Agreeing with much of the above, I think it’s fairly obvious that games can have an impact on the way people approach problems. A city builder like SimCity is essentially a model representing the real world using certain assumptions. Playing the game habituates the player to those assumptions. Essentially, you’re spending a very long time testing the model that the designers put into place, but without any checks back to the real world, as would be done in an academic environment.
The answers here are, respectively: All of them. Yes. Depends on the player.
To be clear, when you’re designing a computer program of any sort, game or otherwise, all the logic that goes behind the game is manufactured. All of the representation of that logic is entirely manufactured as well. Saying that “police stations decrease crime in area X” actually means “object P changes variable C in radius R”, those are all defined arbitrarily. A police station could just as easily generate power, or increase traffic congestion, or anything else. Generally, simulation designers will attempt to make their models match with real-life, because they want the results to make sense to the player, but a lot of things in the real world may have hidden effects that aren’t obvious. What the designer is concerned about more that real-world accuracy is making their model adhere to player expectations. If there are counter-intuitive results borne out in real world data, these may intentionally be modeled incorrectly in order that clearer cause and effect be implemented.
Designing a game is a mix between designing a system (the city simulation) and communicating that system to the players. Particularly for mass-market sim games like SimCity, accessibility demands that the number of totally hidden variables is reduced.
As for designer bias, as with any model (economic, scientific, etc.) is by definition a simplification of the real world that focuses on particular aspects at the expense of others. Even without an intentional point to make, the interests of the designers are going to show through simply in the nature of which elements they decide to represent, and which ones they omit, and how much importance is placed on the elements that are represented.
That all being said, I think you’d have to really look at the data to see if this has any potential impact on real-world policy decisions. You’d need to look at sales numbers, demographic information about how likely simcity players are to vote, etc to see if there’s any kind of meaningful voting block. If it was really huge, like pop-culture saturation / Avatar / Twilight huge, it would be a different story. But I just don’t see that happening. It’s also possible that a single mayor / councilman, etc in the right position playing the game would skew the policy accordingly, but that would be a pretty long shot.