Studies 3 and 4 used non-gamers, who they turned into gamers with training. (Muah-ha-ha!) They actually included women in these samples. For experiment 3, participants were randomly assigned to play a driving game (Mario Kart) or a "nonaction simulation game" (Roller Coaster Tycoon III). Once again, gamers outperformed non-gamers, and also showed improvement in visuomotor-control tasks over multiple sessions (non-gamers did not significantly improve).
Experiment 4 compared a first-person shooter game (Death Match mode of Unreal Tournament 2004) to another simulation game (The Sims 2). Again, gamers outperformed non-gamers.
A few issues with the study: As I said, they did not include women in the first two experiments, and I'm incredibly surprised that in a college student sample, they were unable to find a handful of women who played video games. The sample sizes for each experiment were quite small (around 24 people, 12 in each group), so it seems unlikely there were not at least 6 female gamers on campus. Second, though experiment 1 looked at the impact of gaming on driving, experiments 2-4 used more abstract visuomotor-control tasks (the kinds you see in most lab studies of reaction time, hand-eye coordination, and so on). So it seems to a stretch for them to highlight their study as being about real-world visuomotor-control tasks like driving, since only one study looked at driving ability, or to state that their study supports the use of video gaming to improve driving:
Combined with these results, the findings of the current study support the claim that easily accessible action video games can be cost-effective training tools to help people improve their essential visuomotor-control skills used for driving.But sure, it's nice to have a study show that gamers have some important skills over non-gamers. Now someone needs to study the impact Pokémon Go on hand-eye coordination - I mean, throwing one of those Pokéballs is really good practice. (Yes, I downloaded it - don't judge.)