4; Str 12,
Dex 7, Con --, Int --, Wis 10, Cha 1.
Skills and Feats: Toughness.
Undead Traits: Immunity to all mind-affecting effects, poison, sleep effects, paralysis, stunning, disease, death effects, damage to its
physical ability scores, fatigue, exhaustion, and any effect that requires a Fortitude save (unless the effect also works on objects or is
harmless); not subject to critical hits, nonlethal damage, ability drain, or energy drain; not at risk of death from massive damage, but when
reduced to 0 hit points or less, it is immediately destroyed.
Now I have done this battle two ways: my modified way and the official UK way. My modified way, which works CRs closer to the party's level, STILL results in a TPK here, and using the official UK way, it's just downright nasty.
My way: I don't factor the party's ability scores, but I take the CRs directly from the most Appendix itself, Version 5. I use the old Version 4 method of determining overall EL, though, taking the base EL and increasing it based on the number of enemies. This results in a Level 2 Party at EL 5 each facing SEVEN Troglodyte Zombies to have what is a "standard" encounter (EL 5 in this case). Well, after playing the battle multiple times, there is WAY (short of cheating the dice rolls and/or the monsters rolling bad every time while the players roll great every time) for the party to win this. Between the 29 hp and massive attacks, one slashing weapon simply isn't enough. Don't even try to say the battle was stacked due to only having one slasher, because it isn't. Even the treasured iconics Tordek, Jozan, Mialee, and Lidda use this weapons setup, the only difference being that Mialee has a rapier instead of a dagger; that right there shows that traditionally only the fighter really gets the slashing weapons while spellcasters get bludgeioning weapons (and sometimes piercing weapons) and rogues get piercing weapons. What does this mean? A standard party can't win, period, as the zombies are pretty much immune to all of the party's attacks. Also, as far as Turn Undead goes, there is less than a 50% chance of getting it to work, and the HD are high enough to make it so only one or two get turned if it does work; the Sun Domain doesn't do much to add to that, and the zombies will quickly pound the fighter to mush.
The official UK way: This way is even worse. You see, UK counts the ability scores for the party, something I warned him was a bad idea. He never listened. My party CRs doing it his way are 4.3 (4), 3.8 (3.5), 3.3 (3), 3.9 (3.5); that makes them CR 14 and thus (with four members) EL 16-4=12. Also by his numbers, the same seven troglodyte zombies are CR 2/3*7=4-2/3 (4) and thus (with seven members) EL 9-5=4. See the problem now?! With my way even, the encounter was considered a standard encounter using 25% of the party's resources. Using the official way, this encounter was EL -8, supposedly "Easy" and using only 6.2% of the party's resources. By UK's estimation, the number needed for a normal encounter, well, we'd need 27 troglodyte zombies! CR 2/3*27=18 (18) and thus (with 27 members) EL 17-9=8. It doesn't take a rocket scientist to figure out that 27 of those things could flatten any normal Level 2 party in one round flat (well, two rounds flat because they gotta get there first, heh).
The answer: I took away the massive ability score penalty these things get and then calculated it using the Version 4 numbers to EL method and without counting the party's ability scores. The troglodyte zombie became CR 1.5 and thus EL 3, with TWO being an EL 5 encounter. The party finally won, but had to use that once/day Sun Domain power to do it.
Note: I can find absolutely no encounter anywhere that is "off" due to not counting ability scores while I can find massive numbers of encounters that are "off" due to counting ability scores.
@UK: Checkmate, old friend.
P.S. No, I'm not being a smartass. Me and UK have debated this in a friendly way for over a year. I'm just happy to finally find irrefutable proof on my side.
If someone is dumb enough to create a creature with 5 HD and 200 in every ability score, then he made his own bed and should have to lie in it. The rules should assume a little bit of common sense on the DM's part. All 30s is fine for a Solar, but it doesn't work with a Bugbear. 
