I've found the same with decoys vs. barriers. Generally, most scary monsters have greater resistance than intelligence. Although, if you're playing as the Influence, then some of the human order troops have more intelligence than resistance, plus big guns.
Special forces vs. Monsters, basically.
Decoys also make more sense for two other reasons:
1) Minor, but decoy uses intelligence to make, which free partners have in abundance, whereas barriers take resistance, which is generally found in the more-expensive (more expensive than free and no upkeep, anyways) leased or owned units. Although prisoners are a good workaround, having prisoners attracts even more attacks.
2) Major, the very purpose of intelligence vs. resistance in battle. Barriers deflect low resistance units, but since resistance is hitpoints, low resistance characters are usually easier to beat in combat anyways. Decoys deflect low intelligence units, but since intelligence only determines battle advantage (it can essentially reduce enemy hitpoints, although it might affect how a unit actually fights in combat), high levels don't necessarily make a unit dangerous in combat, at least not in the clear way that resistance does.
The easiest way to quest successfully (in my opinion) is to have one big unit that can win most fights and the other six characters as companions, although I can see logic in going all companions + tons of Decoys. Multiple high-intelligence characters do a better job carrying the load than one, especially since you can only use each companion once per artifact piece search. Plus the other stats are vital, such as resistance, terror (although I usually had to use a merit on terror rolls, even with the painter guy that makes everyone insane), and mobility. I've also occasionally seen combat attributes (short medium long range attack values, short range most often because of the arena territories in my personal experience) pop up.
I'm curious if any attributes increase the number of spontaneous actions from units. Observation seems to rule out any specific stats and favor the hypothesis that only some units have spontaneous actions programmed, thus only some conduct spontaneous action. I'm still wondering if experience factors into it, but I don't have enough research to support that.