The harmfulness manipulation moved ratings strongly and in the intended direction.

Evidence

The harmfulness manipulation produced a large effect in the expected direction, F(2,56) = 91.25, p < .001, η2p = .765. The target was rated as most harmful in the harmful agent condition (M = 5.99, SD = 1.08) and least harmful in the benevolent agent condition (M = 1.88, SD = 1.01), with the target in the neutral agent condition falling in the middle (M = 3.33, SD = .76). All pairwise comparisons were significantly different at p < .001. — Mansur Khamitov, Jeff D. Rotman, and Jared Piazza, Perceiving the Agency of Harmful Agents, 237