Some tests done on precision:

Test method: 
  * Use A2 casters (mainly rus thunder priests) to cast lightning bolt against single size 6 target
  * Range: ~40 tiles (using smallest formation possible for troops with farthest caster at range 40)
  * Have casters scripted as: 1 x Hold + 4((3 for sidhes)) x Lightning Bolt
  * During the hold turn, buffs were cast by other supporting mages
  * 5 battles of 40 casts for thunder priest
  * 7 battles of 30 casts for sidhe lords
  * Hits were calculated by checking the targets battle log

PREC 10 (Thunder priest)
  * 1: 5/40
  * 2: 1/40
  * 3: 2/40
  * 4: 4/40
  * 5: 1/40
  * SUM: 13/200 = 6.5%

PREC 15 (Thunder priest + Wind guide)
  * 1: 1/40
  * 2: 2/40
  * 3: 4/40
  * 4: 6/40
  * 5: 5/40
  * SUM 18/200 = 9%

PREC 19 (Thunder priest + Wind guide + 4 x precision Bless)
  * 1: 8/40
  * 2: 5/40
  * 3: 9/40
  * 4: 4/40
  * 5: 3/40
  * SUM 29/200 = 14.5%

PREC 24 (Thunder priest + Wind guide + 4 x precision Bless + eye of aiming)
  * 1: 4/40
  * 2: 4/40
  * 3: 2/40
  * 4: 2/40
  * 5: 2/40
  * SUM 14/200 = 7%

PREC 23 (Sidhe Lord + Wind guide + 4 x precision Bless)
  * 1: 4/30
  * 2: 4/30
  * 3: 1/30
  * 4: 6/30
  * 5: 3/30
  * 6: 4/30
  * 7: 4/30
  * SUM 26/210 = 12.3%

notes: 
  * the spread pattern got noticeably smaller with higher precision
  * in earlier test with troop, mages seemed to top most units