I do like rating scales, as they have the most relevance to how a child actually functions in the settings in which they spend their time, but they also don't tell you much about why they present the way they do. (Many different factors can disrupt attention and executive function, including transient, acute and chronic causes.) For example, a depressed child can look quite similar to a child with ADHD on a rating scale, depending on the scale, the rater, and the exact presentation of the child, but treatment should be quite different.

Similarly, there is value in the direct measures of some of the core deficits or profiles generally considered to exist in ADHD (most of the EF areas: impulse control/inhibition, sustained attention, regulation of attention, switching, etc.), but having a profile associated with a high likelihood of ADHD doesn't necessarily translate into functional impairments. So if possible, I prefer evaluations that include both direct and indirect measures of attention, impulsivity, and other executive functions.