I am not an expert on opportunity cost (or economics in general), but I've learned to recognize it's application in various game designs. I find it easier to think about the colloquial term "trade-off". I'm sure someone better versed in opportunity cost could point out the differences. I don't think it's important for this discussion.
In three role selection games that I'm fairly familiar with -- Puerto Rico, San Juan, and Race for the Galaxy -- you'll find that on a given turn there is generally an optimal role to choose when you consider only your position. When you start to evaluate other's positions, you'll see that the optimal role is not necessarily the one that is most beneficial to your play. For example, you may be able to build a better building (or world/development) if you trade first and then build. However, when you consider that your opponent is both low on cash and has a good that will trade for a lot of cash, you may find that building now is actually the better choice, forcing your opponent to either miss a build phase, or build a building that is relatively worse than the building you can build.
In these games the value of a role selection is driven by quantity of resources (abstractly, goods and cash) as well as timing (which choice do you benefit more from now relative to your opponents). You are gauging your abundance against your opponents' scarcity in the given moment, or vice versa. This is a simplification; there are more factors to weigh than this. At a high level, it summarizes my point sufficiently.
Evaluation of opportunity cost feels slightly different in Eminent Domain, another role selection game that I have had the opportunity to play several times. Timing is still very important, but quantity feels a bit different. In the case of EmDo, quantity generally refers to the quantity of icons any given player has. Resources on the players planets play into this as well, but a player cannot do anything with the resources unless he has the proper icons. The problem is that the majority of the time, your opponent's quantity is not known. You can infer a lot from what they have discarded (and discards are open information), but you are still limited in your knowledge.
Additionally, scarcity can be created simply due to diverging strategies. If two players are specializing in different roles, scarcity is a common occurrence when those two players are compare their positions when determining which role to choose. On the other hand, even when players have two similar strategies, it can be difficult to gauge scarcity unless one of the players had spent a lot of one type of card on the previous turn. With either diverging or similar strategies, timing is easier to gauge than quantity, and therefore it carries more weight in the decisions of the players.
Having played EmDo, I don't see this as a problem. It feels different than other role selection games, but not in a negative way. Most importantly, it doesn't feel like the game plays itself. However, this is an initial concern of mine with the dice game that I'm working on. Admittedly, it's still in the "theory" phase, so it could be that any concern will disappear once I prototype and play it. The above thought process was important to me so that I could understand a potential problem if I find that the game plays itself.