Teaching physical education requires competencies to conduct the classes and to assess the motor skills of practitioners. Specialists (physical education professionals) and generalists (primary school teachers) differently experienced motor tasks during their academic education. This study aimed to compare the teachers’ ability in assessing the children’s forward and backward rolls from the analysis of the reliability of an evaluation grid of rolling abilities (Information Scale for Agility on the Soil, InfoSAS), which was investigated in a first study with teachers. A second study in young children explored the responsiveness of the InfoSAS to discriminate by skill level or by training effects. When administered by specialists, the InfoSAS resulted in being reliable (forward: p = 0.087 and p = 0.908; backward: p = 0.926 and p = 0.910; intra- and inter-rater reliability, respectively) and responsive in detecting differences due to expertise (gymnasts vs. primary school children; forward: p = 0.003, backward: p = 0.016) or improvements after specific training in rolling (pre- vs. post-children’s training; forward: p = 0.005, backward: p = 0.001). The results support the conclusion that specialists exhibit higher competence than generalists, which allows proper application of the InfoSAS, possibly because of the practice of skills and reflective teaching styles in physical activity they experienced, along with their academic education in sport sciences.