Their control group almost certainly has some people with undiagnosed ADHD, and although they don’t describe their recruitment methods, I suspect they (unintentionally but knowingly—the participants are all college students) excluded a lot of folks through how they identified potential participants.
They fact that three quarters of their ADHD participants were on stimulant medication already indicates they’ve oversampled on those folks, and honestly I’m guessing their stats reflect more the effect of Adderall and Ritalin on visual processing than they do the effects of ADHD.
But one of the things that’s often misunderstood about ADHD is that whatever the physical underpinnings of the sorts of neurodivergence that get classified as “ADHD symptoms”, a key aspect of whether or not someone “has ADHD” is not whether their brain functions in a different way that “typical” people, but instead whether or not those symptoms cause regular distress in their daily lives.
In other words, you may exhibit many of the physical and behavioral markers of ADHD but if those traits aren’t making you struggle, then you don’t “have ADHD”. And so any physical test is going to be unable to ever be entirely reliable, because there are lots of folks living their lives who would flag on the test but who don’t “have ADHD”. The study results even hints at that with a much worse rate of identifying neurotypicals successfully.