Since WFRP3 characters are estimated to be quite a bit tougher than 2e characters, I had to evaluate this on the dice roller:

**Success chances without training:**

Scribe attempting to hit in combat in neutral stance (2 blue, 1 purple) = Success chance: 44% Three success: 0%

Scribe attempting to hit in combat in one stance (1 blue, 1 green, 1 purple) = Success chance: 51% Three success: 0%

**Scribe attempting a 2d athletics check (2 blue, 2 purple) = Success chance: 25% Three success: 0%**

Thief, attempting to hit in combat in neutral stance (3 blue, 1 purple) = Success chance: 59% Three success: 6%

**Thief, attempting a 2d athletics check (3 blue, 2 purple) = Success chance: 38% Three success: 3%**

Soldier, attempting to hit in combat in one stance) 3 blue, 1 green, 1 purple = Success chance: 77% Three success: 22%

**Soldier, attempting to make a neutral atheltics 2d check) 4 blue, 2 purple = Success chance: 51% Three success: 9%**

Trollslayer, attempting to hit in combat, in one reckless stance) 4 blue, 1 red, 1 purple = Success chance: 83% Three success: 36%

Those odds look about right to me for a newbie character.

__ __

__Now, Consider this Strength 4 Thug WITH training:__

Thug, attempting to hit in combat in 1 stance (3 strength, 1 red reckless, 1 yellow, 1 purple) = Success chance: 83% Three success: 39%

**Thug attempting a 2d skill check in neutral stance (4 strength, 1 yellow, 2 purple) = Success chance: 64% Three success: 21%**

Trollslayer attempting to hit in combat in 2 reckless (3 blue, 2 reckless, 1 yellow, 1 purple) = Success chance: 90% Three success: 55%

What changes is that untrained characters do not the the high "triple success" effects.

What's currently happening (according to the OP), is that his group feels like they start maxed out, and there's nowhere to go and nothing makes anyone unique. Essentially, everyone is succeeding in massive doses no matter what their specialty. Removing automatically starting with "yellows" in a bunch of skills, definitely changes the odds of the purple dice significantly.

A GM can certainly expect to use a lot fewer "necessary" purple dice, and there is likely to be a lot more "assist" maneuver going on.

jh